<%BANNER%>

Influences of Verbal Properties on Second Language Filler-Gap Resolution

Permanent Link: http://ufdc.ufl.edu/UFE0022122/00001

Material Information

Title: Influences of Verbal Properties on Second Language Filler-Gap Resolution A Cross-Methodological Study
Physical Description: 1 online resource (229 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: acquisition, dependencies, l2, language, processing, proficiency, psycholinguistics, second, sentence, syntax
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The difficulty of mastering the syntax of a foreign or second language (L2) by adult learners has been the focus of much research. Recent evidence suggests that even advanced latelearning L2 speakers differ as to how they comprehend complex sentences on-line compared to native speakers and early-learning L2 speakers. One area in which these differences occur is in the processing of filler-gap dependencies, e.g. 'Mary saw which toy the child pushed ___while playing.', where 'which toy' is the filler, which maintains a relationship between itself and the underlying syntactic position from which it moved ?___? (as the direct object of 'pushed'). These dependencies may cause difficulty for L2 speakers because the non-canonical syntactic constituent, the filler, cannot be integrated when encountered, and, thus, must be reconstructed at some stage, the gap, during on-line processing. Previous behavioral research has suggested that native speakers utilize knowledge of a verb?s transitivity frequency (verb bias) and its subcategorization frame in making on-line decisions about possible gap locations while processing wh-dependencies; yet, there are few studies looking at L2 on-line sentence processing of filler-gap dependencies. This study includes three experiments investigating the type of information late-learning intermediate and advanced Korean L2 speakers of English use while resolving filler-gap dependencies in English. Two of the experiments also compare results from different experimental methods, self-paced reading and eye-tracking. Eye-tracking is used to investigate whether a more natural reading task is better suited to testing L2 speakers. Proficiency is included as a factor to examine whether L2 speakers of different proficiency levels make use of these verbal properties in a similar manner. The two studies that compare methods examine the use of subcategorization frame, and a third eye-tracking study looks at the use of verb bias in filler-gap resolution. Results across experiments suggest that that intermediate and advanced L2 speakers use these verbal properties in a different manner than native speakers when resolving filler-gap dependencies. Regarding the use of subcategorization frame, results suggest that advanced speakers attempt to use this information in an on-line manner, while intermediate speakers do not. A comparison of results across techniques reveals that the two experimental methods yield slightly different results for both native and non-native speakers. Regarding the use of verb bias, results suggest that intermediate speakers are more sensitive to this factor than native and advanced speakers are when resolving filler-gap dependencies. Overall, results are discussed from a processing demands perspective.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Kaan, Edith.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022122:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022122/00001

Material Information

Title: Influences of Verbal Properties on Second Language Filler-Gap Resolution A Cross-Methodological Study
Physical Description: 1 online resource (229 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: acquisition, dependencies, l2, language, processing, proficiency, psycholinguistics, second, sentence, syntax
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The difficulty of mastering the syntax of a foreign or second language (L2) by adult learners has been the focus of much research. Recent evidence suggests that even advanced latelearning L2 speakers differ as to how they comprehend complex sentences on-line compared to native speakers and early-learning L2 speakers. One area in which these differences occur is in the processing of filler-gap dependencies, e.g. 'Mary saw which toy the child pushed ___while playing.', where 'which toy' is the filler, which maintains a relationship between itself and the underlying syntactic position from which it moved ?___? (as the direct object of 'pushed'). These dependencies may cause difficulty for L2 speakers because the non-canonical syntactic constituent, the filler, cannot be integrated when encountered, and, thus, must be reconstructed at some stage, the gap, during on-line processing. Previous behavioral research has suggested that native speakers utilize knowledge of a verb?s transitivity frequency (verb bias) and its subcategorization frame in making on-line decisions about possible gap locations while processing wh-dependencies; yet, there are few studies looking at L2 on-line sentence processing of filler-gap dependencies. This study includes three experiments investigating the type of information late-learning intermediate and advanced Korean L2 speakers of English use while resolving filler-gap dependencies in English. Two of the experiments also compare results from different experimental methods, self-paced reading and eye-tracking. Eye-tracking is used to investigate whether a more natural reading task is better suited to testing L2 speakers. Proficiency is included as a factor to examine whether L2 speakers of different proficiency levels make use of these verbal properties in a similar manner. The two studies that compare methods examine the use of subcategorization frame, and a third eye-tracking study looks at the use of verb bias in filler-gap resolution. Results across experiments suggest that that intermediate and advanced L2 speakers use these verbal properties in a different manner than native speakers when resolving filler-gap dependencies. Regarding the use of subcategorization frame, results suggest that advanced speakers attempt to use this information in an on-line manner, while intermediate speakers do not. A comparison of results across techniques reveals that the two experimental methods yield slightly different results for both native and non-native speakers. Regarding the use of verb bias, results suggest that intermediate speakers are more sensitive to this factor than native and advanced speakers are when resolving filler-gap dependencies. Overall, results are discussed from a processing demands perspective.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis: Thesis (Ph.D.)--University of Florida, 2008.
Local: Adviser: Kaan, Edith.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2010-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022122:00001


This item has the following downloads:


Full Text

PAGE 1

1 INFLUENCES OF VERBAL PROPERTIES ON SECOND LANG UAGE FILLER-GAP RESOLUTION: A CROSS-METHODOLOGICAL STUDY By ANDREA C. DALLAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2008

PAGE 2

2 2008 Andrea C. Dallas

PAGE 3

3 ACKNOWLEDGMENTS I would first like to thank Dr. Edith Kaan fo r her support over the past 4 years. She has been an extraordinary ad visor. She has served as an academic model, always pointing out and pushing me to seek opportunities to expand my knowledge and academic career. Without her help, I wonder if I could have made it through this process with such success. I truly appreciate everything she has done for me. I would also like to thank Dr. H. Wind Cowles for her willingness to share her lab and discuss my project along the way. Without her help, I could not have completed this work. I also owe thanks to my other committee members, Dr. Lori Altmann and Dr. Teresa Antes. Dr. Altmann was my first teacher of statistics and I will never forget her patience in teaching me how to set up spread sheets, understand statistical measures, and use SPSS. Dr. Antes has been a source of inspiratio n for me since I first entered the program in 2001, when she served as my first academic advisor. There are two other faculty members who have supported me in various ways throughout this process, Dr. Ratree Wayland and Dr. Caroli ne Wiltshire, and I would like to express my appreciation for their support. In addition, I would like to take this opportunity to thank some of my fellow graduate students for being available to discuss academic matters technological issues or my project, as well as for providing emoti onal support over the course of the last 4 years: Mohammed Al-Khairy, Chris Barkley, Mingzhe n Bao, Rui Cao, Milla Chappell, Nicole Chevalier, Michael Gelbman, Ji rapat Jangjamras, Bin Li, Ji n Park, Sangyeon Park, Priyankoo Sarmah, and Weihua Zhu. Lastly, and most impo rtantly, I would like to thank my husband, Juan Diaz, for his unrelenting support of my academic goa ls. He has supported me every step of the way along my journey, for this I have much gratitude.

PAGE 4

4 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................3LIST OF TABLES................................................................................................................. ..........8LIST OF FIGURES.......................................................................................................................10LIST OF ABBREVIATIONS........................................................................................................ 11ABSTRACT...................................................................................................................................12 CHAP TERS 1 INTRODUCTION..................................................................................................................14Overview....................................................................................................................... ..........14Goals of the Study..................................................................................................................15Organization...........................................................................................................................172 NATIVE AND NON-NATIVE SENTENCE PROCESSING ............................................... 18Introduction................................................................................................................... ..........18Native Sentence Processing....................................................................................................19Sentence Processing Models........................................................................................... 21Native Processing of Filler-Gap De pendencies: Active Filler Strategy.......................... 28Non-Native Sentence Processing............................................................................................ 32Non-Native Processing of Filler-Gap Dependencies...................................................... 33Active filler strategy................................................................................................. 33Use of syntactic traces du ring filler-gap resolution.................................................38Non-Native Model of Sentence Processing..................................................................... 41Summary.................................................................................................................................423 THEORETICAL BACKGROUND........................................................................................44Cognitive Theory of Second Language Acquisition.............................................................. 44Native Language Transfer...................................................................................................... 50Non-Native Proficiency......................................................................................................... .53Factors in Relation to Current Experiments........................................................................... 544 INFLUENCE OF SUBCATEGORIZATION FRAME DURING FILLER-GAP RESOLUTI ON..................................................................................................................... ..56Introduction................................................................................................................... ..........56Participants across Experiments............................................................................................. 57Research Questions and General Predictions......................................................................... 58

PAGE 5

5 Why Compare Self-Paced R eading to Eye-Tracking? ...........................................................59Materials across Experiments.................................................................................................65Experimental Predictions........................................................................................................68Experiment 1: Self-Paced Reading.................................................................................. 68Experiment 2: Eye-Tracking........................................................................................... 69Comparison of Experiment 1 and Experiment 2.............................................................69Procedures across Experiments.............................................................................................. 70Pre-Testing Materials...................................................................................................... 70Proficiency measure: Combined English Language Skills Assessment (CELSA)............................................................................................................... 70Background questionnaire........................................................................................ 71Shipley vocabulary task........................................................................................... 71WAIS vocabulary task............................................................................................. 71Reading span task.....................................................................................................72Digit span task.......................................................................................................... 72Post-testing grammaticality judgment questionnaire............................................... 73Experiment 1: Self-Paced Reading......................................................................................... 73Participants......................................................................................................................73Pre-Testing Scores........................................................................................................... 74Proficiency measure: CELSA.................................................................................. 74Background questionnaire........................................................................................ 74Vocabulary and memory measures.......................................................................... 75Procedures..................................................................................................................... ..76Analysis....................................................................................................................... ....76Results.....................................................................................................................................77Self-Paced Reading......................................................................................................... 77Comprehension accuracy......................................................................................... 77Self-paced reading: native speakers......................................................................... 78Self-paced reading: advanced speakers.................................................................... 81Self-paced reading: intermediate speakers............................................................... 84Post-testing grammaticality judgment questionnaire............................................... 87Discussion..................................................................................................................... ...88Experiment 2: Eye-Tracking................................................................................................... 89Participants......................................................................................................................89Pre-Testing Scores........................................................................................................... 90Proficiency measure: CELSA.................................................................................. 90Background questionnaire........................................................................................ 90Vocabulary and memory measures.......................................................................... 91Post-testing grammaticality judgment questionnaire............................................... 91Procedures..................................................................................................................... ..92Analysis....................................................................................................................... ....92Results........................................................................................................................ .....93Comprehension accuracy......................................................................................... 93Native speakers........................................................................................................94Advanced speakers................................................................................................... 98Intermediate speakers............................................................................................. 101

PAGE 6

6 Post-testing grammaticality judgm ent questionnaire............................................. 104Discussion..................................................................................................................... .104Conclusions...........................................................................................................................1055 INFLUENCE OF VERB BIAS DU RING FILLER-GAP RESOLUTION .......................... 110Experiment 3: Introduction...................................................................................................110Why use Eye-Tracking as Opposed to Self-Paced Reading?............................................... 111Participants...........................................................................................................................112Research Questions and General Predictions....................................................................... 113Materials...............................................................................................................................114Experimental Predictions......................................................................................................116Native and Advanced Speakers..................................................................................... 116Intermediate Speakers...................................................................................................116Methods................................................................................................................................116Pre-Testing Materials.................................................................................................... 116Post-Testing Grammaticality Judgment........................................................................117Pre-Testing Scores......................................................................................................... 117Procedures..................................................................................................................... 117Analysis....................................................................................................................... ..117Results...................................................................................................................................118Comprehension Accuracy............................................................................................. 118Native Speakers............................................................................................................. 120Advanced Speakers.......................................................................................................124Intermediate Speakers...................................................................................................128Post-Testing Grammaticality Judgment Questionnaire.................................................132Discussion.............................................................................................................................133Conclusions...........................................................................................................................1346 DISCUSSION AND CONCLUSION.................................................................................. 137Summary of Results............................................................................................................. .137Experiments 1 and 2...................................................................................................... 137Experiment 3................................................................................................................. 138Discussion.............................................................................................................................138Limitations and Future Research.......................................................................................... 143Conclusion............................................................................................................................146A EXPERIMENT 1 AND 2: EXPERI MENTAL ITEMS........................................................ 148B DISTRACTER ITEMS.........................................................................................................157C SUBCATEGORIZATION FRAME S ELF-COMPLETION QUESTIONNAIRE .............. 162D SUBCATEGORIZATION FRAME GRAMMATICALITY JUDGMENT TASK FOR NORMING ...........................................................................................................................167E LANGUAGE BACKGROUND QUESTIONNAIRE..........................................................172

PAGE 7

7 F SHIPLEY VOCABULARY TEST....................................................................................... 176G WAIS VOCABULARY TEST............................................................................................. 178H EXPERIMENT 1: GRAMMATICALITY JUDGMENT TASK POST TESTING ............. 181I EXPERIMENT 2: GRAMMATICALITY JUDGMENT TASK POST TESTING ............. 187J EXPERIMENT 3: EXPERIMENTAL ITEMS....................................................................192K EXPERIMENT 3: DISTRACTER ITEMS..........................................................................200L EXPERIMENT 3: PLAUSIBILITY JUDGME NT QUESTIONNAIRE 1..........................203M EXPERIMENT 3: PLAUSIBILITY JUDGME NT QUESTIONNAIRE 2..........................210N EXPERIMENT 3: GRAMMATICALITY JUDGMENT TASK POST TESTING ............. 216LIST OF REFERENCES.............................................................................................................221BIOGRAPHICAL SKETCH.......................................................................................................229

PAGE 8

8 LIST OF TABLES Table page 4-1. Eye-Tracking Measures.....................................................................................................614-2. Subcategorization Frame Example Materials.................................................................... 654-3. Percentage of Completions............................................................................................... .674-4. Background Data for Non-Native Speakers...................................................................... 754-5. Vocabulary and Memory Scores........................................................................................ 754-6. Native Speakers: Self-Paced Reading Summary ANOVAs.............................................. 794-7. Native Speakers: Mean Self-Paced Reading Time per Critical Word............................... 794-8. Advanced Speakers: Self-Paced Reading Summary ANOVAs......................................... 824-9. Advanced Speakers: Mean Self -Paced Reading Time per Critical Word......................... 824-10. Intermediate Speakers: Self-Paced Reading Summary ANOVAs..................................... 854-11. Intermediate Speakers: Mean Se lf-Paced Reading Time per Critical Word..................... 854-12. Background Data for Non-Native Speakers...................................................................... 904-13. Vocabulary and Memory Scores........................................................................................ 914-14. Native Speakers: First Pa ss Reading Time Summary ANOVAs....................................... 954-15. Native Speakers: Mean First Pa ss Reading Time per Critical Word................................. 954-16. Native Speakers: Mean Regre ssion-Path Time per Critical Word.................................... 974-17. Advanced Speakers: First Pass Reading Time Summary ANOVAs.................................984-18. Advanced Speakers: Mean First Pass Reading Time per Critical Word........................... 994-19. Intermediate Speakers: First Pass Reading Time Summary ANOVAs........................... 1024-20. Intermediate Speakers: Mean Fi rst Pass Reading Time per Critical Word..................... 1025-1. Verb Bias Example Materials.......................................................................................... 1145-2. Native Speakers: First Pa ss Reading Time Summary ANOVAs..................................... 1205-3. Native Speakers: Mean First Pass Reading Time per Critical Word............................... 121

PAGE 9

9 5-4. Native Speakers: Total Gaze Duration Summary ANOVAs...........................................1225-5. Native Speakers: Mean Tota l Gaze Duration per Critical Word..................................... 1225-6. Advanced Speakers: First Pass Reading Time Summary ANOVAs...............................1245-7. Advanced Speakers: Mean First Pass Reading Time per Critical Word......................... 1255-8. Advanced Speakers: Total Gaze Duration Summary ANOVAs.....................................1275-9. Advanced Speakers: Mean Tota l Gaze Duration per Critical Word................................ 1275-10. Intermediate Speakers: First Pass Reading Time Summary ANOVAs........................... 1285-11. Intermediate Speakers: Mean Firs t Pass Reading Time per Critical Word..................... 1295-12. Intermediate Speakers: Total Gaze Duration Summary ANOVAs.................................131

PAGE 10

10 LIST OF FIGURES Figure page 3-1. Central Role of Noticing ....................................................................................................474-1. Comprehension for Experiment 1...................................................................................... 774-2. Native Speakers: Self-Paced Reading Time...................................................................... 804-3. Advanced Speakers: Self-Paced Reading Time................................................................. 834-4. Intermediate Speakers: Self-Paced Reading Time............................................................. 864-5. Comprehension for Experiment 2...................................................................................... 934-6. Native Speakers: Eye-Track ing First Pass Reading Time................................................. 964-7. Advanced Speakers: Eye-Tracking First Pass Reading Time............................................ 994-8. Intermediate Speakers: Eye-Tracking First Pass Reading Time...................................... 1035-1. Comprehension for Experiment 3.................................................................................... 1195-2. Native Speakers: Total Gaze Duration............................................................................ 1235-3. Advanced Speakers: First Pass Reading Time................................................................. 1265-4. Intermediate Speakers: First Pass Reading Time............................................................. 130

PAGE 11

11 LIST OF ABBREVIATIONS ACC Accusativ e AFS Active Filler Strategy CELSA Combined English La nguage Skills Assessment FDH Fundamental Difference Hypothesis L1 First Language L2 Second Language NOM Nominative NP Noun Phrase PP Prepositional Phrase RC Relative Clause RT Reading Time SLA Second Language Acquisition SSH Shallow Structures Hypothesis SMS Stop Making Sense Task VP Verb Phrase WM Working Memory

PAGE 12

12 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INFLUENCES OF VERBAL PROPERTIES ON SECOND LANG UAGE FILLER-GAP RESOLUTION: A CROSS-METHODOLOGICAL STUDY By Andrea C. Dallas May 2008 Chair: Edith Kaan Major: Linguistics The difficulty of mastering the syntax of a foreign or second language (L2) by adult learners has been the focus of much research. Recent evidence suggests that even advanced latelearning L2 speakers differ as to how they comp rehend complex sentences on-line compared to native speakers and early-learning L2 speakers. On e area in which these differences occur is in the processing of filler-gap dependencies, e.g. Mary saw which toy the child pushed ___while playing., where which toy is the filler, which maintains a relationship between itself and the underlying syntactic position from which it moved ___ (as the direct object of pushed ). These dependencies may cause difficulty for L2 sp eakers because the non-canonical syntactic constituent, the filler, cannot be integrated when encountered, and, thus, must be reconstructed at some stage, the gap, during on-line processing. Prev ious behavioral resear ch has suggested that native speakers utilize knowledge of a verbs transitivity frequency (verb bias) and its subcategorization frame in making on-line d ecisions about possible gap locations while processing wh -dependencies; yet, there ar e few studies looking at L2 on-line sentence processing of filler-gap dependencies.

PAGE 13

13 This study includes three experi ments investigating the type of information late-learning intermediate and advanced Korean L2 speakers of English use while resolving filler-gap dependencies in English. Two of the expe riments also compare results from different experimental methods, self-paced reading and eyetracking. Eye-tracking is used to investigate whether a more natural reading task is better suited to testing L2 speakers. Proficiency is included as a factor to examine whether L2 speaker s of different proficiency levels make use of these verbal properties in a sim ilar manner. The two studies that compare methods examine the use of subcategorization frame, and a third eye-tracking study looks at the use of verb bias in filler-gap resolution. Results across experiments suggest that that intermediate and advanced L2 speakers use these verbal properties in a di fferent manner than native speake rs when resolving filler-gap dependencies. Regarding the use of subcategor ization frame, results suggest that advanced speakers attempt to use this information in an on-line manner, while intermediate speakers do not. A comparison of results across techniques re veals that the two expe rimental methods yield slightly different results for bot h native and non-native speakers. Regarding the use of verb bias, results suggest that intermediate speakers are more sensitive to this factor than native and advanced speakers are when resolving filler-gap dependencies. Overall, results are discussed from a processing demands perspective.

PAGE 14

14 CHAPTER 1 INTRODUCTION Overview The difficulty of m astering the syntax of a fo reign or second language by adult learners has been the focus of much research. One of the pr imary concerns is that, unlike children learning a first (L1) or second (L2) language those who begin learning an L2 after puberty are undertaking the acquisition process after the optimal time fo r achieving an underlying native-like competence (grammar). The arguments as to why this ma y be the case revolve around issues of brain plasticity and the mechanisms involved in acqui ring grammar at a young age versus after puberty (Johnson and Newport 1989, Lenneberg 1967, Paradis 1994, Penfield and Roberts 1959, Ullman 2001). Much of the early research dealing with L2 grammatical competence and this critical, or more cautiously termed sensitive, period has been conducted using off-line tasks; by off-line, I mean tasks, such as grammaticality judgment tasks, that do not directly ta p into real time (online), more automatic processing that occurs as the sentence unfolds, but instead allow time for conscious, controlled analysis to take place (Birdsong 1999, Johnson and Newport 1989, Long 1990). Since the mid 1990s, second language researcher s have begun to turn their attention to whether maturational effects impact late L2 learne rs processing of sentences in real time (Hahne and Friederici 2001, Juffs and Harrington 1995, Weber-Fox and Neville 1996). Through the use of psycholinguistic techniques, questions a bout whether maturational effects influence processing strategies of late-learning L2 speakers are beginning to be i nvestigated, including: What types of information do la te learners use when comprehending sentences on-line? Are these learners as sensitive as native speakers to s yntactic or semantic information as they process sentences or are they just sl ower? If these learners do not utilize native-like processing

PAGE 15

15 strategies, does the L1 that they speak influenc e strategy use or does more experience with the L2 matter? Goals of the Study The aim of this dissertation is to examine factors that influence L2 on-line sentence processing. Specifically, it investig ates the types of information la te-learners of English with a Korean native language (L1) background use when comprehending complex English sentences. The complex sentence structure being explored is referred to in the literature as a wh -dependency or a filler-gap dependency. These structures c ontain non-local dependencies, such as those in relative clauses and wh -questions (e.g., The elephant that the boy rode___ was no longer at the zoo, and The teacher saw which girl the boy pushed ___yesterday). Looking at the second example, which girl maintains a non-local relationship be tween itself and its underlying syntactic position in the sentence as the direct object of pushed indicated by a ___. These dependencies may cause difficulty for L2 readers because they contain a syntactic constituent that does not appear in its canonical position cau sing a burden on the system as th e filler cannot be integrated when encountered. To date there has been little research conducted on L2 processing of filler-gap dependencies. While there has been growing inte rest in L2 on-line se ntence processing, many studies have focused on what types of verbal in formation are pertinent and/or transferred when L2 learners process syntactically ambiguous sentences. Fewer studies have looked at the issue from a developmental perspective; and even fewe r have investigated L2 on-line processing of filler-gap dependencies (Marinis et al. 2005, Williams et al. 2001, Williams 2006). Previous monolingual research has suggest ed that native speakers ut ilize knowledge of a verbs transitivity frequency (verb bias) and subcategorization frame in making on-line decisions about possible gap locations while processing wh -dependencies (Crain and Fodor 1985, Fodor 1989,

PAGE 16

16 Goodluck et al. 1991, Stowe 1986, Tanenhaus et al. 1989). Whether or not L2 speakers utilize these verbal properties in the sa me manner while processing fille r-gap dependencies has not been previously investigated. Three e xperiments are included in this di ssertation to investigate the use of verbal properties in L2 filler-gap resolution. The main hypothesis being tested across all three experiments in this dissertation is that L2 speakers with lower levels of proficiency are expected to be less sensitive to the lexical properties of verbs due to greater processing demands associated with resolving filler-gap dependencies. Such results would argue for a developmental L2 parser that uses a hyper -active strategy of gap-filling; meaning th at lower proficiency L2 learners dump the filler at the first verb encountered, regardless of any lexical information suggesting otherwise. Higher proficiency L2 speakers are expected to pattern si milarly to native speakers, in that they will be sensitive to a verbs lexical pr operties during filler-gap resolu tion. Higher proficiency speakers should experience less of a proce ssing burden, allowing the parser to be influenced by the lexical properties under examination. By incorporating proficiency as a factor, the results will provide information as to how the L2 parser develops ov er time. The results also will be analyzed in terms of whether they fit within a recently pr oposed model of L2 heuristic processing (Clahsen and Felser 2006). This model argues that L2 sp eakers make use of a processor that computes shallow predicate-argument structures based on lexical, pragmatic, and world knowledge. Such a model suggests that all L2 speakers, regard less of proficiency, shoul d be guided by lexical properties. From another perspective, L2 processing of filler-gap dependencies may be influenced by processing burdens guiding both advanced and inte rmediate L2 speakers to use a hyper -active strategy where gaps are filled as soon as possible. If differences in proficiency

PAGE 17

17 levels emerge, the results would suggest that L2 parsing strategies change as proficiency increases. A secondary aim of the project is to co llect and compare da ta using different psycholinguistic methods: self -paced reading and eye-track ing. There is currently a methodological debate as to which method of psycholinguistic research is most adequate for investigating L2 sentence proces sing (Frenck-Mestre 2005). L2 use of subcategorization frame is being investigated using self-paced reading and eye-track ing. The cross-methodological design will allow for a comparison of techniques and will demonstrate whether the less sensitive technique of self-paced readi ng is useful for informing a model of L2 processing. Organization This disse rtation is organized in the followi ng manner. Chapter two presents an overview of research dealing with both L1 and L2 sent ence processing; focusing on the role verbal properties, namely subcategorization informa tion and verb bias, play in comprehending sentences, particularly filler-gap dependencies. Chapter three presents a cognitive perspective of second language acquisition, which is included to provide a theo retical perspective of second language acquisition, and more importantly to offer possible reasons why processing complex sentences causes difficulty for adult learners. Chapters four and five are each devoted to presenting experimental findings of the three ex periments undertaken in this work. Finally, chapter six summarizes the findings and discusses the implications and limitations of the total project. Future research di rections are also provided in this final chapter.

PAGE 18

18 CHAPTER 2 NATIVE AND NON-NATIVE SE NTENCE PROCESSING Introduction A rich body of literature exists regard ing sent ence processing for native, monolingual (L1) speakers. Research on non-native (L2) sentence processing is a relatively new endeavor, and as such, the body of literature on this topic is not as extensive. Where distinct models of L1 sentence processing have been put forth, L2 sentence processing models are not as elaborate. Instead, most research focuses on whether L1 and L2 sentence processing is different, and if so, how it differs. Many studies have examined what types of verbal information are pertinent and/or transferred when L2 lear ners process syntactically ambi guous sentences. Fewer studies have looked at the issue from a developmental pers pective; and even fewer have investigated L2 on-line processing of fille r-gap dependencies (e.g., Joan saw which girl the boy pushed ____ yesterday where which girl is the filler, and ___ represents th e gap). Previous monolingual research has suggested that na tive speakers utilize knowledge of a verbs transitivity frequency (verb bias), and its subcategoriz ation frame in making on-line decisi ons, either initially or during revision, when resolving filler-gap dependencie s (Crain and Fodor 1985, Fodor 1989, Garnsey et al. 1997, Goodluck et al. 1991, Pick ering and Traxler 2003, Stowe et al. 1991, Tanenhaus et al. 1989, Trueswell et al. 1993); whether late-learning L2 learners util ize the same verbal properties in the same manner while processing filler-gap dependencies is currently undetermined. The goal of this chapter is to present an overview of the cu rrent research on L1 and L2 filler-gap resolution; more specifically, the discussion concerns the role that verbal properties play during filler-gap resolution. The discus sion does veer at times from filler-gap dependencies, yet it largely addresses research on this structure. The chapter is organized by first exploring L1 models of sentence processing Within this section, a strategy is presented

PAGE 19

19 suggesting that gaps are filled as soon as possible by native speakers. The second section reviews L2 research examining whether L2 speake rs use the same strategy to resolve filler-gap dependencies, and whether full syntactic representations are built during L2 processing. This second section concludes with a discussion of a recently proposed model of L2 processing adhering to the use of heuristi cs. A summary of important fi ndings concludes the chapter. Native Sentence Processing Most researchers in terested in on-line pro cessing would agree that in broad terms, comprehension is constructed through the use of syntactic, lexical, and discourse information. This broad statement, however, does not reve al how or when each type of information contributes to comprehension. To this e nd, researchers have debated over the relative dependence or independence of s yntactic, lexical, and discourse information during processing. Different theoretical perspectives, different expe rimental techniques and the variation in results that those techniques yiel d, as well as in the interpretation of what these results mean, have all contributed to the formulation of two broad a nd somewhat incompatible models of sentence processing. This debate has part icularly focused its attention on lexical properties associated with verbs. As aptly stated in Friederici and Frisch (2000), an overwhelming amount of research has been conducted examining how a nd when verbal properties influence parsing decisions when resolving syntactic ambigu ity and filler-gap dependencies (pg. 477). The verbal properties that have been most investigated are a verbs subcategorization frame, a verbs thematic role assigning capab ilities and a verbs argument structure frequency (verb bias). The term subcategorization frame refers to the constituent(s) a verb takes as its complement(s), such as noun phrase (NP), prepositional phrase (PP), or nothing. For example, a verb like put occurs in a sentence like: She put the candle on the table. The subcategorization frame for put would be: put V, [ __ NP PP]. The same verb cannot occur in: *She put the

PAGE 20

20 candle. Thus, the PP is a required argument for the verb put Phrasal constituents have been considered in much of the pro cessing literature to be a syntactic property of the verb because they define the syntactic category of a verbs arguments. A verbs thematic role assignment has been considered a lexical propert y associated with the verb. Ta king the same verb and examples above, the event structure of put expresses an event involving th ree participants: an agent ( she ), the one performing the action, a patient ( candle ), the thing undergoing the action, and a location ( on the table ). This information is lexically related to the verb and represented in the thematic roles that a verb assigns; it is also known as selectional restriction information because some thematic roles require that they are assi gned to semantically-restricted lexical items1. Not all verbs have absolute subc ategorization frames, as with put. There is also itemspecific information associated with a verb conc erning the frequency with which it appears with phrasal constituents; this is referred to as verb bi as. Verb bias refers to the frequency with which a particular verb appears with a given argument stru cture. An example of a verb with a transitive bias is push, because it occurs most frequently in c onstructions with NP direct objects (e.g., The boy pushed his bike .); whereas a verb with an intransitive bias would be crash because it occurs most frequently in constructions without NP direct objects (e.g., The driver crashed .) (Gahl et al. 2004). Verb bias is considered a lexical property of a verb becau se this information is encoded within a given verbs argument structure. Verb bias differs from plausibility information. For example, the verb push occurs most frequently with NP dire ct objects, but not all NP direct objects are equally plausible (e.g., The man pushed the river ). So, while both types of information constitute lexical information, plau sibility information is not encoded within a 1 P lausibility may interact with a verbs thematic role assigning capabilities in that agents are typically animate. Plausibility could, however, just be related to semantic proper ties of lexical item, as in the children read the song. Typically, songs are sung rather than read (example from Stowe et al. 1991).

PAGE 21

21 specific verbs lexical entry and must be computed in real time on a case by case basis. It is important to note that in this di ssertation, verb bias is being used to refer to the frequency with which a verb takes a given argument structure, while the term subcategorization frame is being used to refer to verbs that have an all-or-nothing requirement, thus, for verbs that require constituents and result in ungrammatical se ntences if those requirements are not met. The debate in the L1 sentence processing li terature has focused on how dependent the parser is on these verbal propert ies when building representations for comprehension. Added to this debate is also the question of whether discourse information su ch as previous context, visual information and real world knowledge play a role in the parsers decision to build a given structure. The main questions that each th eoretical perspective tries to answer are: Does syntactic processing have special status? Can lexical and/or discourse in formation guide parsing decisi ons (have initial influence)? If not, at what point does this in formation become available? When does real world knowledge come into play? Are multiple structures built and evaluated simultaneously? Is a full syntactic represen tation built at all? Sentence Processing Models In this section, two basic pers pectives of sentence processi ng, the m odular perspective and the lexical/interactive perspective, are described. A brief account is provided as to where each perspective stands in relation to the role of syntactic and lexical information in processing with a focus on verbal properties; and st udies are discussed representing bot h perspectives. In order to describe the differences between the two models research looking at ambiguity resolution is presented. Lastly, a heuristic approach to parsing provides a contrast to these two main approaches. From a modular perspective, sy ntactic processing is consid ered to be independent and initially impervious to information processed by other modules, such as the lexical module or the

PAGE 22

22 discourse module (Ferreira and Clifton 1986, Fr azier 1979, Frazier 1989). The motivation for such separation of modules in processing lies in a modular architecture of language, where each module is encapsulated (Frazier 1989). Applied to language processing, this means that on-line comprehension is first based on syntactic inform ation and principles, su ch as phrase structure rules (e.g., S NP VP); and minimal attachment, a prin ciple which guides the parser to build the simplest syntactic structure. As incrementa l processing proceeds, word category information is used by the parser to build its syntactic representation based on wh at is syntactically permissible, regardless of whether lexical information about the specific word suggests otherwise. The syntactic output is then compared with item-specific information to ensure the correct analysis. If the initial structure built is incorrect, revi sion takes place. For this reason, the models that adopt this theoretical perspective have been referre d to as a serial model, a twostaged model, and a garden-path model (Trueswell et al. 1994). Timing issues are central in a modular model. Empirical evidence has to demonstrate that there is no initial influence of lexical informati on (including semantic prope rties associated with the verb) or discourse information on what stru cture is built. A modular model also has the burden of demonstrating when these other types of information beco me available to the parser. These timing issues become particularly hard to untangle when the parser encounters a verb due to the wealth of lexical c ontent encoded in the verb. Although models within the lexical or intera ctive perspective vary slightly, multiple sources of information (or constr aints) are argued to influence st ructural choices during parsing (Boland 1997, Trueswell et al. 1994, Tyler 1989). This perspective is based on the premise that various aspects of a lexical representation are pr ocessed in parallel. Unlike the modular model, lexical and discourse properties including frequency, can have an immediate influence on the

PAGE 23

23 structure the parser builds. Many models with in this category advocate the building and ranking of multiple possible structures (Boland 1997, Fr azier 1989, MacDonald et al. 1992, Trueswell et al. 1994, Trueswell 1996). Ranked constraints (e.g., c ontext, plausibility, s ubcategorization type, verb bias) contribute to structur al preferences. Lower ranked struct ures are kept activated, but at lower levels. If the preferred structure turns ou t to be incorrect, less preferred structures are easily adopted. These two types of models are difficult to diffe rentiate because at times they predict the same empirical results. To illustrate these predictions, consider the sentence: While Anna dressed the baby played in the crib. From a modular perspective, the syntactic ambiguity of the baby should cause difficulty for a reader at played This type of difficulty is referred to as garden-pathing, which means that the first part of the sentence leads a reader in the wrong direction, structurally speaking. The modular model predicts played will cause a reader to garden-path because the baby was attached as th e direct object of dressed since the parser prefers to build the simplest structure. From a lexically-based perspective, played would still cause a reader to garden-path, but the r easoning would differ. The attachment of the baby as direct object is due not to the pa rser building the simplest syntactic structure, but rather to the parsers knowledge that dressed most frequently takes a direct object. This example should demonstrate that teasing these models apart can be a difficult empirical task. There are instances, however, where the models make different predictions. To illustrate these differences, we can look at a slight ly different example, While Anna hurried the baby played in the crib In this case, the modular model would still predict garden-pathing at played but the lexically-based model would not. The reason why the lexically-bas ed model would not predict difficulty is that the parser takes into account the fact that hurried prefers to occur withou t a direct object. As

PAGE 24

24 pointed out in Garnsey (1997), researchers have conducted several crosstechnique comparisons, and results have been mixed as to whether le xical and discourse information can influence immediate decisions or not. One study that found that selectional restrict ions do not play an immediate role in ambiguity resolution is Ferreira and Clifton (1986). This stu dy monitored eye-movements with reduced and unreduced relative clauses (RCs) beginning with eith er animate or inanimate nouns, compared to non-ambiguous equivalents (reg ions of interest underlined). (1a) The defendant examined by the lawyer (1b) The defendant that was examined by the lawyer (1c) The evidence examined by the lawyer (1d) The evidence that was examined by the lawyer Their motivation was to support the modular claim that animacy (a se lectional restriction, in this case) does not inform initial syntactic structure wh en there is ambiguity present. This would be supported if both reduced RCs cause readers to ga rden path, which would mean that animacy has no effect on initial parsing. A lexically-ba sed model would hypothesize that only animate reduced RCs (1a) would cause trouble because the verbs selectional rest riction prevents the inanimate reduced RC (1c) from garden-pathing. First pass and second pass reading times (RT) at the NP, V, and PP (underlined) were examined The results indicated the best case scenario for a modular model: RT for both Reduced RCs (1a and 1c) were longer regardless of animacy; there was no interaction between animacy and ambiguity as a lexically-based model would predict; and there was no increase in RT at di sambiguating PPs in (1c), which means (in)animacy had no effect. Clifton et al. (2003) used a slightly more sophisticated methodology and found similar results. The response from the opposing perspective was that Ferreira and Clifton (1986)s materials were faulty (Trueswell et al. 1994). Many of the nouns in th e inanimate NPs could

PAGE 25

25 legitimately be subjects of senten ces; meaning that regardless of animacy, they still met at least one of the verbs thematic requirements (e.g., The car towed the boat. The car towed by the truck was mine. Cars can tow and be towed even though they are inanimate.). The materials seemed to have purposely led subjects down the garden-path. If all of the nouns had been equally bad as subjects of main clauses, th en it would be clearer whether thematic role information was being taken into account. A nother complaint lodged against Ferreira and Clifton (1986) was that there were some unnatural line breaks in the display mode, which could have introduced differences in RT between re duced and unreduced RCs. Furthermore, they claimed that reduced versus unreduced RCs do not produce a good baseline for comparison; as eye-movements differ for unreduced RCs due to the presence of function words. These problems led Trueswell and colleagues (1994) to replicate Ferreira and Clifton (1986) with three improvements. First, inanimat e nouns in their materials were all inconsistent with a main clause reading; thereby improvi ng the confound between whet her it was the verbs thematic role information rather than animacy that caused (1c) to garden-path. They also improved the display mode, such that all scoring regions were presented on a single line. The third improvement was that unambiguous passive participles (e.g., drawn) were used as a baseline comparison. (2a) The defendant examined by the lawyer (2b) The defendant that was examined by the lawyer (2c) The evidence examined by the lawyer (2d) The evidence that was examined by the lawyer (2e) The poster drawn by the illustrator (2f) The poster that was drawn by the illustrator The modular model predicts that all reduced RCs, (2a), (2c) and (2e), should be more troublesome than the unreduced RCs (regions of in terest underlined); in other words, animacy

PAGE 26

26 has no effect on initial parse. A lexically-bas ed model predicts that (2a) should be more troublesome than (2c) and (2e) due to the contribution of examined and drawn s selection restrictions; in other wo rds, animacy matters in initial parse. In 1st pass reading times, animate reduced RCs (2a) took longer to read at the by phrase than inanimate reduced sentences; only the verb position of the inanimate reduced (2c) showed a slight increase, but the same pattern emerged with the unambiguous inanimate (2e) and (2f). The problem with both of the studies above is whether they genuinely distinguish between these two theoretical perspectiv es. Proponents of the modular perspective could respond that Trueswell et al.s results do not demonstrate that various lexical factors have an immediate effect, as they analyzed first pass reading times, not first fixation duration. First fixation duration is a measure of the first time the eyes fixate on a word, whereas first pa ss reading is the sum of all fixations within a region before a saccade is made to another region (Boland 2004, Rayner and Sereno 1994, Reichle et al. 2003); thus, it may be argued that the results are based on later effects. At the same time, the results seem to indicate that if not immedi ate, these factors come into play quite rapidly. The inc onclusive evidence listed above is representative of a series of articles that have employed slig htly different techniques to ar gue for one perspective over the other when it comes to ambiguity resoluti on (Blodgett and Boland 2004, Boland 1993, Clifton et al. 2003, Clifton et al. 1991, Ferreira and Henders on 1990, Garnsey et al. 1997, Trueswell 1996). More recently, the visual world paradigm has been used to investigate these issues, as well as whether non-linguistic contextual information guide s initial parsing decisions. The results from these experiments seem to indicate that visual in formation has an immediate impact on structural choices (Knoeferle et al. 2005, Sedivy et al. 1999, Tane nhaus et al. 1995).

PAGE 27

27 The commonality in the models of sentence processing discussed thus far is that they assume that a complete and detailed representati on is built during on-line processing, regardless of what influenced the building process and when (Ferreira 2003). A relativ ely recent model, the heuristic model, proposes that sh allow representations (both synt actically and se mantically) are built during on-line processing (Fer reira et al. 2002). Heuristi c-based accounts use a quick-anddirty approach to parsing based on semantics a nd parsing habits, such as using an N-V-N or Agent-Action-Patient strategy in English. This model argues that readers do not build algorithmic, full syntactic representations for se ntences, but rather rely on a superficial, goodenough interpretation, which in some cases ma y turn out to be wrong. Within such a perspective, the parser is still capable of algorithmic parsing, but th is may happen either at a later stage or when the parser runs into trouble (Townsend and Bever 2001, cited in Ferreira 2003). Ferreira (2003) criticizes prev ious research, including her ow n, on the grounds that little information has been gathered about the exact in terpretations readers form. Furthermore, she proposes that the language proce ssor might process sentences heur istically due to time demands associated with on-line comprehension. Intuitiv ely, such a perspective makes sense, at least some of the time. Many people have the experi ence of realizing they interpreted a sentence incorrectly after the fact especially when pressed for time. A heuristic parser has also been recently proposed as a framework from which to base L2 sentence processing (Clahsen and Felser 2006a). Again, this makes sense to some degree given that L2 learners do not have complete competence of the target language and that this models main premise is that full representations are not being built. I continue the discussion of an L2 heuristic model in the section of the paper dealing with non-native sentence processing.

PAGE 28

28 Native Processing of Filler-Gap De pendencies: Active Filler Strategy One proposed strategy for how the parser reso lv es filler-gap dependencies is known as the active filler strategy (AFS) (Clifton and Frazier 1989, Fodor 1989, Frazier and Clifton 1989). Using the AFS, the parser first identifies the fi ller and then actively seeks to assign the filler to the first possible position in a syntac tic structure. Given the sentence, Who did Jane see ____Max with ____?, if the parser adheres to the AFS, it w ill first try to use the filler as direct object of see but when it encounters Max, it must revise its analysis After revisi ng its original analysis and continuing to actively search, the filler finds a suitable gap, or empty NP, to fill when it encounters the preposition with Sentence processing models agree that for L1 speakers filler-gap dependencies are resolved as soon as possible, as opposed to waiting until the end of the sentence for disambiguating information. The primary evidence for this comes from studies which have demonstrated that parsing can be di srupted if the expected gap in a wh -question is filled with an object NP (Crain and Fodor 1985, Stowe 1986): (3) Who had the little girl expected us to sing those stupid French songs for ____ at Christmas? filled-gap gap The posited gap in (3) follows the verb expected. When us is encountered in this gap position, there is a disruption in parsing re flected in increased RT, resulting in what has been termed the filled-gap effect; thus, the filled-gap effect is an effective tool that demonstrates whether readers have filled the gap at the verb. While sentence processing models agree that gaps are posited as soon as possible, they differ in terms of what types of information they view as capable of having an immediate influence on whether the parser pr edicts a gap or not (Boland 1997). For supporters of a modular model, the parser actively posits gaps based on s yntactic information in the structure (Frazier and

PAGE 29

29 Flores D'Arcais 1989). This means that lexical information about specific verbs and discourserelevant information does not play an immediate role in the prediction process. A lexically-based model, however, would argue that verbal properties play an active and influential role in the parsers predictions; and a constraint-based model would argue that several types of information should be available to aid the parser in making initial decisions in filler-gap resolution (Trueswell et al. 1994, Tyler 1989). Not surprisingly, one area where the debate has focused its attention is on properties of verbs, since verbs are often the critical location where the parser first posits a gap (at least in English). Stowe et al. (1991) found that verb bias in fluences whether the parser predicts an upcoming gap at the verb. In particular, they f ound that the parser does not predict a gap after verbs that occur more frequently as intransitives. To test this, self-p aced reading was used and filler plausibility and verb bias were manipul ated. Self-paced reading is an experimental technique that yields RT per word or per fragme nt. The recorded RT allows researchers to see how long participants spend reading a critical word or fragment w ithin a sentence. Differences in RT for critical words or regions are then compar ed for various conditions in the experiment. This technique assumes and exploits the incremental nature of reading; thus, if at a given point in the stimulus, readers encounter a word or region that does not fit their st ructural or semantic expectations or that leads to an increase in complexity, an increase in RT is expected on that critical word or soon after (s pillover effects) (for furthe r discussion see chapter four methodology). Returning to Stowe et al. (1991), items 4a-4b have a transitive verb bias (intransitive bias not shown). (4a) She wondered which book the students read ___ duri ng their class. PLAUS (4b) She wondered which son g the students read ___ duri ng their class. IMPLAUS filler gap

PAGE 30

30 The logic of manipulating filler plausibility was that if the parser was immediately influenced by verb bias, then gaps should not be posited af ter intransitive verbs fo r either plausible or implausible fillers. Indeed, these were the findings: A plausibility effect was found for transitive bias verbs (RT increases at the verb when the filler was im plausible), but not for intransitive bias verbs. In addi tion to indicating that the parser is sensitive to verb bias, this research also demonstrates that manipulating the plausibility of the filler is a useful tool for investigating gap filling. While Stowe et al. (1991) s uggest that verb bias has an immediate influence over whether the parser posits a gap, other research does not support the same conclusions. Pickering and Traxler (2003) argued that the parser does not use verb bias initi ally to posit and fill upcoming gaps. Three experiments, two self-paced reading and one eye-tracking, were conducted that led to the above conclusion. Unlike self-paced readin g, where a reader controls the pace of word presentation, eye-tracking records eye fixations as a reader moves her eyes across a complete text, replicating more natural read ing. Eye-tracking results in a ri cher data set (Reichle et al. 2003), and has been argued to be better capable of determining more precise timing to the onset of difficulties in on-line processi ng (Boland 2004, Frenck-Mestre 2005). The materials tested in the experiments of Pickering and Traxler (2003) differed from those in Stowe et al. (1991), in that Pickering and Traxler varied the frequency with which a verb subcategorizes for an NP or PP. Below is a PP-preference item, as determined by two norming studies (/ indicate regions analyz ed in one of the self-paced read ing experiments and in the eyetracking experiment): (5) Thats the plane / that the pilot /landed_____ carefully/ behind _____ in the fog / at the airport. filler false gap gap

PAGE 31

31 For all three experiments, they found that an ob ject interpretation of the filler was initially preferred; that is, even when the verb had PP-preference (5), plausibility effects were found, indicating that the gap was filled at the verb. Such results are evidence against the claim that verb bias information is immediately influential in predicting gaps, as there should have been no plausibility effect if the parser was influenced by verb bias. In a series of three experiments, one in which filler-gap dependencies are investigated, Staub (2007) replicated aspects of Pickering and Traxler (2003), but found that while verb bias does not matter to the parser, strict intransitivity does; thus, it may be that the case th at in the Stowe et al. (1991) materials some of the intransitive bias verbs we re rarely used transitively. Exploiting filled-gap effects, Goodluck et al. (1991, 1995) found that a verbs subcategorization frame requirements matter to the parser as they make predictions about potential gap locations. This was done using self -paced reading and comparing filled-gap effects with put/push verbs, which have different argum ent structures (6a, 6b). (6a) John wondered what the little boy had put ____the b ook into ____ the desk for ___ before class ended. false gap false gap gap (6b) John wondered what the little boy had pushed ____the book into _____ the desk for ____ before class ended. false gap false gap gap In (6a), the verb put requires two arguments, an NP and a PP; whereas, in (6b) push only requires an NP. The results were that larg er filled-gap effects were found for push-items directly after the verb, whereas for put -items, larger filled-gap effects were found directly after the first preposition, which followed the direct object filled gap; that is, for push, the parser had no obligatory subcategorization arguments left as potential gap locations after it encountered the direct object; for put on the other hand, the parser did have an obligatory argument as a potential gap location, the object of the required prepos itional phrase. Goodl uck et al. (1991, 1995) interpret these findings as a completion of proposition effect; meaning that, because the

PAGE 32

32 proposition of push is complete at the direct object, the pars er prefers the gap to be located at this point, which is why the effects are in a different location for put verbs, which are completed later. Another way to interpret these results is that they suggest that subcategorization frame requirements are actively available to aid in fo rmulating or reformulating gap predictions. To summarize, research on L1 processing of filler-gap dependencies suggests that for L1 speakers of English, gaps are actively posited by the parser, and that verb bias and verb subcategorization frames are used rapidly in this process. Non-Native Sentence Processing Given the findings of the resear ch reviewed in the previous section, the L 1 parser appears to use subcategorization frame and verb bias to resolve filler-gap dependencies. It is unclear, however, whether language learners would demonstr ate the same patterns. The L2 research in on-line processing has largely focu sed on issues of ambiguity reso lution (Heredia and Altarriba 2002), particularly the issue of high versus lo w PP attachment and garden-path sentences (Dussias and Cramer 2006, Felser et al. 2003, Fernandez 2002, Ferna ndez 2003, Frenck-Mestre and Pynte 1997, Hoover and Dwivedi 1998, Ju ffs and Harrington 1996, Juffs 1998, Juffs 2004, Papadopoulou 2005). Relatively few studies have considered L2 processing of filler-gap dependencies (Felser and Roberts 2007, Juffs 2005, Love et al. 2003, Marinis et al. 2005, Williams et al. 2001, Williams 2006), only two of whic h look directly at what types of verbal information is used during L2 on-line processing. While most L2 on-line research indicates that late-learning L2 speakers rely more on lexical and pragmatic properties of language to process sentences, it is currently unknown which verbal pr operties influence filler-g ap resolution. It is also unknown whether L1 transfer and L2 proficie ncy level play a role in whether L1 knowledge or parsing strategies influence L2 processing.

PAGE 33

33 Non-Native Processing of Filler-Gap Dependencies Active filler strategy The non-canonical ordering of constituents in fi ller-gap dependencies m ay create an added burden on the L2 processing system, which c ould cause difficulty during processing. L2 processing has been shown to be slower than native processing, even among highly proficient speakers (Segalowitz 1986, Skehan 1998). Curren t research suggests that L2 speakers do actively posit gaps as they proce ss filler-gap dependencies in real time, but that they may not be influenced by lexical and syntactic information in the same manner as L1 speakers. Research indicates that at least one of the lexical factors that L1 speak ers use to fill gaps, plausibility, may be available and used on-line by L2 speakers, if they focus on it. This section discusses the issues related to the use of an AFS by L2 speakers. One early L2 study on filler-gap dependencies, which suggests that L2 speakers do utilize an AFS, focused on subject versus object movement (Juffs and Harrington 1995). Examples are given below: (7a) What does Mary believe John teaches ______? object gap (7b) Who does Mary believe ______ teaches linguistics? subject gap The primary motivation of this research was to use an experimental tech nique previously unused in L2 research, self-paced readi ng, to see if it could provide more insight than offline measures as to why subject movement caused more difficulty for some late L2 learners (White and Juffs 1998). While the overall results of Juffs and Ha rrington (1995) suggest that these L2 learners had a processing difficulty and not a competen ce problem with subjec t extraction, this is tangential to the current discussion; what is important is that the results suggest th at L2 speakers appear to be actively predicting gaps. Increases in RT, possible integration effect s, found at the

PAGE 34

34 verb teaches in (7a) compared to (7b), were argued to reflect expectation of upcoming gaps on the part of the L2 readers. There were some problems with this study, however. Juffs (2005), reporting an improved replication of Juffs and Harringt on (1995), points out that some of the analyses were conducted on word final positions. Measuri ng RTs at word final positions al so taps into sentence wrap up effects, and, thus, is a confound that should be avoided in experimental paradigms. More importantly, as Williams et al. (2001) point out, Ju ffs and Harrington (1995) did not present clear evidence of the use of an AFS; that is, it canno t be determined whether readers were using the AFS or whether they were waiting until the end of the sentence to resolve the dependency. As Juffs and Harrington (1995) show weak support for the L2 parser adhering to the AFS, the need for more direct evidence motivated Williams an d colleagues to investigate this issue further (Williams et al. 2001, Williams 2006). Results from Williams et al. (2001) do provide stronger support for the use of an AFS by advanced late-learning L2 speakers. To test the us e of the AFS, Williams et al. (2001) used a stop-making-sense task (SMS) and manipulated plau sibility to examine filled-gap effects in nonnative speakers of English. Plausible Filler (8a) Which girl did the man push _______ the bike into ___ late last night? filler false gap gap Implausible Filler (8b) Which river did the man push _______ the bike into ___ late last night? filler false gap gap In an SMS task, participants read sentences one word or fr agment at a time by pushing a button. As each word or fragment is revealed, the previous word disappears allowing for RT per word or fragment to be recorded, as in self-p aced reading. Once a word or fragment causes the

PAGE 35

35 sentence to stop making sense, pa rticipants are instructed to pr ess a different, reject, button. This technique is capable of revealing the location at which participants consider the sentence to stop making sense, as well as revealing difficulties of which the participants may or may not be consciously aware throu gh increases in RT. Use of the AFS by both L1 and L2 speakers was supported by an increased number of reject responses at the verb fo r implausible condition (8b). In terestingly, there were no RT differences found on the verbs for the different conditions. For both L1 and L2 speakers, an increased RT did occur on the filled-gaps ( the bike) in the plausible condition, which suggests that the gap was already filled at the verb and that it was harder to recover if the filler was successfully integrated at the verb. The L1 gr oup, however, did perform differently at the filledgap depending on condition. For L1 speakers there was a difference in RT at the determiner preceding the filled-gap noun ( the bike) in the two conditions (RT was slower at the determiner for implausible conditions); there was no differe nce found for the L2 speakers at this same position. This was argued to suggest that native speakers are sensitive to syntactic cues, and that the L2 readers are insensitive to such syntactic cues, regardless of native language background. Overall, these results suggest that both L1 and L2 speakers are using the same parsing strategy, AFS, and that both groups are sensitive to plausibility. However, two problems with these conclusions are that the task (SMS) coul d have inflated reading times across all groups, masking any small differences, and that it could have been leading L2 sp eakers to actively posit gaps when they might not in a more naturalistic task (as pointed out in Williams, 2006). A follow up study was conducted and found that task-specific demands influenced the performance of both L1 and L2 speakers (Williams 2006). This follow-up study increased the number of words between the postverb determiner and the noun (e.g., the very noisy motorbike)

PAGE 36

36 to examine how quickly L2 speakers recognized the filled-gap, and compared results from a SMS task to those of a self-paced reading task. Across both experiments, all participants were of advanced proficiency, as determined by a proficiency measure and an extensive self-assessment questionnaire. For the SMS task, the results were similar to Williams et al. (2001): Both L1 and L2 groups made more reject decisions in the region preceding motorbike ; thus, all groups were argued to be equally sensitive to structural and non-structural information. The results from the self-paced reading experiment revealed that L2 pa rticipants took longer to read overall, but there was no difference in the pattern per region. More interestingly, there were no plausibility effects for any of the participants, incl uding the L1 participants. Th ere was a comprehension/memory task which indicated that both L1 and L2 speak ers did better at rememb ering the implausible conditions, suggesting that there was some effect of plausibility. Further analysis was conducted by splitting participants into two groups depending on how well they performed on the memory task. For L1 speakers, the good performers showed an effect of plausibility at the determiner ( the) and intensifier ( very), while the poor performers showed effects at the noun. When they did the same analysis on the L2 speakers, a differe nt pattern emerged. The good performers showed plausibility effects at the pr eposition following the noun, while the poor performers showed no plausibility effects at all. Williams argues that some L2 speakers are not able to apply plausibility information during on-line processing of filler-gap dependencies. Results clearly differed depending on task: The results from the SMS task indicated that both L1 and L2 speakers fill gaps immediately (a nd use plausibility information to compute or reanalyze filler-gap dependencies). This mean s that when the task focuses attention on plausibility, plausibility info rmation is used, but when the task does not focus attention on plausibility, then not all reader s use this information. Williams argues that some L2 readers are

PAGE 37

37 not accessing full semantic representations of fill ers, and that it may be difficult for them to access plausibility information on-line; which coul d be due to a variety of reasons. Williams (2006) interpretation of th e results is related to differences in degrees of automaticity; that is, some L2 speakers may be attending to lower orde r processes, such as lexical access, and thus, are more inefficient at applying lexical info rmation when not focusing on it. Taking the Williams (2006) interpretation a step further, it might be argued that some L2 speakers rely on a strategy leading them to dump the filler at th e first verb encountered, regardless of conflicting lexical information. In any case, it is clear fr om the Williams and colleagues studies that more work needs to be done in this area. To my knowledge, there are no published studies looking at the influen ce of verb bias or verb subcategorization frame during L2 filler-gap resolution There is, however, one recent study looking at ambiguity and the use of verb bi as and another at the use of subcategorization frame in L2 ambiguity resolution (Dussias and Cramer 2006, Frenck-Mestre and Pynte 1997). Frenck-Mestre and Pynte (1997) used eye-tracking and found that verb subcategorization frame influenced how L2 speakers resolved ambiguities re lated to prepositional phrases. These results were used to argue that L2 speakers may be driven more by lexical properties of the verb compared to L1 speakers, as earlier effects we re found for L2 speakers. Dussias and Cramer (2006) found that L2 speakers were influenced by L2 verb bias during ambiguity resolution if the speakers had learned those biases. Knowledge of L2 verb bias was determined by an offline task. Such results suggest that verb bias might play a role in L2 filler-gap processing. If verb bias does play a role, it would mean that when processing a filler-gap dependency with an intransitive-bias verb, L2 readers should use th is information to either prevent them from positing gaps at the gap s ite (the verb itself) or to aid in revision.

PAGE 38

38 Use of syntactic traces dur ing filler-gap resolution Overlaid upon the issue of which lexical or s yntactic cues native sp eakers m ay be using to help them predict potential gap locations is the is sue of whether or not abst ract syntactic gaps are being used by the L2 parser when the parser resolves the filler-gap dependency. Two opposing theories as to how filler-gap dependencies are resolved are the direct association hypothesis (DAH) and the trace reactivation hypothesis (TRH). Within the DAH, a filler is directly associated or integrated with its subcategorizer (e.g., verb) when that subcategorizer is encountered, regardless of the proposed gap location (Featherston 2001, Pickering and Barry 1991). This differs from the TRH, which assumes that the parser relies on abstract syntactic elements (i.e. structural traces) for resolving depende ncies. What this means is that at the trace position (the gap), abstract features of the f iller, which are stored in working memory, are reactivated and the abstract trace is then used to resolve the de pendency with its subcategorizer (Featherston 2001, Nicol and Swinney 1989, Swinney et al. 1989). Some L1 research suggests that L1 speakers may use structural traces when processing filler-gap dependencies, whereas L2 research suggests that abstract syntactic trac es are not used by L2 speakers when processing filler-gap dependencies (Felser and Roberts 2007, Marinis et al. 2005). Instead, L2 speakers appear to be using simple heur istics to create shallow meani ng-based representations (discussed below) (Clahsen and Felser 2006a). To explain the difference between how L1 and L2 speakers may resolve filler-gap dependencies, it is necessary to look at a dependency that is st ructurally more complex than those examined above. All of the non-local dependencies discu ssed thus far have focused on dependencies contained within a single syntactic clause. While the linear distance between the filler and the gap has varied, there have not been additional syntactic barriers in the structures. Consider the filler who in (9), which has been extracted from an embedded clause. From a

PAGE 39

39 minimalist perspective, (9) represents all of the successive cyclic moves that the filler who makes in the given sentence as it moves to the front of the se ntence (Felser 2004): (9) Who did you say (that) Mary likes? [CP2 WHO [C did] you [vP2 who [v say] [CP1 who [C(that)] Mary [vP1 who [v likes] who ]]]] [uWH ,Q] [EPP, uQ ] [uWH,Q] [uWH,Q] [uWH,Q] [WH,Q] The sentence and its structure differ from the examples previously discussed in that in (9) there is a greater distance, both in a linear and in a syntac tic sense, between the filler and its gap location. The use of CP and vP along w ith the features in brackets2 are not of great concern to the reader, the important aspect to glean is that these ar e structural boundaries that represent potential intermediate resting places for the filler as it move s to the front/top of the structure. Research on L1 speakers indicates that these resting stops are utilized to refresh the filler in working memory so that when the gap location is reached the fi ller has not been forgotten (Marinis et al. 2005, Nakano et al. 2002). Marinis et al. (2005) f ound that L2 speakers of different language backgrounds do not use these resting stops durin g filler-gap resolution. Using a self-paced reading task, the results revealed that L1 speaker s used intermediate gaps, while L2 speakers did not. All speakers showed an increase in RT at the subcategorizing ve rb in the gap condition relative to a non-gap control (fill er-gap integration effect), but L2 readers did not show intermediate gap effects. It is important to note that all of the participants were at least at an upper intermediate level of proficiency and they were able to comprehend these complex structures according to an offline measure. Thus, the researchers conc lude that there is no transfer of syntactic operations and claim that al l L2 processing is lexically-based, regardless of 2 CP=Complement Phrase, vP=light verb Phrase, EEP=Extended Projection Principle, uWH=Uninterpretable WHfeature, uQ=Uninterpretable Question-Feature; What drives successive cyclic movement is under debate, but one view is that it is driven by uninterpretable features, re presented in the example directly above the arrows (e.g., [uWH]).

PAGE 40

40 language background or proficiency. A crucial point to their argumen t is that L2 speakers do not adhere to the TRH; instead they may be inte grating fillers by linking them to their lexical subcategorizers, as proposed by the direct association hypothesis. Results from a more recent study using the cr oss-modal priming technique are argued to support these conclusions (Felser and Roberts 2007). In a cross-modal priming task, visual stimuli (pictures or words) are presented at critical points during the auditory presentation of the sentence. Participants make pr edefined decisions about the visu al stimuli, while simultaneously listening to the auditory. If th e visual stimulus is primed, then reaction time for making the decision about the visual stimuli is faster. For example, (10) Fred chased the squirrel to which the nice monkey explained the games difficult rules [ ] in the class last Wednesday. (graphic a dded; item from Felser and Roberts, 2007) The structural indirect object gap in (10) is located directly after th e direct object, and it is at this point that the image of the squirr el is presented. Cross-modal prim ing taps more directly into the construction of a filler-gap dependency because if the filler is active at a structural gap then a lexical decision about an associated stimulus will be faster because it is primed. The study also investigated whether or not working memory differ ences interact with an individuals ability to reactivate the filler at the gap site, as previous research has suggested is the case with L1 speakers (Roberts et al. 2007). Th e results indicate that L2 sp eakers, regardless of working memory, do not show localized priming at struct ural gaps. Unlike their L1 counterparts, who were only primed at the structural gap, the L2 sp eakers, all of whom scored at an advanced level on a proficiency measure, were primed at a non-gap control position (after games in item (10)). The authors argue that slowed processing is not a reasonable e xplanation, and instead suggest that L2 speakers cannot apply some of the pa rsing routines that are used in L1 on-line processing.

PAGE 41

41 Interestingly, an earlier L2 cross-modal priming study also found that L2 speakers do not operate in the same immediate manner when it co mes to reactivating and integrating fillers (Love et al. 2003). In this study, object relative sentences with lexically ambiguous words as the fillers (pens two meanings: pencil/jail) were used to test the priming of the two meanings at given positions in the sentence, one of which was the structural position in which the filler had been extracted; as in: (11) The professor insisted that the exam be completed in ink, so Jimmy used the new pen 1 that his mother-in-law recently2 purchased 3___ because the multiple colors allowed for more creativity. g ap In sum, Love et al. (2003) found that L1 speakers of English activated both meanings of the filler at the filler itself (position1), as both meanings were primed, but only the correct meaning at the gap (position3). For the late L2 spea kers priming was staggered: the primary meaning was primed late (position 2), followed by the secondary meaning at the gap (position 3). Love et al. (2003) suggest that for L2 speaker s, certain syntactic processes, such as gap-filling, are less immediate than they are for L1 speakers, and th at native syntactic reflexes are never fully acquired by highly profic ient L2 learners. Non-Native Model of Sentence Processing Felser and R oberts (2007) interpret their findings within th e recently proposed heuristic model of L2 sentence processing or shallow structures hypothesis (SSH). The SSH claims that L2 speakers build lexically-based structures during on-line processing, st ructures that do not contain abstract syntactic elements (Clahsen and Felser 2006a, Clahsen and Felser 2006b, Ferreira et al. 2002). More explicitly, this m odel involves the applica tion of lexically-based comprehension heuristics and high plausibility and/or strong canonical meaning-form patterns (pg. 29, Felser and Roberts, 2007). According to the SSH, L2 speakers process filler-gap dependencies on a semantic level by semantically associating fillers with verbs, and the L2

PAGE 42

42 parser does not build structur al gaps during on-line comprehe nsion. Attempting to fit the Williams et al. (2001) stop-making sense data into this model, one could argue, as the researchers themselves point out, that the highly plausible fillers were driving the filled-gap effects (more SMS decisions at the filled-gap), but that in the more natura listic task, the non-use of plausibility information could have been the result of the use of a strong form pattern, which leads L2 speakers to integrat e the filler at the first verb. While this model of L2 processing is attractiv e, it is important that it be fully evaluated through extensive empirical research. One criticis m is that it is somewhat vague and that clear predictions are difficult to make. Another concern over an L2 heuris tic parser is that it applies to all L2 speakers, regardless of L1 background, prof iciency level, working memory or aptitude differences; thus, an intermed iate and an advanced L2 sp eaker may differ in their L2 competence, but they both adhere to an L2 pars er based on heuristics. The evidence used to support the model is based on studies that looked for and did not find transfer effects at a single proficiency level. Many of the commentaries fo llowing Clahsen and Felser (2006a) point out the need for further research before settling on this model (Birdsong 2006, Dowens and Carreiras 2006, Frenck-Mestre 2006, Sabourin 2006, Sekeri na and Brooks 2006, Steinhauer 2006). Summary This chapter presented an overview of sent ence process ing literatur e from both a native and non-native perspective. Three models of L1 sentence processing were discussed: a modular model, a lexically-based model, and a heuristi c model. Research regarding L1 filler-gap resolution was also discussed. While the results from this research do not differentiate clearly between the L1 models, they suggest that for L1 speakers of English, gaps are actively posited by the parser, and that verb bias and verb subcategor ization frame are used rapi dly in this process. The research to date on L2 processing of filler-gap dependencies reveals both qualitative and

PAGE 43

43 quantitative differences between L1 and L2 speakers. It is unclea r from this research whether verbal properties influence filler-gap resoluti on. Williams results suggest that information concerning the plausibility of the filler may be available and used on-line by L2 speakers, but only if they focus on it. Resear ch investigating the use of synt actic information by L2 speakers suggests that full representations may not be built during on-line processing. A recently proposed L2 heuristic processor is a step in the direction of de veloping a model of L2 sentence processing; yet, it is still in a nascent stage. Further studie s should be conducted using different methodologies and looking at furthe r mitigating factors. It is s till unclear as to what effects mitigating factors (L1 transfer, proficiency, work ing memory, and other individual differences) might have on the parser, or whether results using different psycholinguistic techniques will converge. Dual tasks, such as the SMS task an d cross-modal priming, lead to increased reading times, which may influence the strategies both na tive and non-native speakers use on-line. This disadvantage, along with some limitations of self-paced reading (see chapter four methodology), suggest that as research techni ques, they may not be the best techniques to investigate L2 processing, which is already slowed compared to L1 processing. For these reasons, other more sensitive techniques, such as eyetracking, may be more appropriate. Such advances would provide insight into the underlyi ng nature of late second langua ge learners cognitive processing and whether there are ultimate limits on our la nguage processing abiliti es after puberty.

PAGE 44

44 CHAPTER 3 THEORETICAL BACKGROUND Cognitive Theory of Second Language Acquisition When investigating how late-learners process a second language, it is im portant to consider which theoretical perspective of L2 acquisition provides insight into the sources of difficulty that learners face. Processing as it pertains to th e structure under examination in this work, namely filler-gap dependencies, is assumed to be proces sing on a cognitive, psycholinguistic level. This work does not attempt to draw conclusions rega rding the underlying nature of L2 syntactic representations, that is, whether a specific version of UG theory is at play, or whether these learners have full or limited access to UG. Inst ead, it is grounded in a theory of second language acquisition that takes into account real time demands facing learners; thus, the theoretical perspective is of an information processing na ture. There are many models of information processing that have been proposed to best acco unt for the difficulties learners face during second language acquisition and/or learning : Krashens Monitor Theory, MacWhinneys Competition Model, Robinsons Cognition Hypothesis, and Ullmans Declarative/Procedural Model, to name a few. The work at hand is not informed by a specific theory of second language information processing. Instead, a general cogni tive perspective of language learning based on a limited capacity provides the theoreti cal framework for this work. In my opinion, a cognitive theory of second lang uage learning, such as that put forth by Skehan, best accounts for processing factors that impact acquisition of syntactic and morphological features of language in late-learner s (1998). Furthermore, while some researchers argue that the robustness of a cr itical period is lacking, another line of research focuses instead on the hypothesis that adults do no t acquire L2 in the same manner as L1, and that while nativelike proficiency may be attainable, it is not achi eved by the majority of L2 learners (Birdsong

PAGE 45

45 1999, Clahsen and Felser 2006, DeKeyser 2000, Ioup et al. 1994, Skehan 1998, White and Genesee 1996). The above assumption is what motivates much of the current psycholinguistic research in second language ac quisition (Hahne 2001, Hahne and Friederici 2001, Kroll and De Groot 2005, Weber-Fox and Neville 1999). Bley-V romans fundamental difference hypothesis (FDH), that adults rely on L1 knowledge and dom ain-general problem solving to acquire an L2, captures the perspective described above (1990) Furthermore, when learning begins in adulthood, empirical evidence suggests there may be a greater reliance on lexical and pragmatic aspects of communication than on syntactic a nd morphological aspects (Clahsen and Felser 2006, Hahne and Friederici 2001, Marinis et al. 2 005, Skehan 1998). A cognitive model, such as Skehans, captures these assumptions, and explains how L2 learning can proceed in terms of syntactic and morphological aspects of language. Moreover, the strength of the theory relies in its emphasis on processing, particularly it its focus on the role of cognitive processes such as memory and attention. Below I will outline the theory put forth in A Cognitive Approach to Language Learning (Skehan 1998). Skehans approach to L2 learning is based on the premises encompassed in the FDH. The important differences between adul t and child learners for Skehan are that: adults draw on fully developed schematic and contextual knowledge; adults are able to consider their interlocutors point of view; and adults use strategies to byp ass syntax and instead focus on using language to convey meanings. This perspectiv e adopts the view that language is more lexically-based for adult learners, and that this applies more so dur ing processing due to th e heavy working memory (WM) demands of on-line computation. For thes e reasons, Skehan describes his model as a limited-capacity information-processing system, in which memory and lexical learning play major roles.

PAGE 46

46 The lexicon is not responsible for all of acquisition in this model though. Skehan advocates a dual-processing system, a memorized exemplar system and a rule-based system, influenced by the work of Sinclair (1991). According to Skehan, Si nclair suggested that language learning and use is guide d by two principles, a lexical principle and an open-choice principle. The lexical principle treats uttera nces as idioms, where expressions become whole units. The open-choice principle refers to the ca pacity to combine lexical items in a creative manner. In this system, the idiom principle (Skehans exemplar system) overrides the openchoice principle (Skehans rule-based system). For L2 adult learners, lexicalized exemplars minimize processing demands, and provide learners mo re time to either plan speech or aid in comprehension. One important aspect of Skehans model is that an L2 cannot be acquired solely on the basis of exemplars. If it did, the system would not be able to adequately account for how syntactic or morphological aspects of the target language are lear ned. Skehan claims learners must balance their processing: half on exempl ars and half on rule-based learning. If adult learners fail to balance their learning, they will never achieve a high level of proficiency. The question, then, is how is this accomplished? Skehan adopts a processing perspect ive of L2 learning that gives noticing a central role in processing (Fig 3-1). This view is not unique to Skehans model and is largely based on the work of Schmidt (1990, 1992). In this model, not all noticing is equally valuable, yet input must be noticed in order for the learning system to be able to use it. Six factors contributing to the systems ability to notice are: frequency of the input; saliency of the input; how adequate instruction is at getting the lear ner to notice the inpu t; individual differences in aptitude and dealing with processing demands; the readiness fo r the system to notice the input; and whether the demands of the task allow for noticing to ta ke place. Noticing in this model can be thought

PAGE 47

47 of as attention (implicit or explic it), awareness (consciousness), or control (fluency). In Schmidt (1990), it is argued that awareness is necessary in L2 learning to acquire grammatical features, but that incidental learning thr ough attention can also be affec tive if the learning environment adequately draws attention. Skehan points out th at awareness enables le arners to match their current language system with the input they receive; that it helps them appreciate the instruction they get, and the corrections they get; that it helps them generate the material better; and that it helps them maneuver between the exemplar system and the rule-based system. For example, the rule-based system can generate an utterance that becomes an exemplar and can be used in the future to relieve the burden on WM. Figure 3-1. Central Role of Noticing. (Skehan 1998, pg 57) Working memory is the space where, if the factors converge and allow the input to be noticed, what is noticed becomes available for rehearsal, modification and incorporation into

PAGE 48

48 long term memory. The central pr ocessing that takes place contro ls what happens to elements once they are noticed. The processes that can take place in working memory are complex, and include matching, feedback, and rule-based exem plar generation. The processes in long-term memory (LTM) are: the exemplar-based memory system and the rule-based system, and schematic knowledge. The connections between WM and LTM are constrained by the speed and efficiency of the WM system. Skehan again poi nts to Schmidt to explain how this works: awareness (one type of noticing) enables the learner to lean to ward a rule-based perspective, which will more likely lead to long-ter m change in the interlanguage system3. The key to the theory can be summed up in the following argu ment: the greater the emphasis is on awareness of gaps, on awareness of what feedback can te ll one, on awareness of recombination, and on awareness rule-based generation, the greater the chance that long-term memory will move from accessing exemplars to accessing rule-based represen tations. In turn, this means the greater the likelihood of restructuring the unde rlying interlanguage system. Ou tput (practice) also plays a role in this system because it enables d eclarative knowledge (exemplars) to become proceduralized (automatic) to relieve burdens on WM. The faster this process can take place for the learner, the better. The model places a larg e role on individual differences (IDs) because IDs contribute to the ability of an individual to notice. Skehans theoretical model is also compatib le with a neuropsychological model put forth by Ullman, the declarative/procedural model (D/P model) (2001). The D/P model of L2 learning proposes that late-learning L2 speakers rely on their declarative (e ncyclopedic-episodic knowledge) memory system to a greater extent than their procedural (how-to knowledge) memory system. For L1 speakers, grammatical ru les fall within the procedural memory system. 3 In L2, the interlanguage system refers to the language sy stem at any point throughout learning. It is unlike the L1 system of the speaker and unlike a native sp eaker of the target languages system.

PAGE 49

49 Because this system is not as accessible for late-learners, declarativ e memory handles rule generation, but it is less apt at doi ng so. Within this model, pr actice initiates a shift from the reliance of grammatical processing on declarativ e memory to a greater reliance on procedural memory. As a last note, I think it is important to point out that Skehans mode l does not dictate what the representational system looks like. His information processing approach is more concerned with control of attention duri ng comprehension and production th an what the actual lexical or rule representation is; this means that the mo rphology and syntax may be represented in a UG format, or it may be represente d in a connectionist format. From the perspective of someone who is intere sted in L2 processing, this model of second language learning has several advant ages. First of all, this model is process and capacity-based, which means that there is less focus on the representation of the system and more on whether an individual has the capacity to manage the system in various ways. The model also incorporates basic principles of the FDH, which helps expl ain why adults face greater difficulty in both learning and using a second language. Because the model is capacity-based, there is an emphasis on individual differences, in that aptitude and WM, among other differences, can play a role in the process of learning, as well as in on-line processi ng. The model also easily adapts to neuropsychological models, such as Ullmans D/P model of L2 acquisition. Together the predictions of the theory account for why adult learners have trouble with automatic processing (both on the comprehension end and the production end) and w ith syntactic and morphological elements. In a system that focuses on meani ng and requires noticing to acquire rules, adult learners are faced with an uphill battle in get ting to the point where they produce and process syntactic and morphological aspects of the language with a degree of automaticity near what

PAGE 50

50 native speakers do. I think this model best accounts for the acquisition of morphological and syntactic aspects of a second langua ge because it inherently assu mes that these aspects will be more difficult. While Skehans approach consid ers age of acquisition, it can account for factors such as L1 transfer and proficiency. These fact ors are not just important to consider in the overall acquisition process, but also in how they manifest themselves in L2 on-line processing. Native Language Transfer Roughly, language transfer can be defined as the use of prior linguistic information, whether it be of a syntactic or semantic nature in a non-L1 context (Ga ss 1996). Transfer can be positive or negative in the sense that it can help a l earner or hurt a learner. In an overview of the role of transfer, Gass points out how learner perception of L1 and L2 may play a role in whether something is being transfer to the target langu age or not (1996). Her example is of Czech speakers transferring morphology to Russian but not to English because they perceive it as relevant to Russian but not to English. This implies a certain awareness on the part of the language user, which fits within the cognitive ap proach described above; that is, transfer may rest on the L2 learners ability to perceive of their L1 knowledge as being useful ( this could be perceived implicitly). Negative transfer might work in the opposite way. If learners do not implicitly or explicitly notice whether their L1 is useful or not, the result could have negative implications. Within the L2 pro cessing literature, whether or not L2 learners transfer either L1 knowledge or L1 processing strategies to L2 contexts is under debate. One study using event-related potentials (ERP s) that finds evidence for L1 transfer effects in L2 processing is Sabourin and Haverkort (2003). The motivation of the research was to investigate whether advanced L2 speakers of Dutch with an L1 of German display the same off-line capabilities regarding gender as native sp eakers, and if so, whether the L2 speakers use this knowledge during on-line processing. The Dutch gender system is similar to German in its

PAGE 51

51 handling of definite determiner noun agreement, but differs from German in its handling of indefinite determiner noun agreement. The results from an off-line grammaticality judgment task indicated that L2 speakers have learned de finite gender agreement in Dutch, but struggle when it comes to indefinite agreement. The on-line task, which looked at how Germans would process gender violations in Dutch compared to native speakers, found similar results. The results of the ERP experiment show that German speakers are sensitive to violations of gender agreement based on the parts of the Dutch gender sy stem that are similar to their L1, but are not sensitive to violations based on pa rts that differ from their L1. Th e results are interpreted from a declarative/procedural perspectiv e, suggesting that L1 procedur al knowledge (syntactic) may be used (transferred) by advanced learners of an L2. More recently, Tokowicz and MacW hinney (2005) investigated transfer in the early stages of late-learning. ERPs were used to examine if L1 syntactic cues are used implicitly by early learners. Processing of violations of a similar cue (auxiliary verb), a different cue (determiner number agreement), and a unique cue (determi ner gender agreement) in Spanish by English speakers was examined. The assumptions of the study are based on the competition model, which argues that cue weights in L1 will influe nce cue weights in L2 and play a role in processing. The researchers hypothe sized that ERPs would reveal implicit sensitivity to the L1L2 similar cue, no implicit sensitivity to the di fferent cue due to competition, and sensitivity to the unique cue due to lack of competition. While the results are marginal, the predictions were borne out. They argue that L1 cue knowledge tr ansfers (and competes) with cues being acquired in the L2. While the two studies above argue for transfer there are also two recent studies examining L2 filler-gap resolution that claim there is no L1 transfer during L2 processing (Marinis et al.

PAGE 52

52 2005, Williams et al. 2001). Both studies looked at filler-gap resolution comparing groups of L2 speakers from different L1 backgrounds. The hypot hesis was that if transfer was occurring, the L2 groups from different L1 backgrounds shoul d display qualitatively different patterns of processing; that is, for groups with an L1 wh in-situ background compared to groups with an L1 wh ex-situ background differences in patterns should emerge. Williams et al. (2001) used a stopmaking-sense task along with self paced reading to look at whether an ac tive-filler strategy was used by all groups of L2 speakers (Chinese, Kore an, and German). They found that all groups used this type of strategy. While main eff ects of language were found when the native group was compared to the L2 groups, these effects disappeared when the native group was removed from the analysis. Such a finding suggests lack of transfer. Marinis et al. (2005) looked at the use of intermediate gap positions in the processing of long distance whdependencies using selfpaced reading. The results across groups (Germa n, Greek, Japanese, and Chinese) indicate that none of the L2 groups used intermediate gap positions; thus, no L1 transfer. There are several other studies using self-paced reading and ey e-tracking to investigate L2 processing and the results have been mixed (Frenck-Mestre and Pynte 1997, Frenck-Mestre 2002, Juffs and Harrington 1995, 1996, Juffs 1998, 2005) It has recently been argued that measures of self-paced reading may be obscuring subtle differences between L2 groups (FrenckMestre 2005). In Frenck-Mestre & Pynte (1997), eye-tracking demonstrated transfer of verb bias in L2 French-English learners. The other st udies that found patterns of transfer have used a more sensitive technique (ERPs) for measur ing on-line processing (Sabourin 2003, Sabourin and Haverkort 2003, Steinhauer et al. 2006, Tokow icz and MacWhinney 2005). While more research is needed to uncover whether L1 transfer occurs, much of the L2 ERP research indicates that transfer at an automatic early level is lacking in late -learners (Hahne 2001, Hahne and

PAGE 53

53 Friederici 2001, Mueller 2005). Wh at is important for the study at hand is that L1 background be kept constant, as its role in filler-gap resolutio n is unclear. Non-Native Proficiency Proficiency is a factor that has obvious ram ifications in any study involving second language. It is also a factor that is more difficult to determ ine than L1 background. One of Bley-Vromans motivations for suggesting the F DH was that success rates for late-learners are variable (1990). This means th at while two learners may have been learning the target language for the same amount of time and under the same conditions, their proficiency levels could differ dramatically. In the early studi es on the critical period hypothes is, overlooking proficiency often led researchers to make bold statements about wh at was attainable or what was not (Johnson and Newport 1989, White and Genesee 1996). In L2 sentence processing literature, much of the literature seeks to define their populations at a homogenous proficiency leve l to avoid influences from this factor. Few studies have investigated possible influences. Among much of the behavioral research, very little has been done to directly compare proficiency groups. Frenck-Mestre (2002), however, does report on one study looking at processing of ambiguous sentences, which found differences based on proficiency. The study compared the processing of re lative clause attachment (e.g., Arnold watched the wife of the doctor who was leaving the medical centre ) in two proficiency groups of late-English learners of French to French native speakers. The lowe r English-French group preferred low attachment, whereas the higher proficiency English-French gr oup showed a preference for high attachment. Weber-Fox and Neville (1996) was an early L2 study looking at syntactic processing differences based on age of acquis ition effects. Proficiency measur es were given to participants, but unfortunately they were not kept constant within age groups ; therefore, the results suggest how age contributes to syntactic processing differences, but not proficiency. Other ERP studies

PAGE 54

54 examining differences between native speakers a nd late-learning non-native speakers have also found differences, but have only compared one proficiency group to native speakers (Hahne 2001, Hahne and Friederici 2001). As stated earlie r, these studies have consistently found that late-advanced L2 learners may not be using earl y automatic syntactic processes. A recent ERP study, however, did find that automatic processi ng may vary as a function of proficiency (Steinhauer et al. 2006). This study tested four groups of late-learning L2 speakers: high and low proficiency French-English speakers and high and low proficiency Chinese-English. The materials tested contained phrase structure viol ations, and results suggest that not only is proficiency playing a role in the use of these early automatic syntac tic processes, but also that L1 background may interact with proficiency. The results from the studies above clearly de monstrate that proficiency is an important factor in L2 sentence processing ; furthermore, they demonstrate how the two factors, L1 transfer and proficiency, may interact. What this means for L2 processing studies is that these issues cannot be overlooked when designing experiments or when building theoretical models of L2 sentence processing. Factors in Relation to Current Experiments Som e on-line studies report inte ractions between transfer an d proficiency, while some do not. More research needs to be conducted th at seeks to compare groups of different L1 backgrounds and different proficiency levels. If interactions occur, they should manifest themselves in qualitatively different patterns in th e on-line record, as seen in Steinhauer et al. (2006). As mentioned in Chapter two, a recent heur istic model of L2 processing is proposed (Clahsen and Felser 2006). By heuristic, the au thors mean that L2 learners compute shallow predicate-argument structures based on lexical, pragmatic, and world knowledge without the use

PAGE 55

55 of detailed syntactic and morphological structure. Such a perspective would also be predicted for some L2 learners based on the cognitive model of L2 acquisition outlined above.4 While it is hopeful to see models of L2 sentence processing being put forth, the main concern over a single L2 shallow structures parser is that it applies to all L2 learne rs, regardless of L1 background and proficiency level. Many of the commentaries on the Clahsen and Felser article point out the need for further research before settling on this model (Birdsong 2006, Dowens and Carreiras 2006, Frenck-Mestre 2006, Sabourin 2006, Sekerina and Brooks 2006, Steinhauer 2006). The experiments in the current study do not directly test th e SSH, but instead aim to answer questions about the developmental nature of the L2 parser. Processing demands are predicted to play a central role in what in formation is available during on-line filler-gap resolution. Proficiency level is included as a variable, as it is a ssumed, based on a cognitive approach to second language acquisition, that lower proficiency speakers will be overburdened during on-line processing. These speakers are, thus, not expected to be using verbal information to resolve filler-gap depende ncies as they do not have e nough resources free to notice the significance these properties contribute to resolvi ng these complex structures. In order to avoid the influence of L1 transfer, L1 background is kept constant, and all non-na tive participants share an L1 background of Korean, a wh in-situ language (Dis cussed more in Chap ter four). 4 Some because while the cognitive model places meaning over form, it also takes into consideration individual differences, which would allow for differences in processing based on the capacity of the individual to notice.

PAGE 56

56 CHAPTER 4 INFLUENCE OF SUBCATEGORIZATION FR AME DURING FILLE R-GAP RESOLUTION Introduction Based on L2 sentenc e processing research of filler-gap dependencies, reviewed in chapter two, it appears that advanced late-learning L2 speakers may be sensitive to plausibility information, but more importantly that they use a filler-driven st rategy when processing wh dependencies. Many questions, however, remain regarding how specifi c verbal properties influence L2 filler-gap resolution, as well as if there are differences in strategy based on proficiency level. In this ch apter, two experiments are presen ted which investigate the use of subcategorization frame by late L2 learners of different proficiency levels, advanced and intermediate, while resolving filler-gap dependencies in English. The two experiments compare results from different psycholi nguistic methodologies, self-paced reading and eye-tracking. As stated above, the goal of both experiments is to investigate the role subcategorization frame plays in L2 filler-gap resolu tion. The syntactic structures that are used to investigate this issue are similar to those used by Goodluck and colleagues. The structures involve wh extraction from PP Complements compared to PP Adjuncts, which follow put/push verbs respectively and, thus, differ in argument requi rements (Goodluck et al. 1991, Goodluck et al. 1995). A put verb obligatorily subcategorizes for an NP and a PP (ungrammatical: he put the cart ), while a push verb subcategorizes for only an NP (grammatical: he pushed the cart ). As discussed in chapter two, native speakers have been shown to be influenced by this information during on-line filler-gap resolution. In the context of the current experiments, this means that knowledge of the PP requirement of put -type verbs aids native speak ers in predicting upcoming gaps locations. The question guidi ng the two experiments is whethe r intermediate and advanced

PAGE 57

57 speakers from an L1 background without obligatory wh -movement are equally influenced by this property. Participants across Experiments In both experim ents, L2 participants are late learners who share th e same L1 background, Korean. Late-learning is define d as having learned English after the age of twelve. Twelve was chosen as the cutoff age of acquisition due to maturational effects found in syntactic processing for the age group 11-13 (Weber-Fox and Neville 1999). Korean was chosen as a common L1 background because it is a wh in-situ language, in that wh -questions are formed by replacing the questioned NP with a question word without moving the question word to a non-canonical position in the sentence. (1) Mary.nun caebaeng.eseo muet.ul sat.ni? Mary.NOM bookst ore.at what.ACC bought/past.ques What did Mary buy at the bookstore? Unlike the English translation, in Korean the question-word muet.ul remains in the same canonical position as it would in a declarative sentence. Recent research on Japanese, another wh in-situ language, demonstrates that scrambling of question-phrases leads to strategies for resolving filler-gap dependencies; and that these strategies are ba sed on filling the gap at the first verb the reader encounters, rather than at the first possible gap location (Aoshima et al. 2004). In addition, research on Korean sentence processi ng suggests that reader s use morphological and prosodic information to process scrambled construc tions before verbs are encountered (Kiaer and Kempson 2005). Given that speakers of wh in-situ languages have strategies for resolving dependencies, it is not correct to argue that L1 transfer of filler-gap resolution strategies is not possible; because of this, it could be argued that it is no more interesting to look at learners with an L1 Korean background over learners with an L1 Spanish background. There are, however, a

PAGE 58

58 few reasons why it is still interesting to examin e how L1 speakers of Korean process filler-gap dependencies in English. One reason is that th e research on scrambling implies that L2 learners of English from languages that allow scrambling may be able to transfer strategies from the L1 to the L2, and that given Englishs word order differences with head-final languages, such as Korean, this may mean that Korean L1 participants use the strategy of filling the gap at the first verb. It is also of interest to see if these readers use lexical information during this process. Another consideration is that sc rambled constructions have been reported as being infrequent, at least in Japanese (Yamashita 1997), meaning that speakers from languages which scramble question-phrases, over ones that require movement, may be less likely to transfer such strategies, particularly in the face of increased processing demands associated with processing an L2. In any case, since L1 background could have possible tr ansfer effects, it is important to keep the group of L1 learners as homogenous as possible. For this reason, it wa s better to have all participants come from the same L1 background regardless of whethe r they have L1 strategies to transfer or not. It is also im portant to point out that the emph asis throughout the experiments is on whether strategy differences between groups emerge. Research Questions and General Predictions The research questions which gui de these two experim ents are: Do L2 speakers use subcategorization frames to posit potential gap locations; that is, will they be influenced by the subcategorization type of the verb to inform gap sites? Do strategies change as proficiency increases? Will data from self-paced readi ng and eye-tracking converge? The general prediction is that advanced speakers will be more sensitive to lexical information during on-line filler-gap resolution due to lowe r processing demands associated with greater automatization in L2 reading; this means that they, like native speakers, will be influenced by lexical information associated with the verb to aid in positing eligible gap locations.

PAGE 59

59 Intermediate speakers, on the other hand, are hypothesized to be less sensitive to lexical information due to higher processing demands a ssociated with lesser automatization in L2 reading, on top of the demands associated with resolving non-local dependencies; this implies that unlike native speakers and advanced speakers, they will ignore lexical information during gap positing. Regarding whether results from both methods will yield the same results is unclear. Results are being compared based on a dvantages, disadvantages, and recent arguments concerning both methods. Below, these issues are outlined. Why Compare Self-Paced Reading to Eye-Tracking? Several psycholinguistic m et hods, which have been used to examine native language processing and representation, are also useful for investigating second language (L2) processing and representation. Two on-line methods, which ar e increasingly becoming of interest to second language researchers, are self-paced reading and eye-tracking. The most used method of the two to date for investigating L2 on-line sentence processing has been self-paced reading (Clahsen and Felser 2006, Juffs 2001a). Self-paced reading is a technique used to measure how long readers spend reading a critical word or region within a sentence (See chapter two for discussi on). There are several advantages and disadvantages associated with th is method. One major advantage is that as online methods go, self-paced reading is the most simple and least expensive (Mitchell 2004). The ease of use and low cost may offset the disa dvantage of imprecisi on, rendering this method especially fruitful for preliminary investiga tions; such as when investigating linguistic phenomena of which relatively little is known, or when undertaking an investigation of a different population; as in the case of L2 learners. A disadvantage to this technique is that it is an unnatural way of reading. Even though Just, Carpenter, and Wooley (1982) compared selfpaced reading to eye-tracking techniques and f ound that it produced many of the characteristics

PAGE 60

60 of naturally occurring eye-fixation data, ther e are some significant differences. In normal reading, function words are often skipped enti rely (Rayner and Sereno 1994), whereas they cannot be skipped in self-paced reading. Furtherm ore, as Garnsey et al. (1997) point out, readers may slow down more than normal with self-p aced reading, which argues that as a research technique it is not sensitive to initial interpreta tion, but rather includes revision. In self-paced reading, readers are also unable to regress to an earlier wor d, which may place a burden on working memory. Reading in an L2 may already place burdens on working memory, thus, selfpaced reading may put L2 readers at a further disadvantage (Ardila 2003). In addition, if a researcher is interested in investigating issues related to regression, se lf-paced reading would not be the appropriate measure to use. While self-paced reading has been used to a mo derate degree in L2 research, eye-tracking is a method that has been underused (Frenc k-Mestre 2005a, Frenck-Mestre 2005b, Juffs 2001b). Unlike self-paced reading, where a reader contro ls the pace of word presentation, eye-tracking records eye fixations as a reader moves their eyes across a complete text. The result of recording various fixation measures is a richer data set compared to self-paced reading (Reichle et al. 2003). Eye-tracking has been argued to be better capable of dete rmining more precise timing to the onset of difficulties in on-line proce ssing (Boland 2004, Frenck-Mestre 2002). Eye movements are typically acquired with head -mounted video cameras that measure pupil movement relative to text or images presente d on a computer monitor. Some eye-tracking equipment requires a bite bar to prevent read ers from moving their h eads while reading, but newer technology can compensate for head move ments, making the bite bar obsolete (Mitchell 2004). The typical measures analyzed by resear chers are: first fixation duration, total gaze duration, first pass reading time, second pass read ing time, total reading time, regression path,

PAGE 61

61 and probability of regression (see table 41) (Boland 2004, Rayner and Sereno 1994, Reichle et al. 2003). Table 4-1. Eye-Tracking Measures Measure Definition First Fixation Duration First tim e eyes fixate on a word. First Pass Reading Time The sum of a ll fixations within a region before a progressive saccade is made to another region. This measure may be equivalent to the above if only one fixation is made. Total Gaze Duration The sum of gaze duration and any regressions made back to the word. Second Pass Reading Time Time sp ent re-fixating within a region. Regression Path Sum of all regression fixations from a region and before a saccade is made to a progressive region. Probability of Regression Th e percentage of first pass regressions out of a region. Even though eye-tracking provi des the researcher with mo re information about on-line processing, there is still not an isomorphic relationship between eye movement and cognitive processes (Rayner and Sereno 1994). Researchers take different positions on these linking assumptions. As mentioned in Rayner and Sere no (1994), Just and Carpenter (1992) put forth the immediacy hypothesis, which states that wh en the eyes move from a word, processing for that word is complete. The problem with this linking assumption is that it doesnt account for spillover effects and regressions (Rayner and Se reno 1994). Spillover effects are when fixation times on another word following a word that is more difficult to access or integrate are longer compared to a control. Spillover effects and regression suggest that all processing is not completed before the eye moves away from an ar ea of difficulty. This le ads to the question of why the eye would move at all if processing was not complete. Re search shows that there is a disparity between when a movement is programme d and when it is initiat ed (Rayner and Sereno 1994, Rayner et al. 2003, Reichle et al. 2003). This phenomenon helps to e xplain the skipping of words, short fixations and non-optimal landing sites. Rayner and Sereno (1984) also mention

PAGE 62

62 that fixation times are not just affected by lexi cal access, but may also be influenced by higher order processes (such as conceptual in tegration and syntac tic processes). Boland (2004) tackles the issu e of linking assumptions as it relates to higher order processes in sentence comprehension. Word fr equency, length, predicta bility, and ease of integration into the sentence all have been shown to influence lexical access and integration into the readers representati on, reflected in longer fixations. The problem, as Boland sees it, lies in that it is unknown how uni que processes appear in eye movement data. Boland (2004) reports on severa l studies that examined careful manipulations to tease this linking assumption problem apart. Frazier and Rayner (1982) argue for a linking assumption where first fixation duration reflects that a problem was detected by the reader, whereas regression reflects that the source of the problem was identified. This is not entirely explicit, in that regressions may reflect an in correct initial analysis, or they may reflect the plausibility of competing analyses, or they may reflect syntactic differences between the initial and the revised analyses. Boland (2004) puts forth a new linking assumption: namely, that eyes do not leave a word until it has been structura lly integrated; thus, constraints that influence structure building influence first pass reading time (which incl udes first fixations and regressions). She frames her argument in terms of the EZ reader model of eye-movement control, which does not consider the issue of hi gher cognitive processes (Reichle et al. 2003). In the EZ reader model visual attention moves to the next word when lexical access is complete. There are two stages during lexical access: the first stage co ncerns identifying the orthography of the word, not complete lexical access, but enough to signal the planning of the ne xt saccade; the second stage involves identifying phonological and semantic information about the word necessary for integration, which may not be complete before th e saccade begins (structura l integration need not

PAGE 63

63 be complete either). The EZ reader model only strives to explain what drives normal reading processes, not what happens when a difficulty arises. Boland recommends changing the EZ reader model by including syntactic predictability and frequency of syntactic form into the first stage. Even with such an adaptation, questions remain about the link between cognitive processes and eye movements. Th e point that I want to draw he re is that problems with linking assumptions leave researchers to be somewhat conservative in their claims about initial influences on parsing. Frenck-Mestre holds eye-tracking to be a better method compared to self-paced reading for investigating L2 processing (Frenck-Mestre 2005a, Frenck-Mestre 2005b). She claims that selfpaced reading only provides total reading time, and although there is not a direct relationship between which cognitive processes are entailed in initial parsing decisions, valuable information can be gained by separating first pass from sec ond pass reading. She further suggests that selfpaced reading may be potentially misleading when comparing L1 and L2 groups. In Altarriba et al. (1996), an L2 study looking at lexical access, differences we re found in skipping probability and first fixation times between L1 and L2 read ers, measures not available from self-paced reading. Another L2 study that used eye-tracking is Frenck-Mestre and Pynte ( 1997). In this study, effects of verb argument struct ure on initial syntactic processi ng in ambiguity resolution was investigated for L2 readers compared to native readers in French. The dependent measures examined were first pass reading times, assumed by the researchers to be indicative of initial parsing decisions, and second pass reading times. The structure under inve stigation contained an ambiguous PP, which in English and French pr eferably attaches high (to the VP). (2) They accused the ambassador of espionage but nothing came of it. (translation not available)

PAGE 64

64 Verb argument structure was varied with eith er ditransitive verbs, two arguments, or monotransitive verbs, one argument; as well as manipulating the noun within the PP to prefer high or low attachment. Thus, if a reader was being guided by the lexical information of the verb, and they read a verb with two arguments and encounter a PP, they should prefer to attach high (to the VP). First pass reading time should then be shorter on a PP that semantically belongs to the verb compared to the same c ondition with a monotransit ive verb because the readers expectations were met. On the other hand, if the noun in the PP following the ditransitive verb semantically at taches low (to the NP), first pass reading increases should be observed because expectations were not met. The opposite pattern should be observed for monotransitive verbs, in that first pass readi ng times should increase for PPs that semantically attach high, but should not increase for PPs that sema ntically attach low. First pass reading times demonstrated that both L1 and L2 readers were influenced by a verbs lexical properties in resolving ambiguities. Although, Frenck-Mestre a nd Pynte were interpreting first pass reading times as reflecting initial parsing decisions, it is interesting to note that in first fixation times, there were no effects or interactions for native speakers. For L2 readers, there were effects on first fixation times in that they exhibited longer first fixations at the PPs noun following a monotransitive verb with semantic high attachment interpretation. They interpret this to mean that L2 readers are driven more by lexical properties than native speakers. Eye-tracking, in this case, was the more appropriate experimental method to use. Self-paced reading may have obscured the subtle differences between the two groups. Eye-tracking as a method does appear to have advantages for investigating L2 sentence processing. If the differences between L1 and L2 readers are small, it behooves a researcher to use a more sensitive method. The advantage this method offers, however, may also turn out to

PAGE 65

65 be a pitfall. As Reichle et al (2003) point out, Because an averag e college-level reader can read approximately 300 words per minute, this technique produces a staggering amount of data (449). Researchers, thus, are likely to have more difficultly inte rpreting the results, as evident from the discussion of Frenck-Mestre and Pynte (1997). Materials across Experiments Across both experim ents 1 and 2, the same materi als will be used. Three verbs with two obligatory complements (put, place, stick ), and three verbs with one obligatory complement ( push, drop, lift ) were used to construct 36 quadruplets of the type illustrated in table 4-2 (Appendix A). Table 4-2. Subcategorizati on Frame Example Materials Condition* Critical Words A PP C Gap The card dealer saw what the old player /put/the/cards /under/during/the/ poker game. B PP C No Gap The card dealer saw if the old pl ayer /put/the/cards /under/the/napkin/ during the poker game. C PP A Gap The card dealer saw what the old player /pushed/the/cards /under/during/the/ poker game. D PP A No Gap The card dealer saw if the old play er /pushed/the/cards /under/the/napkin/ during the poker game. *PP: prepositional phrase, C: complement, A: Ad junct. Critical words are separated by /. The area of most interest, the filled-gap region, is un derlined. See Appendix A for complete list of items. Items were divided into four lists through a Latin-square procedure such that each list contained 9 different sent ences for each of the four conditions. A total of 72 distracter items were included (Appendix B), yielding a total of 108 sentences per participant. Eighteen distracter items contained legitimate short gaps with transitive bias verbs (e.g., The press asked what the clever spy found near the gun at the crime scene ). The remainder of the distracter items contained complex clauses (e.g., The carpenter asked when the new designer delivered the new materials for the next phase ).

PAGE 66

66 The six verbs were narrowed down from a list of 32 verbs, 16 PP complement verbs (two obligatory complements) and 16 PP Adjunct verbs (one obligatory complement), compiled from a verb argument structure database (Gahl et al. 2004). In order to ensure that the PP complement verbs required a PP and the PP adjunct verbs di d not, two tasks were carried out. A selfcompletion questionnaire (Appendix C) was first administered to 33 participants, all native, monolingual speakers of English. These participants were recruited from the general student body and from an established participant pool shared by the Communi cation Sciences and Disorders department and the Linguistics program at the University of Florida (henceforth called the LIN-CSD Pool). Participants for this ques tionnaire participated on a voluntary basis or, if recruited through the pool, for c ourse credit. The questionnaire consisted of 32 experimental items and 16 filler items. In order to create tw o different lists of items, noun phrases consisting of a determiner, an adjec tive and an animate noun (e.g., the little girl ) were randomly matched with the 48 verbs, creating two lis ts. The two lists were then pseudo-randomized so that two verbs with a predicted identical argument structure appeared in a row. This completion questionnaire (table 4-3) resulted in four verbs from each list. These four verbs were chosen to be included in the grammaticality judgment task based on the percentage of completions that conformed to either the obligatory nature of the PP complement verb s (PP Completion: e.g., the man put the box on the table ) or the optional nature of th e PP for the PP adjunct verbs (NP Completion: e.g., the man pushed the cart ).

PAGE 67

67 Table 4-3. Percenta ge of Completions Verb Verb type NP Completion PP Completion Put PP Comp 0% 100% Place PP Comp 0% 100% Stick PP Comp 0% 85% Strap PP Comp 0% 88% Push PP Adjunct 30% 57% Lift PP Adjunct 46% 46% Drop PP Adjunct 63% 33% Spill PP Adjunct 42% 57% The grammaticality judgment task (Appendix D) consisted of 48 items, thirty two of which were fillers. For each verb in table 4-3, two versions were created, one grammatical and one ungrammatical. For the PP complement verbs, the four ungrammati cal items had direct objects, but no prepositional phrases, while the four gr ammatical items had both. For the PP adjunct verbs, the four ungrammatical items had no objec ts, while the four grammatical items had only direct objects. Two pseudo-randomized lists were created altering the agent in the subject noun phrase to ensure that semantic fit was not infl uencing participants choices. The questionnaire was administered to 20 native, monolingual speaker s of English, who were recruited in the same manner described above. The responses of interest were the grammati cal versions of the PP adjunct verbs and the ungrammatical versions of the PP complement verbs. The PP complement verb strap was eliminated from the list because 40 % of participants rated items such as The lazy clerk strapped the books as grammatical. For the purposes of keeping the number of verbs repeated in the main task equal across verb type, spill was eliminated as a PP adjunct verb; resulting in three verbs for each verb type. A ll other items had opposite responses at a range of 0%-20%, and were thus considered accurate repr esentatives of the resp ective argument structure categories.

PAGE 68

68 Experimental Predictions If both self-paced readin g and eye-tracking are equally sensitive measures, experimental predictions are as follows. Experiment 1: Self-Paced Reading Native and Advanced Speakers. D ue to the use of the active-filler strategy, all gaps will be filled at the verb regardless of verb type. This will lead to filled-gap effects, that is, increases in reading time (RT), at the filled-gap NP or s oon after (Critical Words 2 and 3, table 4-2) in the gapped conditions compared to the non-gap conditions. Subcategorization frame, however, will influence parsing decisions for both groups of sp eakers; that is, there will be an interaction between verb type and gap in the filled gap region (Critical Words 2 and 3) or soon after. What this means is that relative to the non-gap controls the difference in RT at critical words 2 or 3 between conditions A and B (PP Comps) will be smaller than the difference in RT between conditions C and D (PP Adjuncts). The l ogic behind these RT pa tterns is that the subcategorization frame of the put -type verbs will influence pote ntial gap sites; thus, while participants may expect the gap to be at the verb in the PP Comp Gap (A) condition, the subcategorization frame associated with this ve rb type will facilitate positing a gap at the required PP if the NP following the verb is filled. While not central to the main predictions of this experiment, there is also the issue of whether the advanced speakers will recognize the filled-gap at the determiner or at the noun. W illiams et al. (2001) argued that L2 readers do not appear to be sensitive to the syntactic cue represented by the determiner, so differences may emerge between native speakers and advanced speakers regarding whether RT increases at critical words 2 or 3. Intermediate Speakers. As with the native and advanced speakers, intermediate speakers are expected to fill gaps at the verb regardless of verb type due to the use of the active-filler

PAGE 69

69 strategy. This will also lead to filled-gap eff ects at the filled-gap NP or soon after (Critical Words 2 and 3) in the gapped conditions compared to the non-gap conditions. Subcategorization frame, however, will not influence parsing decision s; that is, there will not be an interaction between verb type and gap in the filled gap regi on (Critical Words 2 and 3) or soon after. The reason for these RT patterns is that the lower pr oficiency participants are not expected to be influenced by the subcategorization frame differe nces between the two types of verbs. If the predictions for both L2 groups are borne out, it woul d suggest that the parser develops strategies as proficiency increases. Experiment 2: Eye-Tracking Native and Advanced Speakers. A ssuming that both methods are equally sensitive, the results will be similar to those described above. Intermediate Speakers. As with the native and adva nced speakers, the results are expected to be similar to those described above. Comparison of Experiment 1 and Experiment 2 If both techn iques are equally sensitive, the eye-tracking data will support the results from the self-paced reading data, regardless of whethe r the predictions above are supported. If the predictions above are borne out, it is unclear which eye-tracking measur e will support the selfpaced reading data since self-paced reading has been criticized as being reflective of revision; thus, effects present in the se lf-paced reading data may not show up in first pass reading measures, but instead may be found in total ga ze duration or regressi on-path measures. It is also possible that self-paced reading may not be sensitive enough to pick up on subtle differences between L2 speakers. In this case, th e self-paced reading data for advanced speakers would suggest that advanced speakers are not sens itive to subcategorization frame; while results from the eye-tracking data analyses would yield an interaction between verb type and gap, as

PAGE 70

70 described above, for advanced speakers, which w ould be lacking for intermediate speakers. Results along the lines of the latter would indicat e that eye-tracking is a better technique to use when investigating L2 sentence processing. Procedures across Experiments Pre-Testing Materials A given set of tasks were adm inistered in a systematic order in two sessions for L2 participants and in one se ssion for native speakers. Session one: o Proficiency Test: CELSA o Language Background Questionnaire o Two Vocabulary Measures (Oral & Written) o Two Memory Tests (Reading Span & Digit Span) Session two: o Main Task o Grammaticality Judgment Task on PP Comp verbs All L2 speakers underwent pre-testing. Five of the L2 participants across both experiments chose to complete the entire experiment in one day. All five were required to take at least a one hour break in between the pre-testing and the ma in task. The approximate time it took for L2 speakers to complete the pre-testing session wa s 1.5 hours. The self-paced reading task took approximately 1 hour for L2 speakers, while the eye-tracking task took 1.5 hours due to mandatory breaks. For native speakers, everythi ng except the CELSA was administered in one session and it took approximately 1.5 hours to complete, regardless of main task. Proficiency measure: Combined Englis h Language Skills Assessment (CELSA) L2 speakers of English were pr e-screened using a standardized proficiency test designed for adult speakers of English, th e Combined English Language Skills Assessment (CELSA). The CELSA is published by the Association of Classroom Teacher Testers, and places individuals within one of seven levels of prof iciency from Low Beginning to Advanced Plus.

PAGE 71

71 Participants had 45 minutes to complete the C ELSA, which requires them to read passages of increasing difficulty and fill in the blanks by choo sing one of four options. CELSA scores were used to break L2 speakers into two groups, inte rmediate and advanced in order to carry out statistical analysis. All speakers who scored in the CELSA range for Advanced Plus were grouped together and all those who scored belo w the Advanced Plus threshold were grouped together as the intermediate group for analysis purposes. Background questionnaire All participants were asked to com plete a questionnaire concerni ng language background (Appendix E). Native speakers were give n a shortened version of the Background Questionnaire. The L2 questionnaire gathered information concerning learning experience, age of acquisition, age of significant exposure, and a self-rating of language pr oficiency in the four different modalities (listening, read ing, writing, and speaking). Partic ipants were also required to self-assess their proficiency using a scal e adapted from the European Unions Language Portfolio ( http://europass.cedefop.eu.int/ ). Shipley vocabulary task All par ticipants were given the Shipley Vocabulary Task to complete. The task is a written 40-item multiple-choice vocabulary task. Participants were required to circle the word that most closely matches the target. (Appendix F) WAIS vocabulary task The W AIS vocabulary task is an oral task re quiring participants to define words in a manner that displays they know the meaning of the word. The task consists of 35 items and has a total possible score of 70. Answers are scored two points for a good synonym, a major use, a definite or primary feature, a general classification, a correct figurativ e use, a good example of an action or causal relation; one point for a vagu e or less pertinent syno nym, a minor use, an

PAGE 72

72 attribute which is not definitive, an example that merely uses the word, a concrete instance; and zero points for wrong answers, vague or trivial answers (Appendix G). Reading span task Even though working m emory is not a main factor in the experiments, all participants were given an English updated version of the R eading Span Task developed by Daneman and Carpenter (1980). The score will serve as a m easure of working memory for the purpose of maintaining the option of explor ing effects of this factor. The task was administered in PowerPoint and progress wa s controlled by the principal investig ator. In this task, participants read aloud sets of sentences (ranging from two to six sentences per set) presented on a computer screen. Upon completion of reading the final wor d, the principal investigator pressed a button to present the next sentence in the se ries. After the final sentence in a given set, a recall prompt ? appeared to indicate that the fi nal words of each sentence should be recalled by the participant. Participants were given three tw o-item practice sets before beginning, and were instructed that the number of sentences would increase after five sets of a given level. The testing ceased when a participant could not recall at leas t four out of the five sets. L2 participants were instructed to not struggle with proper pronunc iation of the words, but instead to focus on understanding. Reading span was determined at the level at whic h the participant correctly recalled three out of the five sets. If they recalled two out of the five, they were given a credit of .5. Digit span task Again, even though working m emory is not a main factor in the experiments, all participants were given an audito ry digit span task in English. The scores serve as a measure of working memory for the purpose of maintaining the opt ion of exploring effects of this factor. In this task, participants listened to a male voice read strings of digits between one and nine, beginning with a set of three digits. After hearin g the string of digits, pa rticipants typed in the

PAGE 73

73 digits they heard in the order that they heard th em. The number of digits per string was repeated five times and then an additional digit was added up to nine digits total. If the participant reported less than three of the five times inco rrect, the lower number of digits per string was repeated. Digit span was determined at the leve l at which the subject correctly recalled three out of five digit strings. If they recalled two out of the five, they were given a credit of .5. The task was administered in E-prime software and the script was altered from the digit span task listed on the Carnegie Mellon databa se of E-prime Scripts ( http://step.psy.cmu.edu/scripts-plus/ ). Post-testing grammaticality judgment questionnaire Upon com pletion of the main experimental tas k, participants were given a grammaticality judgment task consisting of 48 items, half of which were ungrammatical (Appendix H). The instructions asked participants to rate items as grammatical or ungrammatical on a scale of 1 to 5 with 1 being completely ungrammatical and 5 bei ng completely grammatical. 12 of the 48 items tested the participants knowledge of the verb argument structure a ssociated with the verbs in the main task: put, place, stick, push, drop, and lift. All other items were distracter items (36). This task took native speakers approximately 10 minutes to complete and non-native speakers approximately 15 to 20 minutes to complete. In an effort to determine the source of the ungrammaticality, participants were asked to circle the word or phrase that cau sed the sentence to be ungrammatical. Experiment 1: Self-Paced Reading Participants All participants for experim ent 1 were recrui ted from the University of Florida student body and the surrounding Gainesville community via flyer and through the LIN-CSD Pool. In total there were 32 native speakers between the ages of 18 and 35 ( M =21), 18 female. One participant was excluded as she was a non-native speaker of English. The target for each group

PAGE 74

74 began at 30 and data for native speakers were collected first to ensure the manipulation warranted testing L2 speakers. For the advan ced group there were a to tal of 19 participants, between the ages of 23 and 35 ( M =29), 10 female; and for the intermediate group, 19 participants, between the ages of 22 and 35 ( M =29), 9 female. Korean participants were more difficult to recruit, which led to the lower numbe r per group compared to the native speakers. Pre-Testing Scores Proficiency measure: CELSA CELSA scor es were used to divide the 38 L2 speakers into two groups: advanced and intermediate. In order for sp eakers to qualify for the advanced group, their scores had to fall within the raw score range of 68-75, denoting an advanced plus proficie ncy level in terms of reading, grammar and vocabulary. 19 speakers scored within this ra nge with an average score of 70 ( SD 2.17). 19 speakers scored below the threshold of Advanced Plus and were grouped together as being the intermediate group for the purposes of this study. It should be noted, however, that while none of the speakers fe ll below the CELSAs benchmark score for an intermediate level of proficiency, the scores were diverse, with sc ores ranging from High Advanced to High Intermediate with an average score of 60 ( SD 5.28). This group of speakers, thus, does not represent a true group of intermediate proficiency L2 speakers because some speakers scored in the Low and High advanced range. Background questionnaire The data collected from the Background Ques tionnaire is summarized in table 4-4. The data suggest that the two groups ar e not all that different in thei r background profiles. The main difference is that the advanced group arrived in an English-speaking country slightly earlier. These data mask differences in terms of each in dividuals purpose for living abroad. For some, they came for academic reasons and, thus, use English on a daily basis in the academic

PAGE 75

75 environment. For others, they are living abroad to accompany their spouses who were graduate students at the University of Fl orida. Unfortunately, I did not collect systematic data on this issue. I learned of many indi viduals reasons through conversation that occurred during the data collection process. Table 4-4. Background Data for Non-Native Speakers Group Age of Acquisition Age of Arrival Years in U.S. Years Learning Advanced 14 (2.4)* 25 (3.4) 3 (2.1) 14 (3.7) Intermediate 13 (.89) 27 (3.8) 2 (1.9) 16 (4.0) *standard deviation in parentheses Vocabulary and memory measures The data collected from the vocabulary and me mory measures are summarized in table 4-5. Table 4-5. Vocabulary and Memory Scores Group WAIS Oral Shipley Written Reading Span Digit Span Native 61 (8) 34 (3) 3 (.7) 7.6 (.8) Advanced 45 (11) 26 (5) 3 (.6) 8.1 (.8) Intermediate 32 (13) 20 (8) 2 (.4) 7.3 (1.1) *standard deviation in parentheses One-way ANOVAs were conducted examining performance by the three groups on each measure with language as a between subjects fact or. For all measures, there were significant differences between language groups: WAIS [ F (2,66) = 47.57, p < .001]; Shipley [ F (2,66) = 39.5, p < .001]; Reading Span [ F (2,66) = 15.67, p < .001]; Digit Span [ F (2, 66) = 3.6, p <.05]. Post hoc tests were conducted to investigate wh ether differences between specific groups were significant. For all measures ex cept digit span, differences betw een all groups were significant. Regarding digit span, native speakers did not diffe r from either group of L2 speakers (advanced

PAGE 76

76 vs. native, [ t = n.s]; intermediate vs. L2, [ t = n.s] ), but advanced speakers differed from intermediate speakers (advanced and intermediate, [ t (36) =2.7 p <.05] ). The data suggest that except for digit span, the other measures are re lated to proficiency in English, with native speakers scoring the highest. Procedures For the self-paced readin g task, a participant initially saw a se ries of dashes on the screen representing one sentence. U pon pressing the ENTER key on an external button-box, the first word of a given sentence appeared. When partic ipants finished reading that first word, they pressed ENTER again for the second word and so fort h. Participants were instructed to read at a normal pace and to not pause for extended periods of time within each sentence. After each sentence, a comprehension probe appeared. Part icipants answered yes or no by pressing designated buttons. Analysis A series of ANOVAs were conducted for each proficiency group: native, advanced an d intermediate per each of the six critical words (see Critical Words 1 through 6, table 4-3), by subjects (F1) and by items (F2). Each ANOVA was a 2 x 2 within subjects ANOVA with verb type (complement vs. adjunct) and gap (gap vs. non-gap) as the within subjects factors. All statistical analyses were conducted on RT for correct comprehension it ems only. Reading times were prepared for analysis by c onducting relative cut-offs to e liminate outliers. Any reading time per word that exceeded SD mean RT was replaced by the mean SD A 3 x 2 x 2 splitplot ANOVA was also run on the comprehension que stions with proficiency (3) as the between subjects factor and verb type (2) and gap (2) as the within subjects factors.

PAGE 77

77 Results Self-Paced Reading Comprehension accuracy Results on the com prehension probes for all groups are presented in figure 4-1. Figure 4-1. Comprehension for Experiment 1 Looking at the graph, it is evident that L2 speakers performed more poorly on gapped constructions (see table 4-2 for materials) co mpared to native speakers, but that on the nongapped structures, L2 speakers performed at the same level of comprehension. An ANOVA was conducted and main effects of gap were found [ F (1,66) = 176.32, p < .001], as well as an interaction between proficiency x gap [ F (2,66) = 20.06, p < .001]. Bonferroni post-hoc tests on between group differences suggests that all three groups performed differently compared to each other [native and advanced, p < .001; native and intermediate, p < .001; advanced and

PAGE 78

78 intermediate, p < .02]. Further analysis was perf ormed on the non-native data to investigate the nature of the differences between these two gr oups. When the native group was dropped from the analysis, there was no proficiency x gap interaction [ F < 1, n.s.]. The only significant effect was a main effect of gap [ F (1,36) = 153.06, p < .001]. These results suggest that gap conditions are difficult for both advanced and intermediate speakers in this expe riment, regardless of proficiency level. To some extent, this is no t surprising given that th ese structures are more complex and much more infrequent than direct object extraction in English. What this result may suggest is that wh -extraction from prepositio nal phrases is a structure that presents difficulty for L2 speakers, and as such, comprehension suffers. Self-paced reading: native speakers The predic tion for native speakers is that they would be sensitive to subcategorization frame when resolving filler-gap dependencies. Such an influence is expected to manifest itself in a smaller RT difference between condition A and B (PP complement gap and non-gap conditions) compared to C and D (PP adjunct gap and non-gap conditions) at or soon after the filled-gap (table 4-2). Table 4-6 summarizes the within-subjects (F1) and within-items (F2) ANOVA results for native speakers for each critical word, table 4-7 presents mean RT and standard error, and figure 4-2 displays th e means associated with these effects. The area of most interest for the predictions is the filled-gap region, cr itical words 2 and 3. In the analysis by subjects, the beginning of th e filled-gap region, at the determiner after the verb, a main effect of gap was found to approach significance [ F (1,30) = 3.6, p = .068], as was an interaction between verb type and gap [ F (1,30) = 3.86, p = .059].

PAGE 79

79 Table 4-6. Native Speakers: Se lf-Paced Reading Summary ANOVAs Source F1 (1,30) p F2 (1,34) p At the Verb (Critical Word 1) Verb Type (VT) 3.24 .082 .030 .864 Gap (G) .450 .509 2.15 .152 VT*G 2.14 .153 .012 .914 At the Det1 (Critical Word 2) Verb Type (VT) .204 .655 1.55 .221 Gap (G) 3.60 .068 .413 .525 VT*G 3.86 .059 .487 .490 At the Noun (Critical Word 3) Verb Type (VT) 1.03 .319 .538 .468 Gap (G) 10.59 .003** .367 .549 VT*G 4.99 .033* 2.98 .094 At the Prep (Critical Word 4) Verb Type (VT) .016 .900 .042 .839 Gap (G) 2.81 .104 4.89 .034* VT*G .038 .847 .016 .900 After the Gap (Critical Word 5) Verb Type (VT) 1.85 .184 .037 .849 Gap (G) 2.05 .163 .739 .396 VT*G .831 .369 .496 .486 After the Gap (Critical Word 6) Verb Type (VT) 1.15 .292 1.38 .248 Gap (G) 10.10 .003** .275 .604 VT*G .983 .329 .880 .355 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 Table 4-7. Native Speakers: Mean Self -Paced Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 418 (19.7) 411 (15.7) 425 (20.4) 432 (13.8) 417 (20.5) 443 (22.2) B 442 (17.4) 456 (22.1) 402 (15.4) 408 (17.7) 406 (13.4) 416 (20.7) C 419 (18.8) 435 (16.1) 461 (23) 432 (17.5) 412 (17.7) 444 (23.3) D 408 (16.5) 443 (17.6) 389 (12.2) 404 (16.8) 384 (12.6) 384 (10.7) CW = critical Word, (Standard Error ).

PAGE 80

80 Figure 4-2. Native Speakers: Self-Paced Reading Time From examining the graph, it can be seen that th ese effects, at the determiner, are not in the direction predicted. One reason fo r this is that the non-gap contro l conditions took longer to read at this position. These results do not provide clear support or counter evidence for the Williams et al. (2001) argument that native speakers are sensitive to the s yntactic cue of the filled-gap represented by determiner; thus, if the determiner does serve as a syntactic cue for native speakers, it is not evident in this group of subjects. These effect s are also not present in the F2 by items analysis.

PAGE 81

81 More importantly, at the noun of the filled-ga p, the predicted reading time patterns are present. At the noun, there is a main effect of gap [ F (1,30) = 10.59, p < .01], and an interaction with verb type and gap [F (1,30) = 4.99, p < .05]. From examining the mean RT, the noun in the PP Complement condition was read 36 ms faster than the noun in the PP Adjunct condition. More relevant to the predictions, the difference between the PP complement condition (A) versus the non-gap control condition (B) (23ms) was not as great as the difference between the PP adjunct condition (C) versus the non-gap control condition (D) ( 72ms). These results suggest that native speakers are using subc ategorization frame to influence parsing decisions; that is, the native speakers are applying their knowledge that the put -type verbs have a nother potential gap location coming up because those verbs require a pre positional phrase for grammaticality. The F2 analysis does not fully support th at this effect is independent of the subjects tested, as the interaction of verb type and gap is only marginally significant at this location [ F (1,34) =2.98 p = .09]. Self-paced reading: advanced speakers The prediction for advanced speakers is that they, like native speakers, would be sensitive to subcategorization frame when resolving filler-gap dependenc ies. Such an influence is expected to manifest itself in the same patterns in RT as predicted for native speakers, a verb type by gap interaction in the filled-gap region. If these results are not found in the results from the self-paced reading experi ment, there is the possibility that th e lack of results is related to the task and may be found in the eye-tracking expe riment. Table 4-8 summarizes the withinsubjects (F1) and within-items (F2) ANOVA results for advanced speakers per critical word, table 4-9 presents mean RT and standard error, and figure 4-3 displays th e means associated with these effects.

PAGE 82

82 Table 4-8. Advanced Speakers: Self-Paced Reading Summary ANOVAs Source F1 (1,18) p F2 (1, 26) p At the Verb (Critical Word 1) Verb Type (VT) .205 .656 .009 .925 Gap (G) .824 .376 .005 .943 VT*G .746 .399 .838 .368 At the Det1 (Critical Word 2) Verb Type (VT) .870 .363 .309 .583 Gap (G) .004 .949 2.78 .107 VT*G .993 .332 .193 .664 At the Noun (Critical Word 3) Verb Type (VT) .004 .951 .086 .772 Gap (G) .921 .350 .765 .390 VT*G 4.13 .057 2.71 .112 At the Prep (Critical Word 4) Verb Type (VT) 1.22 .284 4.16 .052 Gap (G) 3.72 .069 .048 .828 VT*G .042 .840 .105 .749 At the Prep2 (Critical Word 5) Verb Type (VT) .103 .752 .179 .676 Gap (G) 5.73 .028* 7.79 .010* VT*G .510 .484 .204 .655 At the Det2 (Critical Word 6) Verb Type (VT) .693 .416 1.06 .311 Gap (G) 11.26 .004** 7.21 .012* VT*G .113 .741 .601 .445 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance, **p < .01,*p < .05 Table 4-9. Advanced Speakers: Mean Se lf-Paced Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 867 (68.2) 806 (54.1) 585 (37.5) 748 (64.2) 654 (40.9) 694 (91.3) B 797 (39.5) 840 (38.3) 598 (27) 672 (35.4) 619 (38.9) 463 (25.8) C 811 (60.2) 875 (55.9) 631 (42.1) 785 (47.9) 672 (54) 729 (65.5) D 806 (52.9) 846 (50.8) 554 (22.9) 695 (38.3) 577 (27.8) 479 (23.1) CW = critical Word, (Standard Error )

PAGE 83

83 Figure 4-3. Advanced Speakers: Self-Paced Reading Time Based on the results represented in the graph above, it is evident that the RT patterns for advanced speakers are not identical to those of th e native speakers. The in teraction for advanced speakers at the filled gap noun is marginally significant [F(1,18) = 4.13, p = .057], and the gap effects begin at critical word 5, the second preposition after the actual gap [F1(1,18) = 5.73, p <.05; F2 (1,26) = 7.79, p < .05]. These main effects of gap also spill over onto critical word 6. The marginally significant interaction at the noun in the filled-gap region may reflect an attempt on behalf of the advanced readers to apply their knowledge of the subcategorization frame associated with these verbs in an on-line manner. The source of this interaction is based

PAGE 84

84 on the difference in means between conditions A and B (PP Comp) compared to C and D (PP Adjunct), as it is with the native speakers. The mean difference in RT between the PP complement gap condition (A) and the non-gap cont rol (B) is 13 ms (not in the anticipated direction), whereas the mean difference in RT between the PP adjunct gap condition (C) and the non-gap control (D) is 77 ms. Since only a marginal interaction at the noun of the filled-gap was found, there is no evidence that the advanced speakers are using the determiner as a syntactic cue. These results are similar to Williams et al. (2001), yet it is not clear that the native speakers are using the determiner as a syntactic cue either. The main eff ects of gap are also later than that of the native speakers, which most likely reflects spill-over. The source of these spil l-over effects could be related to difficulty in processing filler-gap depe ndencies. To summarize, the advanced speakers may be attempting to apply subcategorization fram e to aid in filler-gap resolution, but what is clear is that processing filler-g ap dependencies causes difficulty for advanced readers after the filled-gap region; which implies delayed processing compared to native speakers. Self-paced reading: intermediate speakers The predic tion for intermediate speakers is th at they, unlike native speakers and advanced speakers, would not be sensitive to subcategorizati on frame when resolving filler-gap dependencies. This prediction would manifest itself in a lack of an interaction at or soon after the filled-gap. For intermediate speakers, the results are similar to neither th e native group nor the advanced group. Table 4-10 summarizes the w ithin-subjects (F1) and within-items (F2) ANOVA results for advanced speakers per criti cal word, table 4-11 presents mean RT and standard error, and figure 4-4 displays th e means associated with these effects.

PAGE 85

85 Table 4-10. Intermediate Speakers: Self-Paced Reading Summary ANOVAs Source F1 (1,18) p F2 (1,30) p At the Verb (Critical Word 1) Verb Type (VT) 3.39 .082 1.39 .246 Gap (G) .247 .635 .06 .809 VT*G .004 .953 1.28 .267 At the Det1 (Critical Word 2) Verb Type (VT) 8.75 .008** 1.64 .210 Gap (G) 5.67 .028* .45 .510 VT*G .851 .368 1.88 .181 At the Noun (Critical Word 3) Verb Type (VT) .548 .469 .12 .731 Gap (G) .120 .733 .32 .576 VT*G 1.552 .260 1.73 .198 At the Prep (Critical Word 4) Verb Type (VT) 2.075 .167 2.57 .119 Gap (G) .279 .604 3.38 .076 VT*G 3.017 .099 1.75 .195 At the Prep2 (Critical Word 5) Verb Type (VT) 2.56 .121 .045 .833 Gap (G) 2.79 .112 .024 .879 VT*G 2.65 .121 .061 .807 At the Det2 (Critical Word 6) Verb Type (VT) 1.84 .192 .686 .414 Gap (G) 26.75 .000** 4.51 .042* VT*G 1.35 .261 .699 .410 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance, **p < .01,*p < .05 Table 4-11. Intermediate Speakers: Mean Self-Paced Reading Time per Critical Word Cond CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 970 (76.8) 719 (48.4) 574 (40.6) 759 (34.2) 721 (53.5) 752 (81.1) B 993 (72.6) 815 (45.5) 556 (37.6) 741 (37.6) 629 (39.6) 490 (43.6) C 858 (56.9) 855 (51.6) 529 (38.4) 677 (46) 632 (49.4) 639 (55.2) D 888 (45.9) 873 (54.8) 565 (39.3) 724 (43.6) 617 (40.3) 475 (35.6) CW = critical Word, (Standard Error )

PAGE 86

86 Figure 4-4. Intermediate Speakers: Self-Paced Reading Time Focusing on the filled-gap region (Critical Words 2 and 3) there is no interaction between verb type and gap. There are main effects at th e determiner, but they are difficult to interpret, and are not present on the following critical word, critical word 3. At the determiner, the gap conditions (A and C) take intermediate reader s less time (marginal mean: 787 ms) compared to the non-gapped conditions (B and D) (marginal mean : 844 ms). There is also a main effect of verb type with the adjunct conditions (C and D) taking longer (marginal mean: 864 ms) than the complement conditions (A and B) (marginal mean : 767 ms). The verb type effects could be spillover from the verb, as there is a marginally si gnificant main effect of verb type at the verb

PAGE 87

87 itself [ F1 (1,18) = 3.39, p = .082], which might be related to length. Even though a marginally significant interaction is present in the F1 analysis at the noun in the filled-gap, this is not in the predicted direction. While the results are difficult to interpret, it is clear that these speakers are not being influenced by subcategorization frame in the sa me manner as native speakers, nor does it appear that they are attempting to apply this informa tion when they encounter the filled-gap, as the advanced speakers appear to be doing. The comprehension data supports that these readers may not be successfully resolving th ese types of dependencie s overall, since they often performed at chance on the comprehension questions. The main effect of gap at the verb, however, suggests that what is more likely is that are actively filling the gap, but when they encounter the filledgap, they cannot revise their init ial interpretations. For both groups of L2 speakers, first pass eye-tracking measures may be more useful in ex amining what influences these readers as they encounter these filled-gaps. Post-testing grammaticality judgment questionnaire One obvious question at this poi nt is whether or not these non-native speake rs have the sufficient off-line knowledge of thes e particular verbs subcategorization fra me. The purpose of the post-testing grammaticality judgment task was to ensure that L2 speakers had this knowledge. As previously stated, Korean subcat egorization frames for th e verbs tested in the current study are reported to be similar (personal communication, Sangyeon Park, Jin Park); but the task was to ensure that the speakers were fa miliar with the frames in English. Overall, L2 speakers rejected sentences with put -type verbs that did not have PPs in comparison to push-type verbs [ t (226) = -4.67 p<.001]. Furthermore, in a 2 x 2 split-plot ANOVA with verb type (2) as the within-subjects factor and proficiency group (2) as the between-subjects factor, no differences were found between how the two groups performed [ F (1,36) = .063, p =.804]. What

PAGE 88

88 these results suggest is that all L2 speakers ha d some knowledge of the subcategorization frames associated with the verbs being tested. The re sults from the off-line task suggest that the speakers have difficulty applying this off-line knowledge in an on-line manner when resolving filler-gap dependencies. Discussion The m ost significant results to come out of this experiment are that the results from Goodluck et al. (1995) are supporte d in that native speakers are influenced by subcategorization frame information while resolving filler-gap de pendencies on-line. This influence does not, however, manifest itself in comprehension as there were no signi ficant differences found between the two gap conditions (A and C). More relevant to the study at hand is that the advanced speakers also appear to be influenced by verb subcategorization frame during the resolution of these types of dependencies. The effect was only marginally significant, suggesting that processing load, a lesser degree of automa ticity, and possibly lower levels of vocabulary may play a role in how this information is used in an on-line manner. It is important to note that the verbs examined in the current study are repor ted to have the same argument structure in Korean; furthermore, on the post-testing gramma ticality judgment task, advanced speakers rejected put -type verbs without PPs more often than push-type verbs without PPs. Lastly, in comparison to advanced speakers and native speake rs, intermediate speakers did not demonstrate any influence of subcategorization frame when re solving filler-gap dependencies. Again, like the advanced speakers in the post-test gram maticality judgment task, they did reject put -type verbs without PPs more often than push -type verbs without PPs. The key finding of this experiment concerns the differences between the groups described above. What may be happening is that as proficie ncy increases, L2 speake rs are able to more efficiently attend to lower level processes duri ng reading, leading to a greater level of these

PAGE 89

89 processes being automatized. This might mean that for more advanced speakers processing resources are freed to attend to higher level pro cessing, such as revising in itial predictions about gap locations and using lexical cues associated with the verb to form new predictions. The advanced speakers in this study, however, are not on par with native speak ers, and the question remains as to whether they can achieve the same level of subcategoriz ation frame usage during this type of complex processing. This ques tion may, however, be answered by using a more sensitive technique. As pointed out by Frenck-Mestre (2005), us e of verb argument structure during the processing of ambiguous sentences by a dvanced French learners was only revealed by first pass reading times collected through ey e-tracking. Self-paced reading may not be measuring true first pass readi ng since overall reading is slower, and as mentioned earlier, it has been argued that self-paced r eading may reflect total reading time (Garnsey, 1997). In addition, self-paced reading may place more demands on working memory since readers may already be having difficulty remembering the beginning of the sentence by the time they reach the critical areas of the sentence. As a last point, self-paced reading may also produce effects that would not occur in normal reading since readers are first, slowed down, and second, forced to read every word in the sentence; something they may not do in normal reading. Experiment 2: Eye-Tracking This experim ent is a replication of experime nt 1 using eye-tracking instead of self-paced reading. Participants All participants for experim ent 2 were also r ecruited from the University of Florida student body and the surrounding Gainesville community via flyer and through the LIN-CSD Pool. In total there were 22 native speakers between the ages of 18 and 29 ( M =20), 12 female. The target for each group began at 20. For the advanced group there were a tota l of 11 participants,

PAGE 90

90 between the ages of 20 and 35 ( M =29), 5 female; and for the intermediate group, 13 participants, between the ages of 19 and 35 ( M =23), 10 female. Korean participants were more difficult to recruit, which led to the lower number per group compared to the native speakers. Pre-Testing Scores Proficiency measure: CELSA CELSA scor es were used to divide the 24 L2 speakers into two groups: advanced and intermediate. Groups were determined in the same manner as in experiment 1. 11 speakers scored within the range necessary to be considered advanced, with an average score of 71 ( SD 1.8). 13 speakers scored below the threshold of Advanced Plus and we re grouped together as being the Intermediate group for the purposes of this study. It should be noted as it was in experiment 1 that, while none of the speakers fell below the CELSA benchmark score for an intermediate level of proficiency, the scores we re more diverse, with scores ranging from high advanced to low intermediate with an average score of 50 ( SD 8.9). Background questionnaire The data collected from the Background Qu estionnaire is summarized in table 4-12. Table 4-12. Background Data for Non-Native Speakers Group Age of Acquisition Age of Arrival Years in U.S. Years Learning Advanced 13 (.8)* 24 (5.9) 5 (4.5) 16 (4.8) Intermediate 15 (3.6) 21 (2.9) 2 (2.7) 8 (5.0) *standard deviation in parentheses The background data supports the observati on that the intermediate group in this experiment is closer to an intermediate proficiency than was the intermediate group in experiment 1. These speakers have spent approx imately half the time learning English compared

PAGE 91

91 to the advanced speakers. They have also been immersed in an English speaking country for approximately half the time of advanced learners. Vocabulary and memory measures The data collected from the vocabulary measures are summarized in table 4-13. Table 4-13. Vocabulary and Memory Scores Group WAIS Oral Shipley Written Reading Span Digit Span Native 62 (4.9) 33 (2.4) 3 (1) 7.6 (1) Advanced 45 (11.9) 28 (3.4) 2 (2.3) 7.2 (1) Intermediate 20 (10.3) 19 (4.8) 3 (1) 7 (1) *standard deviation in parentheses One-way between-subjects ANOVAs were conduc ted examining performance by the three groups on each measure. Significant differences were found between groups for both vocabulary measures: WAIS [ F (2,43) = 97.86, p < .001]; Shipley [ F (2,43) = 66.83, p < .001]; but no differences were found with regard to the memory measures: Reading Span [ F < 1, n.s.]; Digit Span [ F < 1, n.s.]. Post hoc tests were conducte d on the vocabulary measures to investigate whether differences between specific groups were significant. For bot h vocabulary measures, differences between all gr oups were significant. Post-testing grammaticality judgment questionnaire Upon com pletion of the main experimental tas k, participants were given a grammaticality judgment task consisting of 65 items, half of which were ungrammatical (Appendix I). The instructions asked participants to rate items as grammatical or ungrammatical on a scale of 1 to 7 with 1 being completely ungrammatical and 7 bei ng completely grammatical. 12 of the 65 items tested the participants knowledge of the verb argument structure a ssociated with the verbs in the main task: put, place, stick, push, drop and lift For each verb, there were two sentence types:

PAGE 92

92 one with a direct object and a prepositional phrase, and another with only a direct object. This task took native speakers approximately 10 minutes to complete and non-native speakers approximately 15 minutes to complete. Procedures For the m ain task, participants read senten ces presented on a computer screen in their entirety. A head-mounted binocul ar eye tracking device, the EyeL ink II, was used to monitor participants eye movements while they read sentences. After a calibration procedure, participants were presented w ith one full sentence per scree n. The participants were all instructed to read at a normal pace and to press a button upon completion of the sentence. Following the item was a yes/no comprehension probe. Analysis A series of ANOVAs were conducted for each proficiency group: native, advanced an d intermediate per each of the six critical words (see Critical Words, table 4-3), by subjects (F1) and by items (F2). Each ANOVA was a 2 x 2 with in subjects ANOVA with verb type (comp vs. adjunct) and gap (gap vs. non-gap) as the within subjects factors. All statistical analyses were conducted on first pass reading times, total gaze du ration (total reading time) and regression-path time for correct comprehension items only. First pass reading times were prepared for analysis by conducting absolute cut-offs as suggested in Albrecht and Clifton, 1998: Fixations that were above 800 ms or below 140 ms were excluded from analysis. The total percentage of fixations excluded across all subjects ( N =46) was 6.9%. A 3 x 2 x 2 split-plot ANOVA was also run on the comprehension questions with proficiency (3) as the between subjects factor and verb type (2) and gap (2) as the with in subjects factors.

PAGE 93

93 Results Comprehension accuracy Results on the com prehension probes for all groups are presented in Figure 4-5. Figure 4-5. Comprehension for Experiment 2 From looking at the graph, it appears that as proficiency decreased, comprehension decreased. Statistical tests support this result. An ANOVA was conducted and main effects of gap were found [ F (1,66) = 176.23, p < .001], as well as an interaction between proficiency x gap [ F (2,66) = 20.22, p < .001]. Bonferroni post-hoc tests on between group differences suggests that all three groups performed differently co mpared to each other [native and advanced, p < .001; native and intermediate, p < .001; advanced and intermediate, p < .02]. Further analysis was performed on the non-native data to investig ate the nature of the differences between these two groups. When the native group was dropped fr om the analysis, there was no proficiency x

PAGE 94

94 gap interaction [ F < 1, n.s.]. The only significant e ffect was a main effect of gap [ F (1,22) = 18.03, p < .001]. These results suggest that gap cond itions (for example materials see table 4-2) are difficult for both advanced and intermediate speakers in this expe riment, regardless of proficiency level; again, as pointed out in experiment 1, these result s are not surprising given that these structures are more complex and much more infrequent than direct object extraction in English. Native speakers Based on th e experimental predictions outlined earlier, it is unclear if first pass reading times will demonstrate the use of subcategorization frame by native speakers. If subcategorization frame information is not used immediately, there is th e possibility that the expected interaction between verb type and gap in the filled-gap region (Critical Words 2 and 3; table 4-2) may be detected in eye-tracking measures that reflects later processing, such as total gaze duration or regression-path time, which is the sum of all fixations, including regressions, before any fixations are made to the right of that region. The resu lts from a 2 x 2 ANOVA on first pass reading times are summarized in table 4-14, table 4-15 presents mean RT and standard error, and figure 4-6 displays the me ans associated with these effects.

PAGE 95

95 Table 4-14. Native Speakers: First Pass Reading Time Summary ANOVAs Source F1 (1, 21) p F2 (1,34) p At the Verb (Critical Word 1) Verb Type (VT) .383 .542 .193 .661 Gap (G) 1.437 .244 .133 .718 VT*G .221 .643 .000 .999 At the Det1 (Critical Word 2) Verb Type (VT) .410 .529 .409 .527 Gap (G) 4.50 .047* .835 .368 VT*G 3.20 .089 2.02 .166 At the Noun (Critical Word 3) Verb Type (VT) .023 .882 .375 .545 Gap (G) 7.92 .010* 5.12 .029* VT*G 1.12 .303 3.16 .085 At the Prep (Critical Word 4) Verb Type (VT) .220 .644 .053 .820 Gap (G) 15.06 .001** 12.31 .001** VT*G .065 .801 .394 .535 After the Gap (Critical Word 5) Verb Type (VT) .017 .896 .556 .461 Gap (G) 9.97 .005* 2.04 .163 VT*G .135 .171 .234 .632 After the Gap (Critical Word 6) Verb Type (VT) .782 .387 2.69 .111 Gap (G) 2.24 .150 2.68 .112 VT*G .027 .871 .529 .473 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 Table 4-15. Native Speakers: Mean First Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 314 (11) 269 (12.1) 281 (12.8) 287 (15) 287 (14.5) 273 (15.5) B 301 (13.3) 263 (12.5) 247 (8.2) 256 (9.4) 250 (10.5) 255 (8) C 305 (13.8) 277 (12.1) 273 (14) 293 (12.5) 283 (11.2) 267 (15.6) D 299 (15.5) 241 (10) 258 (9.5) 257 (9.3) 257 (14.3) 245 (8) CW = critical Word, (Standard Error ).

PAGE 96

96 Figure 4-6. Native Speakers: Eye-Tr acking First Pass Reading Time In experiment 1, there was a significant inte raction between verb t ype and gap at the noun in the filled-gap, position 3. In first pass r eading times, no significant interaction was found at this position. In the by subjects analysis, th ere are, however, main effects of gap, and a marginally significant interaction at the de terminer, position 2, of the filled-gap [ F (1,21) = 3.20, p = .089]. Figure 4-6 helps to illustrate th e nature of this marginal interaction. The difference between the PP complement condition (A) and the non-gap control (B) is only 6 ms; whereas the difference between the PP adjunct condition (C) and the non-gap control (D) is 36 ms. This interaction, however, is on ly marginally significant, nor does the by items

PAGE 97

97 analysis support the result. A few explanations may help to cl arify why these effects are not present in first pass reading times. One explanati on is that subcategorizat ion frame may not have an immediate influence when native speakers revise their initial predictions as to the location of the gap when they encounter a filled gap. In th is case, an interaction should be found in eyetracking measures that reflect later processes. In examining total gaze duration, there is also no interaction between verb type and gap at either of the filled-gap positions or at later pos itions. A marginally significant interaction is found, however, for regression-path time at the noun of the filled gap, critical word 3 [ F1 (1,21) = 3.64, p = .071], with a 651 ms mean difference in RT between conditions C and D, compared to a 47 ms mean difference in RT between cond itions A and B; in both cases, the gap condition taking longer than the non-gap condition (table 4-16 presents the means for regression-path time). Table 4-16. Native Speakers: Mean Regr ession-Path Time per Critical Word Cond CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 402 (49.9) 896 (290.4) 444 (45.2) 347 (21.6) 647 (152.2) 1103 (209.3) B 401 (40) 356 (40.4) 379 (34.2) 371 (44.8) 304 (31.9) 310 (18.4) C 432 (64.8) 380 (43.9) 556 (65.1) 399 (42.2) 590 (121.5) 1057 (291.7) D 398 (36.1) 347 (55.6) 307 (25.2) 272 (11.2) 284 (18) 328 (27.7) CW = critical Word, (Standard Error ). This interaction suggests that subcategoriz ation frame prevents difficulties for native speakers, as evident by fewer regressions at the noun of th e filled-gap after put -type verbs. Taken together, the marginal interaction in firs t pass RT at the determin er of the filled-gap (critical word 2) and the marginal interaction in the regression-path time (critical word 3) suggest

PAGE 98

98 that native speakers overall have more difficulty revising initial predictions when the gap has been extracted from an adjunct position. Advanced speakers Given that s ubcategorization frame played a later role in aiding native speakers during filler-gap resolution, it is expected that advanced speakers may also use subcategorization during later processing. It is still important, however, to examine first pass RT to see whether subcategorization frame plays an early role for advanced speakers. The results from a 2 x 2 ANOVA on first pass reading times are summarized in table 4-17, table 4-18 presents mean RT and standard error, and figure 4-7 displays the means associated with these effects.. Table 4-17. Advanced Speakers: Firs t Pass Reading Time Summary ANOVAs Source F1 (1,10) p F2 (1,28) p At the Verb (Critical Word 1) Verb Type (VT) 19.23 .001** 6.08 .020* Gap (G) .011 .918 .648 .428 VT*G .409 .537 2.39 .133 At the Det1 (Critical Word 2) Verb Type (VT) .603 .455 .109 .745 Gap (G) .030 .866 .464 .504 VT*G 1.38 .267 .030 .864 At the Noun (Critical Word 3) Verb Type (VT) .803 .391 2.88 .102 Gap (G) .192 .671 .155 .737 VT*G 1.16 .306 .456 .506 At the Prep (Critical Word 4) Verb Type (VT) 1.63 .229 .053 .821 Gap (G) 1.67 .226 .659 .426 VT*G .045 .836 1.44 .243 After the Gap (Critical Word 5) Verb Type (VT) .102 .756 .135 .717 Gap (G) 22.47 .001** 26.52 .001** VT*G 20.45 .001** .463 .503 After the Gap (Critical Word 6) Verb Type (VT) 1.13 .313 .571 .461 Gap (G) 7.58 .020* 6.85 .019* VT*G .543 .478 .935 .348 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05

PAGE 99

99 Table 4-18. Advanced Speakers: Mean Firs t Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 376 (21.1) 268 (15.8) 387 (20) 315 (19.5) 347 (20.1) 342 (17.9) B 367 (14.3) 295 (13.7) 395 (16.2) 302 (20.1) 273 (18.1) 387 (25.7) C 422 (28.3) 281 (24.2) 387 (12.5) 338 (23.3) 376 (23.4) 344 (18.8) D 434 (16.9) 260 (24.6) 366 (18.3) 316 (20.4) 256 (13.3) 414 (18.6) CW = critical Word, (Standard Error ) Figure 4-7. Advanced Speakers: Eye-Tracking First Pass Reading Time

PAGE 100

100 While there are no significant results in the filled-gap region (Critical Words 2 and 3), there are later effects at critical word 5, the word after the pre position, which is the actual gap. At this position, there is a main ef fect of gap by subjects and by items [ F1 (1,10) = 22.47, p < .001; F2 (1,28) = 20.45, p < .001], and an interaction between gap and verb type, by subjects only [ F (1,10) = 20.45, p < .001]. When looking at the directio n of the interaction (figure 4-7), these results suggest that there may be an advant age for these advanced sp eakers for dealing with filled-gaps after put -type verbs. The difference between the PP complement condition (A) and the non-gap control (B) is 74 ms; whereas the difference between the PP adjunct condition (C) and the non-gap control (D) is 120 ms. This interaction appears late compared to the results from self-paced reading, in which there is a marginally sign ificant interaction within the filled-gap region. It is likely that these effects are spill-over effects, but because they are so late, it does raise the question as to whether the active filler strategy is being used by L2 speakers during normal reading tasks. To my knowledge, there are no studies usin g eye-tracking that examine th e question as to whether L2 speakers use this strategy, only self-paced reading st udies. It is feasible given the results from self-paced reading studies on fille r-gap processing may actually draw L2 speakers attention to the filler-gap dependency as they are forced to comprehend sentences on a word by word basis; meaning that the task guides L2 speakers to pay attention to both lexical and syntactic structures in a manner different to that of how they would read normally. Another po ssibility is that selfpaced reading reflects revision, as well as initial processing; as su ch, it can be argued that total gaze duration and regression-path time are better m easures to compare with self-paced reading, since they reflect later processe s. Yet, the total gaze duration and regression-path time analyses did not result in any interactio ns between verb type and gap.

PAGE 101

101 One effect that has been overlooked in the disc ussion thus far is the main effect of verb type at the verb itself, position 1 [F1 (1,10) = 22.47, p < .001; F2 (1,28) = 20.45, p < .001]. These results suggest that L2 speakers are trea ting the two classes of verbs differently. One reason for this may be because the PP complement verbs are shorter than the PP adjunct verbs. This was unavoidable given the process by which these verbs were chosen. Another, more empirically supported explanation of these verb t ype effects is related to the morphology of the PP complement verbs. Two out of the three verbs are irregular past tense forms. As pointed out in Portin and Laine (2001), native speakers pr ocess morphologically complex words in an equal amount of time, regardless of whether they are ir regular or regular; wher eas, non-native speakers process mono-morphemic irregular forms faster than regular forms that encode the same grammatical information through multiple morphemes. As discussed in the next section, data from intermediate speakers displa y the same verb type effects at the verb, which reinforces the idea that irregular past tense forms may be the source of this effect. Intermediate speakers So me of the results from the 2 x 2 ANOVAs fo r intermediate speakers are similar to those of advanced speakers. The main difference, however, is a lack of a verb type by gap interaction in first pass reading time. Table 4-19 summari zes the results of the ANOVAs by critical word, table 4-20 presents mean RT and standard error, and figure 4-8 displays the means associated with these effects.

PAGE 102

102 Table 4-19. Intermediate Speakers: Fi rst Pass Reading Time Summary ANOVAs Source F1 (1,12) p F2 (1,22) p At the Verb (Critical Word 1) Verb Type (VT) 14.41 .003** 15.05 .001** Gap (G) 1.79 .206 1.34 .258 VT*G 1.42 .256 .707 .409 At the Det1 (Critical Word 2) Verb Type (VT) .412 .533 .986 .333 Gap (G) .895 .363 .927 .347 VT*G 1.22 .290 1.76 .200 At the Noun (Critical Word 3) Verb Type (VT) .738 .407 .371 .549 Gap (G) .146 .709 .015 .904 VT*G .151 .705 .188 .668 At the Prep (Critical Word 4) Verb Type (VT) .165 .692 2.72 .116 Gap (G) .404 .537 .259 .617 VT*G .062 .808 .338 .568 After the Gap (Critical Word 5) Verb Type (VT) .857 .373 .434 .517 Gap (G) 47.67 .000** 21.04 .001** VT*G .000 .990 .149 .704 After the Gap (Critical Word 6) Verb Type (VT) .734 .408 .103 .752 Gap (G) 3.46 .087 3.32 .087 VT*G .001 .971 .735 .404 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 Table 4-20. Intermediate Speakers: Mean First Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 CW 5 CW 6 A 374 (12) 260 (11.7) 418 (19.4) 333 (12.9) 398 (14.5) 387 (29.3) B 400 (11.9) 287 (22.4) 407 (14.8) 339 (20.9) 275 (14.4) 428 (19.3) C 449 (16.1) 261 (13.1) 424 (14) 335 (21.9) 387 (17.1) 371 (22.1) D 446 (13.4) 265 (12.4) 424 (18.9) 350 (14.3) 264 (14.8) 414 (21.3) CW = critical Word, (Standard Error )

PAGE 103

103 Figure 4-8. Intermediate Speakers: Ey e-Tracking First Pass Reading Time Main effects of gap were found at the word following the preposition, critical word 5 (during/the), which is after the actual gap, as with the advanced speakers [ F1 (1,12) = 47.67, p < .01; F2 (1,22) = 21.04, p < .001]. Figure 4-8 illustrates this main effect of gap. Like the advanced group, there is a main effect of verb type at the verb itself [ F1 (1,12) = 14.41, p < .01; F2 (1,22) = 15.05, p < .001]. Both types of effects raise th e same questions discussed in relation to the advanced results. The que stion of whether intermediate speakers are using the active-filler strategy is further called into question by the fact that in the self-paced reading time data the main effects of gap are late. These late effects o ccurred in the analyses of total gaze duration as

PAGE 104

104 well, where gap effects did not appear until position 5 [ F1 (1,12) = 27.63, p < .001]. Overall, there was no evidence that intermediate speake rs were sensitive to subcategorization frame during filler-gap resolution. There were no interactions between verb type and gap found in any of the eye-tracking measures, suggesting that intermediate speakers are not sensitive to subcategorization frame during filler-gap processing; a conclusion also drawn from the selfpaced reading data. Post-testing grammaticality judgment questionnaire Again, the question m ust be revisited as to whether or not these particular non-native speakers have the sufficient off-line knowledge of th ese verbs subcategoriza tion frame. Overall, L2 speakers rejected sentences with put -type verbs that did not have PPs, in comparison to pushtype verbs [ t (142) = p<.001]. Furthermore, in a split -plot ANOVA with verb type (2) as the within-subjects factor and language group (2) as the between-subjects factor, no differences were found between how th e two groups performed [ F (1, 21) = .041, p =.766]. What these results suggest is that all L2 speakers had so me knowledge of the subcategorization frames associated with the verbs being tested. The re sults from the off-line task suggest that the speakers have difficulty applying this off-line knowledge in an on-line manner when resolving filler-gap dependencies. Discussion To summ arize the eye-tracking results, re sults suggest that native speakers use subcategorization frame in a rapid manner, at the determiner of the filled-gap, during filler-gap resolution. For native speakers, when a verb's subcategorization frame does not require an additional complement from which wh -extraction could occur, diffi culties arise during revision of initial gap predictions. The results also s uggest that advanced speakers are influenced by subcategorization frame, but the effects occur la ter than when reading in a self-paced reading

PAGE 105

105 paradigm. Finally, there was no evidence that intermediate speakers used subcategorization frame. The use of eye-tracking allowed for first pass reading time, which is indicative of initial influences on processing, to be separated from to tal reading time and regr ession-path time. Since the presentation mode involved in collecting ey e-tracking data also resembles more natural reading, it presents a more accurate picture of how readers build comprehension. The results from eye-tracking suggest that task influences processing for native and advanced speakers, but not for intermediate speakers, who struggle with comprehension regardless of task. Conclusions The research questions investigated through these two experim ents were: Do L2 speakers use subcategorization frames to posit potential gap locations; that is, will they be influenced by the subcategorization type of the verb to inform gap sites? Do strategies change as proficiency increases? Will data from self-paced readi ng and eye-tracking converge? Even though the data from self-paced reading and eye-tracking is not entirely convergent, such non-convergence does not impact the answer s that the data provide for the first two questions. In both experiments, there was evidence that advanced speakers were sensitive to subcategorization frame during filler-gap resolu tion, while intermediate speakers were not. As such, it does appear that strategies change as proficiency increases. The hypothesis that runs through both of these experiments is that intermediate speakers woul d not pay attention to lexical information associated with the verb due to in efficient processing. Intermediate speakers were proposed to be less automatic at lower level proc esses such as decoding skills and lexical access; thus, when higher order skills are needed to reconstruct non-local dependencies, processing resources were predicted to be overburdened. Overburdened resources were posited to prevent the intermediate L2 parser from using information that is available off-line because there is not

PAGE 106

106 enough attention and working memory left for th e parser to access this more subtle lexical information associated with the verb. Bo th experiments support this hypothesis. Overall, the results imply that as proficie ncy increases, processing becomes more nativelike. Advanced speakers appear to be developing a level of awareness of this factor, but not to be applying it in the same efficient manner as native speakers do. From a processing perspective, the results imply that intermediate speakers may not have enough processing resources to pay attention to, notice or apply this informa tion while resolving a complex dependency, while more advanced speakers do. Given that both advanced and intermediate speakers display familiarity with these verbs subcategorization frames in an off-line task, this is an interesting finding because it demonstrates difficulties at a processing level, not at a knowledge level. The answer to the third question concerns native and advanced speakers since the conclusions for intermediate speakers in both expe riments are the same. It appears that selfpaced reading for native and advanced speakers is re flective of more than just initial influences on processing. As stated earlier, self-paced re ading may place more demands on a reader, as words can only be read once and, therefore, must be kept in working memory as the reader progresses through the sentence. Eye-tracking, on the other hand, allows readers to regress back to earlier words, as in normal reading, which mean s that if readers forget or encounter difficulty during the sentence, they can retu rn to earlier words and reread ar eas of difficulty. For native speakers in these experiments, the effects of su bcategorization frame manifested themselves in the filled-gap region regardless of task. The key difference for native speakers is that regressionpath time demonstrated that s ubcategorization frame plays a la te role. With the advanced speakers, it appears that self-paced reading slows readers down e nough so that they are able to

PAGE 107

107 apply subcategorization frame at an earlier posit ion in the sentence; while in the more natural reading task, the use of this information plays a late role in terms of having an influence at a later position. As a result, task demands associated wi th self-paced reading gives the appearance that advanced speakers are quicker at recognizing a filled-gap and using lexical information associated with the verb to revise analysis. In this sense, these subtle differences suggest that eye-tracking may be the better method to use when comparing native and non-native processing. On the other hand, when looking at differences between two different proficiency levels of nonnative speakers, both methods produced similar results. There are two aspects in this discussion that have been overlooked. One is that the L2 speakers experience difficulty at the preposition and not in the filled-gap region during revision, while the native speakers experi ence difficulty at the noun in the filled-gap; and the other concerns comprehension performance by L2 speake rs. In addressing the first issue, the only indication that any of the L2 sp eakers are using an active filler-s trategy to resolve dependencies is from the advanced speakers self-paced readin g data. The interaction for this group at the noun shows that these readers were disrupted in the filled-gap region, which suggests that the gap had been filled at the verb. None of the measures from the eye-tracking data support the use of the active filler strategy by L2 speakers. To the contrary, the eye-tracking data indicates that gap effects for both L2 groups occur after the actua l gap. If eye-tracking is more indicative of natural reading strategies these results suggest that L2 read ers wait until the gap to resolve this type of dependency. This conclusion does not co ntradict the results of the previous research investigating L2 filler-gap resolution (Felse r and Roberts 2007, Juffs 2005, Love et al. 2003, Marinis et al. 2005, Williams et al. 2001, Williams 2006). Previous studies using tasks other than eye-tracking (Williams 2006) found effects onl y for good performers, and those were very

PAGE 108

108 late in the filled gap, or at th e location of the gap. In conclu sion, the results from both of the experiments in this chapter may indicate that unless L2 readers have enough processing resources available and are drawn to use an active filler strategy by the task, they may not engage in the use of this strategy. The other issue that needs to be considered is comprehension. For both experiments, analyses on comprehension performance indicates that while the mean of the advanced speakers was higher than intermediate speakers while proce ssing gapped structures, these differences were not statistically significant. Poor performance on these tested structures indicates that all of the L2 speakers experienced difficulty with wh -extraction from prepositional phrases. It may be the case that the L2 speakers are not familiar with ex traction for prepositional phrases, thus, lacking the off-line knowledge necessary to resolve such depe ndencies. To investigate this issue, future research should include a grammaticality judgment task that directly tests speakers familiarity with this type of extraction. The other possi bility is that L2 speakers may not have enough resources to resolve these dependenc ies. If this is the case, th ere could be two consequences. First, they could give up tr ying to comprehend the sentence or they could keep trying to comprehend but not have enough resources left to access the verbs argument structure. By looking only at the comprehensi on data, it would appear that neither L2 group has enough resources left to deal with extraction from PPs; ye t, the reading time patter ns suggest that the L2 speakers are trying to use this information during revision, but ar e not successful. Without a direct measure of speakers familiarity with th is type of extraction, there is a confound across both experiments as to whether th e patterns found are due to processing difficulty or a lack of knowledge.

PAGE 109

109 Overall, it is still unclear whether all L2 sp eakers utilize the active filler strategy during normal reading. The evidence does support the us e of such a strategy for advanced speakers during a self-paced reading task. The next chapter further explores the issue of the active filler strategy, but instead of focusing on revision processe s, which manifest themselves in the filledgap region, the next experiment in this dissertation focuses on effects that are predicted to manifest themselves on the verb itself. It does this by looking at verb bias while manipulating filler plausibility.

PAGE 110

110 CHAPTER 5 INFLUENCE OF VERB BIAS DURING FILLER-GAP RESOLUTION Experiment 3: Introduction W hile the experiments in the last chapter addressed the issue of whether a verbs subcategorization frame aids in filler-gap proces sing, the experiment discussed in this chapter, experiment 3, looks at the issue of whether verb bias prevents the positing of gaps at the verb itself. Previous monolingual research has sugge sted that native speakers utilize knowledge of a verbs transitivity frequency (verb bias) in making on-line predictions about upcoming gap locations when resolving filler-gap dependencies (Garnsey et al. 1997, Stowe et al. 1991, Tanenhaus et al. 1989); whether late -learning L2 learners utilize the same verbal properties in the same manner while processing filler-gap depende ncies has not previously been investigated. As defined in chapter two, verb bias is item -specific information concerning the frequency with which a particular verb appears with a given argument structure. Recall the example presented in chapter two: Push is a verb with a transitive bias because it occurs most frequently in constructions with NP direct objects; whereas crash is a verb with an intransitive bias because it occurs in most frequently in constructions without NP direct objects (Gahl et al. 2004). An important distinction, also made in chapter two, is that verb bias differs from plausibility information. For example, the verb push occurs most frequently with NP direct objects, but not all NP direct objects are equally plausible, for example, The man pushed the river So, while both types of information constitute lexical info rmation, plausibility information is not encoded within a specific verbs lexica l entry but must be computed on a case by case basis. As discussed in chapter two, the results from Stowe et al. (1991) demonstrate that manipulating verb bias and filler plausibility is a useful tool for investigating whether speakers use verb bias to predict upcoming gaps. Stowe et al. (1991) mani pulated filler plausibility and verb bias and

PAGE 111

111 found that gaps were not posited after intransitive verbs; the ev idence in support of verb bias preventing gap-positing is that a plausibility effect was found for transitive bias verbs (RT increases at the transitive verb after implausible fillers), but not for intransitive bias verbs. The influence of verb bias on filler-gap resolution has not been previously tested in L2 speakers. Williams et al. (2001) and Williams ( 2006) did investigate L2 sensitivity to filler plausibility during filler-gap re solution, but did not manipulate verb bias in either experiment. Their results suggest that L2 sp eakers may be influenced by the le xical property of plausibility when resolving filler-gap dependencies, but only under certain circumstances. The current experiment uses eye-tracking to inve stigate the issues outlined above. Why use Eye-Tracking as Opposed to Self-Paced Reading? As has already been discussed, the results from Williams et al. (2001) and Williams (2006) suggest that non-nativ e speakers may be influenced by th e lexical property of plausibility when filling gaps. The tasks that were used by these studies were the stop-making-sense task and the self-paced reading paradigm. The stop-ma king-sense task required conscious judgments to be made as each word was revealed, pl acing emphasis on these judgments and increasing reading time per word. The task, therefore, may have produced unnatural resu lts in that the task could have been training L2 speakers to use bot h the active filler strategy and plausibility, as opposed to testing whether L2 speakers would use th is lexical property in a natural reading task (Williams et al. (2001) acknowledge this problem). Williams (2006) which used self-paced reading only found effects of plausibility for good L2 performers when resolving filler-gap dependencies. Another potential problem with se lf-paced reading is that it has also been found to slow readers down, and may place added demands on readers working memory. As Garnsey et al. (1997) point out, re aders may slow down more than norm al with self-paced reading, which

PAGE 112

112 argues that as a research techniqu e it is not sensitive to initial in terpretation, but rather includes revision. In further support of using ey e-tracking to test L2 speak ers, Frenck-Mestre (2005a;b) argues that it is a more accurate method for re vealing differences betw een native and non-native speakers, and that self-paced read ing may in fact be misleading. She points out several studies whose results could only have been revealed th rough eye-tracking. One such study examined the contribution of sentential context on lexical access in light of comp eting models (Altarriba et al. 1996). While the details are not re levant for the current study, th e point Frenck-Mestre makes is that differences emerged only in first fixation duration and skipping proba bilities; two measures not available in self-paced read ing. More relevant, is anothe r study, which found differences in first pass gaze durations and first fixation duratio ns (Frenck-Mestre and Pynte 1997). In this study, the role of a verbs argument structure in ambiguity resolution was investigated. It was found that verb type impacted first pass gaze du rations, and that non-native speakers showed a greater sensitivity to this lexical property than native speakers, as revealed by first fixation duration differences. The author argues that verb bias may play a larger role for non-native processing, and that overall eyetracking, which provides multi-faceted data, is a better technique for investigating non-native senten ce processing. Experiment 2 of the current study, presented in chapter four, also suggests that eye-tracking may be a more a ppropriate method, given that it revealed different areas of di fficulty for advanced speakers compared to native speakers. Participants As with the experim ents discusse d in chapter four, L2 particip ants in this experiment are all late learners who share the same L1 background, Korean. Late-learning is defined as having learned English after the age of 12, and Kor ean was chosen as a common L1 background because it is often classified as a wh -insitu language (See chapter four for discussion). The

PAGE 113

113 participants for this experiment were also the sa me individuals who particip ated in experiment 2. The way that this was accomplished is that th e experimental items from experiments 1 and 2 (which are identical) served as part of the dist racter items in experiment 3, while experimental items from experiment 3 served as part of the di stracter items for experiment 2. In total, there were 22 L1 speakers between the ages of 18 and 29 ( M =20), 12 female. For the advanced group there were a total of 11 participan ts, between the ages of 20 and 35 ( M =29), 5 female; and for the intermediate group, 13 participants, between the ages of 19 and 35 ( M =23), 10 female. Research Questions and General Predictions The research questions which guide this experim ent are: Will a natural reading task reveal that plausibility influences L2 filler-gap resolution for both groups of L2 speakers? Will there be evidence that verb bias (transitivity frequency) influences L2 filler-gap resolution? Overall, do L1 and L2 sensitivities to verb bias and plausibility di ffer? Do these differ among L2 proficiency groups? The general prediction is that advanced speakers will be more sensitive to lexical information during on-line filler-gap resolution due to fewer processing demands associated with greater automatization in L2 reading; this means that they, like native speakers, will be influenced by lexical information associated with the verb to aid in positing gap locations. Intermediate speakers, on the other hand, are hypo thesized to be less sensitive to lexical information due to higher processing demands associated with lesser automatization in L2 reading, on top of the demands associated with resolv ing non-local dependencies; mean ing that, unlike native speakers and advanced speakers, intermediate speaker s will ignore lexical information during gap positing.

PAGE 114

114 Materials In this exper iment, filler plausibility and ve rb bias were varied and eye-movements were recorded. Verbs with either a transitive bias or intransitive bias we re used to construct 32 quadruplets of the type illustrated in Table 5-2 (Appendix J). Item s were divided into four lists through a Latin-square procedure such that each list contained 8 different sentences for each of the four conditions. A total of 72 distracter it ems were included (Appendix K), yielding a total of 108 sentences per participant. 36 of the distracter items per list were the experimental items from experiments 1 and 2 (Appendix A). Table 5-1. Verb Bias Example Materials Condition* Critical Words A T, PL The manager wondered which waiter the customer /asked/rudely/a bout/during/ the lunch rush. B T, IMPL The manager wondered which plate the customer /asked /rude ly/about/during/ the lunch rush. C I, PL The manager wondered which waiter the customer /rushed/rude ly/around/during/ the lunch rush. D I, IMPL The manager wondered which mop the customer /rushed/r udely/around/during/ the lunch rush. *While the verbs vary in transitivity frequency from high to low, for the sake of convenience, they are labeled T or I for tran sitive and intransitive; PL a nd IMPL stand for plausible and implausible. Critical words are separated by /. Fillers are underlined. The verbs chosen were narrowed down from a list of 45 intransitive bias verbs and 48 transitive bias verbs, which were compiled from a verb argument structur e database (Gahl et al. 2004). All verbs had at least a 3 to 1 ratio in favor of either bias. In total, 15 intransitive bias verbs were used ( crashed, cheered, escaped, floated, fle w, hurried, marched, pointed, raced, rested, rushed, stood, sunk, walked, yelled ) and 16 transitive verbs were used ( advised, asked, attacked, baked, called, cleaned, copied, guarde d, heard, killed, lost, painted, read, reviewed, saved, visited ). The verb choice was based on the bias ratio, as well as how easy it was to construct both plausible and impl ausible conditions with those verbs. No verb appeared more than three times per list.

PAGE 115

115 In order to construct plausible and impla usible conditions, two plausibility rating questionnaires were administered. First, a plau sibility judgment questionn aire was administered to 31 participants, all native, m onolingual speakers of English. Thes e participants were recruited from the general student body of the University of Florida and the LIN-CSD Pool. Participants for this questionnaire pa rticipated on a voluntary basis or, if recruited th rough the pool, for course credit. The questionnaire consisted of 86 experimental item s (43 plausible and 43 implausible) and 14 filler items (Appendix L). Partic ipants were instructed to rate each item as to how plausible it was on a scale of 1 to 7 with 1 being implausible and 7 being plausible. An example item is, How plausible is it for a driver to crash a truck? The logic behind rating this type of structure is that the direct object woul d be serving as the filler in the experimental conditions of the main task; for example, The man saw which truck the driver crashed carelessly into after the wild party. All plausible items that receiv ed a score below 5 were eliminated as a potential plausible item, and all implausible items that received a score above 3 were elim inated as a potential implausible item. Items that failed to fall into the proper range of plausibi lity were revised and a second plausibility questionnaire was administered to 15 partic ipants. This second questionnaire contained 75 experimental items and 25 distracters (Appendix M). For some verbs, as many as three different items were tested to ensure a thir d round of plausibility judgment was not needed. The same criteria were used to eliminate items. After the second ques tionnaire, 32 item sets were established with an average rating of 2.1 on the scale of 1 to 7 for implausible items and 6.1 for plausible items. A t-test confirms that overall the ratings across the 32 items sets for implausible and plausible condi tions differed significantly [t (126) = -34.42, p <.001].

PAGE 116

116 Experimental Predictions Native and Advanced Speakers Verb bias is expected to guide parsing decisions for both of these groups. This m eans that an intransitive verb bias should prevent both native and advanced speakers from filling the gap at the verb. Plausibility is the tool used to dete ct this, which means that an interaction between verb bias and plausibility is expected at the verb or soon af ter. Because these effects are expected to have a rapid effect on the active filler strategy, this interaction is expected in first pass reading times. Specifically, it is expected th at first pass reading time for condition A, T/PL, will be shorter than that of condition B, T/IMPL and that first pass reading time for T/IMPL will be the longest compared to all other conditions. Intermediate Speakers Verb bias is not expected to guide gap predicting for interm ediate speakers, as they are expected to use a hyper -active gap filling strategy, whereby verb bias is ignored. This prediction could manifest itself in two different results. One result is that there would be a plausibility effect at the verb regardless of verb type, which would argue that plausibility is being computed in an on-line manner. The other possibility is th at all gaps are filled at the verb, regardless of verb bias or plausibility of f iller. In this case, there should be no significant main effects or interactions at or soon after the verb. Methods Pre-Testing Materials The pretesting m aterials were admi nistered in the systematic orde r outlined in ch apter four.

PAGE 117

117 Post-Testing Grammaticality Judgment Questionnaire Upon com pletion of the main experimental tas k, participants were given a grammaticality judgment task consisting of 48 items, half of which were ungrammatical (Appendix N). The instructions asked participants to rate items as grammatical or ungrammatical on a scale of 1 to 7 with 1 being completely ungrammatical and 7 bei ng completely grammatical. Participants were instructed to rate items that were not completely grammatical or completely ungrammatical as a 3, 4, or 5; items such as The man imitated accurately. The purpose of administering this task was to test whether or not speakers accepted a transi tive reading of an intransitive verb and vice versa. It should be acknowledged that it is diffi cult to test participants knowledge of verb bias; so instead, the task was designed to ensure reve rse readings were acceptable. Data for one participant was not collected due to an error by the experimenter. Pre-Testing Scores As the participants were the sam e individuals who participated in experiment 2, all of the scores on pre-testing can be f ound in chapter four under Experiment 2, Pretesting Scores. Procedures For the m ain task, participants read sentence s in their entirety presented on a computer screen. A head-mounted binocular eye-tracking device, the EyeL ink II, was used to monitor participants eye movements while they read sentences. After a calibration procedure, participants were presented w ith one full sentence per scree n. The participants were all instructed to read at a normal pace as they w ould in a book or a magazine and to press a button upon completion of the sentence. Following the item was a yes/no comprehension probe. Analysis A series of ANOVAs were conducted for each group: native, advanced an d intermediate per each of the four critical words (see Critical Words, table 5-1), by subjects (F1) and by items

PAGE 118

118 (F2). Each ANOVA was a 2 x 2 within subj ects ANOVA with verb type (T vs. I) and plausibility (PL vs. IMPL) as the within subjects factors. All statistica l analyses were conducted on first pass reading times, total gaze duration (total reading time), and regression-path time for correct comprehension items only. First pass re ading times were prepared for analysis by conducting absolute cut-offs as suggested in Albrecht and Clifton, 1998: Fixations that were above 800 ms or below 140 ms were excluded from analysis. The total percentage of fixations excluded across all subjects ( N =46) was 12%. A split-plot ANOVA was also run on the comprehension questions with proficiency (3) as the between subjects fact or and verb type (2) and gap (2) as the within subjects factors. Results Comprehension Accuracy Results on the com prehension probes for all groups are presented in figure 5-1. A significant main effect of verb type was found [ F (1,43) = 16.96, p < .001], as was an interaction between verb type and proficiency [ F (1,43) = 6.27, p < .01]. A main effect of proficiency [ F (2,43) = 16.75, p < .001] was also found. Bonferroni post-hoc tests on between group differences suggests that there were differences in the performance between all three groups overall [native and advanced, p < .05; native and intermediate, p < .001; advanced and intermediate, p < .05]. Overall, the L2 speakers perfor med more poorly compared to the native speakers (mean percentages: native, 84; advanced, 74; inte rmediate, 64). The post-hoc tests demonstrate that there is decline in comprehensi on as proficiency level d ecreases. A separate ANOVA conducted on the L1 data confirms that native speakers did not perform differently based on verb type.

PAGE 119

119 Figure 5-1. Comprehension for Experiment 3 The interaction between verb t ype and proficiency is a result of better performance by both groups of L2 speakers on the sentences with intr ansitive bias verbs, whereas the native speakers performed similarly on both verb types (see table 5-1 for materials). These results indicate that sentences were easier for L2 speakers to comp rehend when the verb was intransitive, which suggests that L2 speakers had difficulty resolving the dependency when the verb bias reflected a preference for a direct object, ev en when the filler was implausi ble. Sentences containing verbs that prefer a direct object may ha ve presented conflicting informati on in that L2 readers preferred a direct object gap location after transitive bias verbs, even when filling that gap resulted in an implausible event (e.g., The student copied the last seat. ). On the other hand, when the sentence

PAGE 120

120 contained an intransitive verb, this information a ppears to have helped L2 speakers in terms of comprehension, regardless of proficiency level. Native Speakers The predic tion for native speakers is that they would be sensitive to verb bias when resolving filler-gap dependencies. Such an in fluence is expected to manifest itself in an interaction between verb type and plausibility at the verb or soon after in first pass reading times. Specifically, the difference in first pass reading tim e is expected to be larger between conditions A and B with condition B taking the longest (for example mate rials see table 5-1), while no differences are expected between conditions C and D. The reason for this is that gaps should not be posited at an intransitive verb, so there should be no plausibility effect for intransitive conditions (C and D) or for the plausible transitive verb condi tion (A). Table 5-2 summarizes the within-subjects (F1) and with in-items (F2) ANOVA results for native speakers for each critical word, and table 5-3 presents the mean s and standard errors per critical word. Table 5-2. Native Speakers: First Pass Reading Time Summary ANOVAs Source F1 (1,21) p F2 (1,31) p At the Verb (Position 1) Verb Type (VT) 1.89 .183 1.05 .313 Plausibility (P) 3.17 .089 1.97 .170 VTP 0.35 .558 .051 .821 At the Adverb (Position 2) Verb Type (VT) .009 .922 .426 .518 Plausibility (P) .179 .675 .362 .551 VTP .088 .768 .124 .726 At the Prep1 (Position 3) Verb Type (VT) .705 .410 1.01 .325 Plausibility (P) .382 .543 .366 .550 VTP .191 .666 3.04 .095 After the Prep2 (Position 4) Verb Type (VT) 2.50 .128 1.41 .245 Plausibility (P) .283 .600 .396 .534 VTP .682 .417 1.16 .289 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05

PAGE 121

121 Table 5-3. Native Speakers: Mean First Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 A 292 (12.6) 323 (13.5) 300 (14.8) 290 (16.7) B 307 (14.9) 325 (13.2) 288 (13) 292 (16.1) C 299 (16.6) 318 (15.2) 288 (16.8) 323 (18.6) D 326 (13.4) 328 (14.1) 287 (14.5) 304 (12.8) CW = critical Word, (Standard Error ). The analyses of first pass reading times did not reveal any significant differences between the conditions for native speakers. One observation to make based on these results is that verb type does not play an immediate role in whether these speakers posit a gap at the verb or not. The results of Pickering and Traxler (2003) cert ainly suggested this conclusion; yet, a main effect of plausibility was found in their study. In this experiment, there is a hint that native speakers are influenced by plausibi lity at the verb, as indicated by a marginally significant main effect of plausibility, which the F2 analysis does not support. This marginal effect is a result of implausible conditions taking longer at the verb [316 ms for implausible conditions vs. 295 ms for plausible conditions]. Due to the marginal nature, the results indica te that plausibility information may not play an immediate role in ga p-filling; thus, examini ng later measures, such as total gaze duration or regression-path time, may be more re vealing. It must also be acknowledged that small sample size could be dr iving the marginal nature of this effect. Table 5-4 summarizes the ANOVAs conducted on total gaze duration, and table 5-5 presents the means and standard errors per critical word.

PAGE 122

122 Table 5-4. Native Speakers: To tal Gaze Duration Summary ANOVAs Source F1 (1,21) P F2 (1,34) p At the Verb (Position 1) Verb Type (VT) .424 .521 .024 .875 Plausibility (P) 2.28 .145 1.22 .278 VTP 1.77 .197 2.34 .130 At the Adverb (Position 2) Verb Type (VT) .976 .334 .062 .436 Plausibility (P) .789 .384 .752 .392 VTP 3.50 .075 1.01 .323 At the Prep1 (Position 3) Verb Type (VT) 8.57 .009** 5.33 .030* Plausibility (P) .245 .625 1.87 .184 VTP 2.14 .157 .046 .831 After the Prep2 (Position 4) Verb Type (VT) .123 .728 .191 .665 Plausibility (P) 4.26 .051 4.79 .037* VTP .577 .455 .649 .427 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 Table 5-5. Native Speakers: Mean Total Gaze Duration per Critical Word Condition CW 1 CW 2 CW 3 CW 4 A 610 (55.4) 600 (63.6) 528 (45.1) 582 (41.9) B 718 (74.1) 666 (72.8) 555 (55.9) 537 (41.7) C 632 (59.5) 613 (54.9) 468 (38.8) 612 (47.6) D 647 (87.7) 593 (58.1) 417 (31.9) 521 (60.2) CW = critical Word, (Standard Error ) The results suggest that during revision native speakers are sensitive to both verb type and plausibility. First, there is a marginally signi ficant interaction at the adverb. When examining the direction of the interaction, it appears that the implausible transitive condition takes longer overall to read. While the interaction is weak, it does imply that it is most difficult to recover when the verb bias and the filler plausibility support filling the gap at the verb. At the preposition (about/around), which is the gap location, total read ing time with transitive verbs

PAGE 123

123 takes longer (figure 5-2), sugges ting that there is a preference for gaps after prepositions following intransitive bias verbs. Finally, in th e last critical position, th ere are main effects of plausibility, suggesting that fill ers that are plausible at the ve rb have a lingering effect. Figure 5-2. Native Speakers: Total Gaze Duration Regarding regression-path time, no significant main effects or interactions were found. Since regression-path time calculates regressive fixations before progressing beyond a critical word, the lack of effects suggests that participants read through a ll of the critical words before returning to reread areas of di fficulty. Overall, these results suggest that verb bias and plausibility play a late role in revision for native speakers, alt hough plausibility may have a more immediate impact.

PAGE 124

124 Advanced Speakers The prediction for advanced speakers is that th ey, like native speakers, should be sensitive to verb bias when resolving filler-gap dependencie s. Given that the predicted effects were not found for native speakers and the low number of advanced partic ipants, such a re sult is unlikely. Table 5-6 summarizes the within -subjects (F1) and within-items (F2) ANOVAs conducted on first pass reading time for advanced speakers per cr itical word, table 5-7 presents mean RT and standard error, and figure 5-3 displays the means associated with these effects. Table 5-6. Advanced Speakers: Firs t Pass Reading Time Summary ANOVAs Source F1 (1,10) p F2 (1,20) p At the Verb (Position 1) Verb Type (VT) .001 .970 1.11 .303 Plausibility (P) .307 .592 3.13 .091 VTP 1.19 .302 1.30 .265 At the Adverb (Position 2) Verb Type (VT) 1.91 .199 .216 .646 Plausibility (P) 4.35 .066 .065 .801 VTP 2.53 .145 1.20 .286 At the Prep1 (Position 3) Verb Type (VT) .017 .897 .060 .809 Plausibility (P) .040 .844 .034 .855 VTP .008 .966 2.57 .134 After the Prep2 (Position 4) Verb Type (VT) 6.60 .037* .193 .665 Plausibility (P) 1.13 .321 .000 .986 VTP .772 .408 .012 .912 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05

PAGE 125

125 Table 5-7. Advanced Speakers: Mean First Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 A 439 (34.3) 306 (19.7) 342 (19.9) 236 (23.7) B 480 (28.6) 315 (21.8) 338 (23.3) 234 (18.2) C 466 (25.3) 306 (15.4) 341 (30.7) 266 (11.6) D 451 (15.9) 373 (29.1) 335 (20.7) 300 (19.3) CW = critical Word, (Standard Error ) A marginally significant main effect of plausibility was f ound for the advanced speakers during first pass reading of the adverb, as was a main effect of verb type at the second preposition, although these results are not supported by F2 analyses The marginally significant plausibility effect at the adverb is likely a spillover effect from the verb, which suggests that the advanced speakers may be slower than native speakers at applying this lexical information. Although this effect is one word downstream compared to native speakers, the advanced speakers also take longer for implausible condi tions [344 ms for implausible conditions vs. 306 ms for plausible conditions]. Re turning to the inconclusive results from experiment 2, what the plausibility effect at the adverb suggests is that advanced learners do appe ar to be using an active filler strategy of gap filling; ot herwise, plausibility should not have any impact at this position. The second notable result is the ma in effect of verb type at th e second preposition. Looking at figure 5-3, it is evident that the intransitive verb conditions take longer to read in this position, which might be spillover of an integration effect from the real gap position, which is one word before this second preposition. Compared to native speakers, advanced speakers first pass reading time suggests that they rely more on lexi cal factors than native speakers, although these

PAGE 126

126 factors have an influence later th an expected. It is difficult to tease apart whether these effects are genuine first pass spillove r effects or revision. Figure 5-3. Advanced Speakers: First Pass Reading Time Turning to the late effects as reflected in total gaze duration and regression-path time, advanced speakers do not demonstrate the same e ffects as native speakers do. In fact, the only significant effects in total gaze dura tion occurred in the F2 analysis at the verb (See table 5-8 and 5-9). Average RT at the verb for transitive conditions was 1421 ms, while average RT at the verb for intransitive conditions was 1175 ms; a di fference of 246 ms. Advanced speakers spent more time overall reading the verb when it was tr ansitive, regardless of whether the filler was plausible. Since total gaze dura tion reflects total reading time, r eaders already know the gap is at the preposition and not the verb, as such they ma y be experiencing difficulty with transitive bias

PAGE 127

127 verbs because these verbs prefer to take a direct object. Intransitive bias verbs may be easier to integrate at the preposition downstream, because th ey do not typically take a direct object; thus, readers may prefer PP extraction following intran sitive bias verbs. These results are also supported by the comprehension accuracy, where th ese speakers performed better when the verb bias was intransitive. As with the native speakers, no significant main effects or interactions were found for regression-path time. Table 5-8. Advanced Speakers: Total Gaze Duration Summary ANOVAs Source F1 (1,10) p F2 (1,20) p At the Verb (Position 1) Verb Type (VT) .174 .685 4.54 .045* Plausibility (P) 1.24 .289 3.94 .060 VTP 1.94 .193 .262 .614 At the Adverb (Position 2) Verb Type (VT) .319 .585 .607 .444 Plausibility (P) .575 .467 .280 .602 VTP .276 .611 .346 .562 At the Prep1 (Position 3) Verb Type (VT) .087 .744 3.19 .090 Plausibility (P) .171 .688 .049 .827 VTP 1.01 .341 .457 .511 After the Prep2 (Position 4) Verb Type (VT) .719 .415 .063 .803 Plausibility (P) .254 .625 .000 .976 VTP 3.91 .076 .298 .591 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 Table 5-9. Advanced Speakers: Mean To tal Gaze Duration per Critical Word Condition CW 1 CW 2 CW 3 CW 4 A 1116 (157.1) 1213 (183.6) 777 (100.8) 758 (102) B 1325 (212.1) 1354 (168.9) 735 (95.5) 730 (97) C 1255 (195.1) 1337 (225.5) 671 (87.1) 725 (101.6) D 1273 (180.9) 1361 (243) 797 (160.7) 831 (104.1) CW = critical Word, (Standard Error ).

PAGE 128

128 Intermediate Speakers The predic tion for intermediate speakers is th at they, unlike native speakers and advanced speakers, would not be sensitive to verb bias when resolving filler-gap dependencies; thus, either a plausibility effect at the verb regardless of verb type or no significant main effects or interactions at or soon after the verb are expected. Table 5-10 pr esents first pass reading time for intermediate speakers, table 5-11 presents mean RT and standard error, and figure 5-4 displays the means associated with these effects. Table 5-10. Intermediate Speakers: Fi rst Pass Reading Time Summary ANOVAs Source F1 (1,12) P F2 (1,18) p At the Verb (Position 1) Verb Type (VT) 12.41 .004* 5.15 .042* Plausibility (P) .450 .515 1.83 .200 VTP 1.02 .332 1.58 .232 At the Adverb (Position 2) Verb Type (VT) 3.48 .088 .217 .648 Plausibility (P) .042 .840 3.95 .066 VTP .161 .174 .000 .999 At the Prep1 (Position 3) Verb Type (VT) .644 .437 .732 .414 Plausibility (P) 1.97 .185 .680 .430 VTP 1.05 .324 1.50 .251 After the Prep2 (Position 4) Verb Type (VT) .071 .794 .679 .420 Plausibility (P) 1.29 .281 .000 .988 VTP .598 .456 .012 .911 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05

PAGE 129

129 Table 5-11. Intermediate Speakers: Mean Fi rst Pass Reading Time per Critical Word Condition CW 1 CW 2 CW 3 CW 4 A 489 (34.5) 296 (16.7) 362 (18.4) 286 (24.9) B 475 (24.9) 324 (21.8) 375 (27.8) 325 (31.1) C 407 (16.4) 362 (27.6) 329 (17.8) 300 (17.4) D 454 (23.7) 323 (19.9) 387 (20.3) 300 (27.3) CW = critical Word, (Standard Error ). Interestingly, unlike both na tive speakers and advanced speakers, intermediate speakers appear to be influenced by verb type at the verb ; both F1 and F2 analyses support this result. At the verb, the intransitive bias conditions (430 ms) are read appr oximately 50 ms faster than transitive bias conditions (482 ms). Intermedia te speakers may be demons trating that they are more sensitive to the lexical prope rty of verb bias when processi ng filler-gap dependencies, but it is unclear why the transitive bias verbs are taking l onger to read than the intransitive bias verbs. In Stowe et al. (1991), verb bias was found to be used only becau se an interaction was found at the verb, with transitive bias verbs and implausible fillers taking the longer to read. The longer reading time for intermediate speakers for transiti ve bias verbs may reflect integration of the filler at the transitive bias verbs, or it may reflect readers having more difficulty with gaps at transitive bias verbs. In addition, this result suggests that intermediate speakers, like advanced speakers, do appear to be filling gaps at the verb, hence using the active filler strategy. The surprising aspect of these results is that there ar e no plausibility effects at any position in the F1 analyses. There is a plausibility effect at the adverb in the F2 analyses with implausible bias conditions taking 57 ms longer to read.

PAGE 130

130 Figure 5-4. Intermediate Speaker s: First Pass Reading Time In looking at total gaze duration (see table 5-12), there is a marginal effect of plausibility at the adverb, which was not found in the F2 analys is. Since total gaze duration reflects later processing, it may be the case that intermediate speak ers return to all of the critical regions in an attempt to fully comprehend these complex sentences but overall their attention is not drawn to different areas across conditions. As for regressi on-path time, the only significant finding is a main effect of plausibility at the first preposition (about/aroun d), where items with plausible fillers take 249 ms longer to read th an items with implausible fillers [ F (1,12) = 5.74, p < .05]. This means that intermediate speakers regressed to earlier parts of the sentence and then back to the preposition more when the fill er was plausible. This finding contradicts one interpretation of the first pass reading time effects at the verb, whic h were argued above to suggest that gaps were

PAGE 131

131 not being filled at intransitive bias verbs. Longer regression-path time at the preposition suggests that the speakers did fill the gap duri ng first pass reading for all plausible items, including intransitive verbs, and because of this, they regressed out of and reread the preposition longer to revise the initial analysis. Table 5-12. Intermediate Speakers: Total Gaze Duration Summary ANOVAs Source F1 (1,12) p F2 (1,18) p At the Verb (Position 1) Verb Type (VT) .456 .511 .121 .733 Plausibility (P) 2.62 .131 3.33 .092 VTP 2.89 .114 2.79 .120 At the Adverb (Position 2) Verb Type (VT) 1.78 .208 .414 .529 Plausibility (P) 3.46 .089 2.54 .132 VTP .167 .690 .276 .606 At the Prep1 (Position 3) Verb Type (VT) .443 .518 .866 .376 Plausibility (P) .009 .922 .648 .441 VTP .217 .649 .174 .686 After the Prep2 (Position 4) Verb Type (VT) .001 .976 .003 .951 Plausibility (P) 1.61 .227 3.70 .070 VTP .451 .514 1.31 .265 df are the same across all ANOVAs and are lis ted with the column labels. marginal significance,**p < .01,*p < .05 The main effect of plausibility in regre ssion-path time might also be reflective of plausibility becoming available late to speakers of lower levels of proficiency; meaning that when these readers encounter the verb on firs t pass, they do not compute how plausible the dependency is, just whether the dependency can be resolved based on verb bias alone. Then at some later stage, but before reading the preposit ion, plausibility information becomes available, causing reanalysis based on plausi bility. The preposition, which repr esents the gap location, then surprises readers, forcing them to regress back to earlier parts of the sentence and then reread the preposition.

PAGE 132

132 Post-Testing Grammaticality Judgment Questionnaire Upon com pletion of the main task, all partic ipants were administered a grammaticality judgment task to investigate whether participants accepted transitive readings of the intransitive bias verbs and intransitive readings of the transitive bias verbs (e.g., Intrans: The printer hurried the job.. and Trans: The policeman guarded all night long. ). The ratings, which were on a scale of 1 to 7 with 1 being ungrammatical and 7 bein g grammatical, were analyzed in a split-plot ANOVA with proficiency (3) as th e between subjects condition and verb bias (2) as the withinsubjects condition. Main effects of verb bias [ F (1,42) = 16.47, p < .001] and proficiency [ F (2,42) = 11.18, p < .001] were found, as was an interacti on between verb bias and proficiency [F (1,42) = 4.00, p < .05]. Bonferroni post-hoc tests on pr oficiency demonstrate that there were differences between native speaker s and both L2 groups, but not betw een the L2 groups in their overall ratings [native and advanced, p < .001; native and intermediate, p < .001; advanced and intermediate, p < .999]. In general, native speakers rated both (a) transi tive readings of the intransitive bias verbs and (b) intransitive readings of the transitive bias verbs higher than L2 speakers did (native mean rating for (a) 5.2 and (b) 5.9; advanced for (a) 4.3 a nd for (b) 4.7; intermediate for (a) 4.5 and (b) 4.6). Across proficiency groups, participants pref erred (b) intransitive readings of transitive verbs (mean rating for (a) 4.6 and (b) 5.1). Thes e differences across group indicate that L2 speakers are less tolerant, whereas native speakers are more tolerant, of the infrequent argument structure associated with each bias. It is diffi cult to say whether this reflects participants knowledge of verb bias frequency because the qu estionnaire does not directly tap into this knowledge; consequently, these results may suggest th at the verb bias eff ects for L2 speakers are a result of a lack of knowledge that a transitive verb can occur intran sitively in English. As a last note, when examining mean rating per item, two intransitive verbs ( hurried and marched) fell

PAGE 133

133 below 4, the midpoint on the scale for native sp eakers; whereas thirteen verbs fell below this midpoint for advanced speakers (Intrans: escaped, flew, hurried, marched, rested, rushed, stood, sunk, yelled; Trans: advised, called, killed, saved ), and ten verbs fell below this midpoint for intermediate speakers (Intrans: cheered, escaped, flew, marched, walked; Trans: advised, called, guarded, killed ). Discussion Overall, the results of this experim ent are not as expected. Read ing time patterns per measure for each group suggest that all three grou ps are influenced differently by the factors investigated in this experiment. For native speakers, the results suggest that verb bias does not play an immediate role in whether native speaker s posit gaps at the first available gap location, the verb, but that plausibility might. The support fo r this conclusion is the lack of an interaction between verb bias and plausibility in first pass reading, whereas plausibility has a weak effect at the verb. In addition, in measures that reflect revision, native speakers do seem to take verb bias and plausibility into account, suggesting that thes e lexical factors play a role in revision. For advanced speakers, the predicted result of an interaction did not occur in any of the eye-tracking measures, but main effects of plausibili ty and verb type occurred at different critical words in first pass reading time. Regarding interm ediate speakers, effects in first pass reading were found, suggesting that verb bias may be more influential for lower prof iciency speakers. In conjunction with the results from the grammati cality judgment questionnaire, these effects may be due to a gap in L2 knowledge concerning th e verbs tested. Based on the results of the questionnaire, however, it is not clear why intermediate speakers would rely on verb bias more than advanced speakers. One remaining question, which has not been ad dressed, is whether there were main effects of plausibility in total ga ze duration at the fillers ( waiter/plate/waiter/mop see table 5-2 or any

PAGE 134

134 figure), a critical word not reported on above. An effect at this position might indicate that readers spent more time rereading fillers that were implausible at the verb. There were no significant effects of plausibility in total gaze duration for any of the groups, which suggests that readers did not return to reread implausible fillers any more than they did for plausible fillers. Conclusions The research questions investigated by the current experim ent are: Will a natural reading task reveal that plausibility influences L2 filler-gap resolution for both groups of L2 speakers? Will there be evidence that verb bias (transitivity frequency) influences L2 filler-gap resolution? Overall, do L1 and L2 sensitivities to verb bias and plausibility di ffer? Do these differ among L2 proficiency groups? Before attempting to answer these questions it should be acknowledge d that the general hypothesis guiding this experiment was not supporte d. It was expected th at the two factors under examination, verb bias and plausibility, would in teract because both factors would be used by native and advanced readers as they encountered the verb. The hypothesis, as such, does not treat these three research questi ons as having separate answers; yet, the possibility that these factors would not interact allows each factor to be treated as a different question. It is not entirely clear from the results that plausibility influen ces whether L2 speakers posit gaps at the verb. If this were the case, a main effect of plausibility would have been found in first pass reading at the verb. The effect of plausibility in first pass reading time was marginally significant for advanced speakers at the adverb, which was interpreted as spillover effects from the verb, and suggest s that advanced speakers are infl uenced by plausibility slightly later than native speakers. For intermediate speaker s, there was no effect of plausibility in first pass reading, but plausibility did play a role in revision. For inte rmediate speakers, the lack of the use of plausibility may be a result of not co mputing this information in the same time frame

PAGE 135

135 as more advanced speakers while processing comp lex structures. From a processing demands perspective, dependency resolution may be more ta xing for these readers, interfering with their ability to focus on plausibility information. Cruciall y, plausibility of filler is not intrinsically tied to the verb, whereas verb bias is, which intermediate speakers appear to focus on more. In comparing all three groups, advanced speakers, like native speakers, were quicker to use plausibility information during filler-gap resolution. It is clear from the data that ve rb bias played some role in L2 processing, particularly for intermediate speakers. For both groups of L2 speak ers, there were effects of verb bias, albeit at different critical words and in different ey e-tracking measures. First pass reading time demonstrated that advanced learners spent mo re time reading a few words downstream following a transitive bias verb, while intermediate sp eakers appeared to spend more time reading the transitive bias verb itself. This implies th at across proficiency level, these speakers were sensitive to the different argument structure pref erences of these verbs. Advanced speakers appear to have had more difficulty with PP extrac tion after transitive bias verbs, as reflected in total reading time difficulties at the actual gap. For intermediate speakers, on the other hand, it is not entirely clear how these preferences played a role in resolving dependencies. The results of this experiment do more than provide answers to the research questions put forth. They also suggest that L2 readers utili ze the active filler strategy (AFS). In self-paced reading tasks, like that used in experiment 1, which was presented in chapter four, and in Williams (2006), evidence of the AFS was confo unded because effects for conditions containing gaps appeared late, either at or after the gap. There was also the possibility that self-paced reading was either slowing processing or drawing readers atten tion to lexical information and gapped constructions. The current experiment, however, revealed effects of gap filling in first

PAGE 136

136 pass reading, at the verb for intermediate speaker s and at the adverb for advanced speakers. One of the disadvantages of this experiment is that non-gap control sentences were not used, which would have provided stronger support for L2 use of the active filler strategy.

PAGE 137

137 CHAPTER 6 DISCUSSION AND CONCLUSION The experim ents in this disserta tion set out to test whether late -learning L2 speakers of two different proficiency levels were influenced by lexical propertie s associated with verbs during filler-gap resolution. One of the goals of the study was to investigate developmental aspects of the L2 sentence processing mechanism, and a nother goal was to compare results collected through two different psycholinguis tic techniques. The first tw o experiments, presented in chapter four, investigated the use of subcategor ization frame and compared results from selfpaced reading and eye-tracking. The third experiment, presented in chapter five, dealt with L2 use of verb bias during filler-gap resolution. The main hypothesis tested across all experiments in this study was that L2 speakers with lower le vels of proficiency were expected to be less sensitive to verbal properties due to greater processing demands a ssociated with resolving fillergap dependencies. Results from the first tw o experiments show evidence of supporting this hypothesis, whereas results from the third experiment do not. Summary of Results Experiments 1 and 2 The significant result to com e out of experi ment 1, which used the self-paced reading paradigm, was that advanced speakers appear to be influenced by verb subcategorization frame during filler-gap resolution, alt hough less so than native speakers. In comparison to advanced speakers and native speakers, intermediate sp eakers did not demonstrate any influence of subcategorization frame when resolving filler-gap dependencies. It was determined that the intermediate speakers performance was not due to a lack of knowledge concerning the subcategorization frames associated with the English verbs tested because the Korean verb equivalents shared the same frames and b ecause both groups of L2 speakers rejected put -type

PAGE 138

138 verbs without PPs more often than push -type verbs without PPs in an off-line task. When the results from experiment 1, using self-paced readi ng, are compared to results from experiment 2, it appears that the results for na tive and advanced speakers sligh tly differ depending on task, but that the overall results were the same. Experiment 3 Results from this experim ent suggest that all three groups are infl uenced differently by plausibility and verb bias. In general, the pa tterns that emerged for the different eye-tracking measures are more difficult to interpret than thos e in experiments 1 and 2. The expected results for native speakers were not found in that verb bi as and filler plausibil ity did not interact. Plausibility did appear to have a weak effect at the verb, and in measures that reflect revision, verb bias and plausibility app eared to be used by native spea kers. Regarding L2 speakers, effects in first pass reading were found for both groups, but on different cr itical words. These results suggest that verb bias may be more infl uential for non-native spea kers in resolving fillergap dependencies. Yet, verb bias did not intera ct with plausibility for either advanced or intermediate speakers. Of the two groups, the intermediate group appears to be influenced more quickly by verb bias; although based on later effects, it is not clear what role verb bias played in filler-gap resolution for this group. The advanced speakers did appear to have more difficulty resolving dependencies between fillers and PPs when the verb has a transitive bias compared to the intermediate speakers. Success in terms of comprehending transitive bias conditions was worse for intermediate speakers. Discussion The m ain hypothesis was not supported by the results from all three experiments. Experiments 1 and 2 supported the finding that in termediate speakers were not sensitive to a verbs subcategorization frame in an on-line ma nner. The differences that emerged for L2

PAGE 139

139 speakers between the two experiments suggest th at the L2 parsing mechanism may develop as proficiency increases in that as lower level processes become mo re automatic, resources can be freed up to focus on lexical properties in more demanding situations. The cognitive perspective of language le arning based on a limited capacity, which was discussed in chapter three, provi des a possible explanation for the above results. Recall that in this model, noticing is central to processing; mean ing that input must be noticed in order for the learning system to be able to use it. At the sa me time, not all noticing is equally valuable. Six factors were discussed as to what contributes to the systems abil ity to notice: frequency of the input; saliency of the input; how adequate instruction is at getting the learner to notice the input; individual differences in aptit ude and dealing with processing demands (working memory); the readiness for the system to notice the input; a nd whether the demands of the task allow for noticing to take place. The key factor in the di scussion is the last, whether demands of the task allow for noticing to take place. What I mean by this is that subcategorization frame is one piece of lexical information available to the processor when resolving these complex dependencies. If processing capacity limits have already been reached because a filled-gap has disrupted resolution of the dependency, the L2 processor cannot attend to, or notice, what contribution subcategorization frame makes toward buildi ng a representation for comprehension. For intermediate speakers, demands appeared to obstr uct their ability to notic e this subtle lexical information, regardless of task. It is important to note that the eye-trackin g data reveal that the first critical word at which all of the L2 speakers begin to use s ubcategorization frame is at the preposition from which the filler was extracted. Therefore, the results across experiments 1 and 2 beg the question of whether self -paced reading leads advanced speakers to use the active filler

PAGE 140

140 strategy, which will be discussed in the context of future research. Experiment 3 provides some support for L2 use of the active filler strategy. In turning to how the third study did not support the main hyp othesis across all studies, it was found that intermediate speakers appear to be driven by verb bias more so than advanced speakers and native speakers. Fi rst, it important to note that none of the groups displayed the effects predicted based on Stowe et al. (1991). The results that did emerge were argued to suggest that intermediate speaker s are unlikely to fill the gap at the verb if the verb has an intransitive bias, but as discusse d in chapter five, this conclusi on is not supported by later effects of plausibility. In any case, the fact that verb bias had an i mmediate effect for intermediate speakers is an interesting finding because they are the only group who demonstrated such a pattern. Although results were marginal, both na tive and advanced speakers showed effects of plausibility before verb bias. Intermediate speakers may not be fully accessing the semantic content of the filler as they read the verb. Instead the evidence s uggests that they may be relying on whether or not the verb preferen tially takes a direct object. Co mpared to the results from the Williams studies (2001, 2006), this interpretation ma kes sense. In the stop-making sense task, plausibility did appear to be us ed in an on-line manner, but in the more naturalistic task, selfpaced reading, there was non-use of plausibility information by poor performers on the memory task. What this means is that intermediate sp eakers may be adhering to a processor that builds shallow structures. The heuristic model of the L2 parser, also re ferred to as the shallo w structures hypothesis (SSH), put forth by Clahsen and Felser (2006b) was discussed in chapter two. The SSH claims that L2 speakers build lexically-based structures during on-line processing, structures that do not contain abstract syntactic elements (Clahsen and Felser 2006a, Clahsen and Felser 2006b,

PAGE 141

141 Ferreira et al. 2002). More explicitly, this m odel involves the applica tion of lexically-based comprehension heuristics and high plausibility and/or strong canonica l form-meaning patterns (pg. 29, Felser and Roberts, 2007). According to the SSH, L2 speakers process filler-gap dependencies on a semantic level by semantically associating fillers with verbs, and the L2 parser does not build structural gaps during on-line comprehension. While the structures in this study cannot answer questions abou t whether or not L2 speakers are building structural gaps, they can provide some information about th e use of canonical form -meaning patterns and lexically-based comprehension heuristics. Th e third experiment suggests that both formmeaning patterns and lexically-based heuristics ma y be at play as L2 readers comprehend these structures on-line, but that they may vary according to proficiency. From a developmental perspective, the two grou ps appear to be usi ng different approaches to processing these types of st ructures. The canonical pattern that the intermediate speakers might be using is a wh -filler-as-direct-obj ect-of-the-verb strategy as long as the verb prefers a direct object. This may override the use of the semantic nature of the filler, as no plausibility effects were found for these speakers. The adva nced speakers, on the other hand, appeared to use plausibility first whereas verb bias came in la te. Plausibility of filler requires that readers not only focus on resolving the gap, but also that they maintain the semantic content of the filler as they do so. Again, this implies that the lack of a plausibility effect for intermediate speakers could be related to a limited capacity system be ing overburdened by the resolution process. They may have forgotten the word and its semantic content, but realize they are still seeking a gap for the filler. Unlike interm ediate speakers, advanced speakers appear to be accessing the semantic content of the filler during on-line processing, wh ich may be related to having more resources available to attend to, or notice, this information. The effect of plausibility is manifested at the

PAGE 142

142 adverb, not the verb. Such a late effect may be a spillover effect, as noted in the discussion in chapter five. If this effect is genuinely spillover, it implies that as long as the filler is plausible, advanced speakers are happy to resolve the dependency at the verb. Advanced speakers, thus, may have enough resources to remember the filler and seek a gap, but not enough to immediately tend to the verb bias. One word of caution ne eds to be acknowledged here related to sample size. There may not be enough power due to small sample sizes (discussed further below); yet, if these results are accurate, they suggest that the two groups are using a different strategy to aid in filler-gap resolution. The fact that both verb bias and plausibility did not interact at the verb for the advanced speakers may not be that surprising given that these effects were also not found for native speakers. Pickering and Traxler (2003) and St aub (2007), discussed in chapter two, also found initial effects of plausibility, but no interactions with verb bias. The tendency of advanced speakers and native speakers to use plausibility of filler over verb bias may be a reflection of parsers that are driven by lexically-based comp rehension heuristics, whereas the intermediate parser is driven by form-meaning patterns. Other th an the marginal effect of plausibility at the verb, native speakers only demons trated difficulty during revision. L1 results suggest that these speakers are processing these dependencies on a shallow basis during firs t pass reading and only returning after the fact to sort out any aspects of the first parse that did not make sense. While the argument outlined above tries to fit the observe d patterns neatly into a model of heuristic processing, it must be acknowledge that this is te ntative. As stated above, the manipulation in both experiments does not test the use of abstract syntactic elements, such as intermediate gaps in long-distance dependencies (dis cussed in chapter two). Furt hermore, the model itself is vague in its descriptions of what is meant by heuristics, which has been one of the main

PAGE 143

143 criticisms of such a model in both the L1 and L2 literature. One of the concerns in the L1 literature has been on when native speakers use heuristic parsing. The research at hand raises a similar question in that if the L2 parser is based only on build ing shallow structures, it appears that the strategies change according to proficiency; yet, the exact nature of the strategies used by L2 speakers is vague. Instead, a processing de mands approach can still serve to explain why differences between parsing strategies exist, ev en if the overall hypothesis of the study was not supported. One final observation to make is based on the poor comprehension performance across both experiments, regardless of proficiency leve l. In general, it appears that L2 speakers encounter difficulty with filler-gap dependencie s that are not filled at the verb, even if differences in reading time patterns emerged. This suggests that L2 speakers may not be familiar with extraction from PPs, or that they have less experience with resolving these types of dependencies. These structures are more infreq uent in English overall, and it is doubtful if nonnative speakers have been formally taught that su ch structures are permissible. The purpose of the experiments in this study was not to test L2 speakers knowledge of these structures, but rather to investigate what happens when they encounter filled-gaps (experiments 1 and 2) and what happens when they read verbs with two different biases and fillers that are either plausible or implausible. The point here is that there is a co nfound present as to whether the L2 speakers lack of knowledge about extraction from PPs contri buted to some of the effects that were found. As mentioned in chapter four, future research sh ould consider more care fully how to incorporate a test of this knowledge into an off-line task. Limitations and Future Research Several lim itations are present in this study. While limitations do not make a positive contribution to the research at ha nd; they do contribute to the possibi lity of future research on the

PAGE 144

144 issues explored in this dissertat ion. The first and most obvious limita tion is the lack of statistical power due to small sample size. As noted in the sections regarding participants, it became increasingly difficult to recruit Korean participan ts. In hindsight, a few solutions come to mind as to how this challenge could ha ve been overcome. One solution is that the same participants could have participated in both st udies, reading different lists. This would have exposed these participants to similar structures, but a manda tory break between meetings may have been enough to prevent any effects of previous exposure. Another option would have been to open up the participant pool to speakers from other L1 backgrounds that are wh -insitu. A fair number of particip ants would likely be achieved if Japanese and Chinese speakers of English were included. As with Korean, there is debate over the status of both of these langua ges as to whether they are insitu languages, or whether the movement that does occur in insitu languages allo ws for a transfer of parsing strategies. The current study did not set out to test L1 transfer. To do so another group of L2 speakers who share an L1 background that is wh -exsitu would have had to been included. In any case, both of these ideas concerning how to get larger sample sizes could easily be implemented in the future. Another limitation of the study is that us ing the cross-methodological technique would have been useful in examining verb bias as well; that is, the e xperimental items from experiments 1 served as fillers for experiment 3, which allowed for experiment 2 and 3 to be collapsed. The experimental items from experi ment 3, however, did not serve as fillers in experiment 1. In the future, allowing experimental items to serve as fillers in experiments using different psycholinguistics methods provides a nice comparison and is relatively simple to implement. The timing of when data was collec ted for the experiments in this study prevented the use of items from experiment 3, as they were not written when data collection on experiment

PAGE 145

145 1 began. The results from this study suggest that L2 processing of comple x structures appears to be quite variable; certainly this is not surprising considering variability in levels of ultimate attainment. Given this knowledge, however, researchers should continue to use multiple psycholinguistic tools to look at di fferent syntactic structures. So far, limitations have been used to inform future research, yet the cross-methodological design of the experiment has revealed that futu re research should focus on the potential use of self-paced reading for instructi on. Eye-tracking appears to suppor t some of the results of the self-paced reading, but effects in natural reading are either non-ex istent or occur later. This suggests that self-paced reading may be a useful tool in developing syntactic processing in latelearners. Self-paced reading requires readers to focus on each word presented. In addition, it slows reading. Taken together, these two factors may help read ers develop syntactic prediction skills and to exercise working memory; thus, enha ncing L2 readers ability to incrementally build representations for comprehension. The stop-maki ng sense paradigm could also be useful in helping L2 readers to develop syntactic reflexes. In the Williams studies (2001, 2006), stop making sense demonstrated that when decisions we re required on a per word basis, L2 readers noticed lexical information that they did not in self-paced reading. A follow-up to the current research would be to develop training program s using the two techniqu es and to see whether either method has an impact on the use of verbal properties during normal reading, such as that recorded during eye-tracking. One last area of future research that could be pursued would be to compare results from a technique more sensitive than eye-tracking. Studi es using Event Related Potentials (ERPs) to measure brain activity during processing have not yet examined filler-gap resolution; instead, ERP studies of L2 processing have been focused on how the processing of violations can inform

PAGE 146

146 models of L2 sentence processing (Hahne 2001, Hahne and Friederici 2001, Ojima 2005, Rossi et al. 2006, Sabourin and Haverkort 2003, Steinhauer et al. 2006, Tokowicz and MacWhinney 2005, Weber-Fox and Neville 1999). Like eye-tracking, ERPs provide a continuous record of language comprehension processes, but with a greater temporal re solution and greater functional significance than both eye-tracking and self-paced reading. As w ith the other two methods, predefined regions of interest are co mpared to a control or other e xperimental conditions. The real advantage this method offers is that because the temporal resolution is on a millisecond to millisecond level, and because it measures brain activ ity, it allows more direct relationships to be drawn between the data and processing theories. Although still not entirely isomorphic in terms of cognitive processes being reflected, ERPs bring researchers one step closer to creating a theoretical explanation of senten ce processing. In addition, this t ype of research could provide insight into the underlying natu re of late second language learners cognitive processing and whether there are ultimate limits on our language processing abilities after puberty. Conclusion Overall, all three stud ies provide answers to questions regarding L2 processing that have been previously uninvestigated. It is evident fr om the discussion of futu re research that much more needs to be done to uncover how late-learn ing L2 speakers process filler-gap dependencies in real time. Individual differences also remain a concern for those interested in this area of research. This study has demonstrated that even when testing late-lea rners who all share the same L1 background special challenges are presen t; thus, future work needs to carefully weigh how L1 background, proficiency, as well as ot her individual differences might manifest themselves in the results. Eventually, info rmation gained on second language performance can be used to support theories of second language acquisition, as well as for the more practical purpose of designing language learning experiences for classroom teaching. The field of second

PAGE 147

147 language acquisition must move in the direction of how processing in second language interacts with the acquisition process and what this implie s for the human ability to attain languages after puberty.

PAGE 148

148 APPENDIX A EXPERIMENT 1 AND 2: EXPERIMENTAL ITEMS 1. A, B, C, D The teacher wondered what the little child p laced the radio on before the fire drill. Did the little child place something on th e radio before the fire drill? No The teacher wondered whether the lit tle child placed the radio on the table before the fire drill. Did the teacher place something on the radio before the fire drill? No The teacher wondered what the little child dropped the radio on during the fire drill. Did the little child drop something on the radio before the fire drill? No The teacher wondered whether the little child dro pped the radio on the table during the fire drill. Did the teacher drop something on the radio before the fire drill? No 2. A, B, C, D The director asked what the crazy artist placed the picture near before the party. Did the crazy artist place the picture near something? Yes The director asked if the crazy artist placed the picture near the fireplace before the party. Did the director ask about the crazy artist placing the picture near something? Yes The director asked what the crazy artist dropped the picture on before the party. Did the crazy artist drop the picture on something? Yes The director asked if the cr azy artist dropped the picture on the cake before the party. Did the director ask about the crazy arti st dropping the picture on something? Yes 3. A, B, C, D My father asked what the young doctor placed the medicine into after the examination. Did the young doctor place something into the medicine? No My father asked whether the young doctor placed the medicine into the trash after the examination. Did my father place something into the medicine? No My father asked what the young doctor dropped the medicine into after the examination. Did the young doctor drop something into the medicine? No My father asked whether the young doctor dropp ed the medicine into the trash after the examination. Did my father drop something into the medicine? No 4. A, B, C, D The clerk saw what the nice woman placed th e clock on after a child knocked it down. Did the nice woman place the clock on something? Yes The clerk saw if the nice woman placed the cl ock on the counter after a child knocked it down. Did the clerk see if the woman pl aced the clock on something? Yes The clerk saw what the nice woman dropped th e clock on after a ch ild knocked it down. Did the nice woman drop the clock on something? Yes The clerk saw if the nice woman dropped the cl ock on the counter afte r a child knocked it down. Did the clerk see if the woman dr opped the clock on something? Yes

PAGE 149

149 5. A, B, C, D The minister saw what the clever mother placed the flowers on before the wedding. Did the clever mother place something on the flowers before the wedding? No The minister saw whether the clever mother pl aced the flowers on the altar before the wedding. Did the minister place something on th e altar before the wedding? No The minister saw what the clever mother dropped the flowers on before the wedding. Did the clever mother drop something on the flowers before the wedding? No The minister saw whether the clever mother dr opped the flowers on the altar before the wedding. Did the minister drop something on the altar before the wedding? No 6. A, B, C, D The crowd saw what the foreign writer placed the coffee on during his speech. Did the foreign writer place the coffee on something? Yes The crowd saw if the foreign writer placed the coffee on the podium during his speech. Did the crowd see if the writer placed the coffee on something? Yes The crowd saw what the foreign writer dropped the coffee on during his speech. Did the foreign writer drop the coffee on something? Yes The crowd saw if the foreign writer dropped the coffee on the podium during his speech. Did the crowd see if the writer dropped the coffee on something? Yes 7. A, B, C, D The students saw what the strange professor placed the plant on during the demonstration. Did the strange professor plac e something on the plant? No The students saw if the strange professor placed the plant on the chair during the demonstration. Did the students place something on the chair? No The students saw what the strange professor dropped the plant on during the demonstration. Did the strange professor drop something on the plant? No The students saw if the strange professo r dropped the plant on the chair during the demonstration. Did the students drop something on the chair? No 8. A, B, C, D The photographer asked what the short touris t placed the camera on during the trip. Did the short tourist place the camera on something during the trip? Yes The photographer asked whether the short tourist pl aced the camera on the seat during the trip. Did the photographer ask about the touris t placing the camera on something? Yes The photographer asked what the short touris t dropped the camera on during the trip. Did the short tourist drop the camera on something during the trip? Yes The photographer asked whether the short tourist dropped the camera on the seat during the trip. Did the photographer ask about the tourist dropping the camera on something? Yes 9. A, B, C, D The nanny knew what the mean father pl aced the keys under before leaving. Did the mean father place something under the keys? No The nanny knew if the mean father placed the keys under the doormat before leaving. Did the nanny place something under the keys? No

PAGE 150

150 The nanny knew what the mean father dropped the keys under before leaving. Did the mean father drop something under the keys? No The nanny knew if the mean father dropped the keys under the doormat before leaving. Did the nanny drop something under the keys? No 10. A, B, C, D The mechanic knew what the lazy driver placed the tools into before driving off. Did the lazy driver place the tools in to something before driving off? Yes The mechanic knew if the lazy driver placed the tools into the trunk before driving off. Did the mechanic know if the driver placed the tools into something? Yes The mechanic knew what the lazy driver dropped the tools into before driving off. Did the lazy driver drop the tools into something before driving off? Yes The mechanic knew if the lazy driver dropped the tools into the trunk before driving off. Did the mechanic know if the driver dropped the tools into something? Yes 11. A, B, C, D The reporter knew what the new president placed the file under during the coup. Did the new president place something under the file? No The reporter knew if the new president placed the file under the car pet during the coup. Did the reporter know if the new president placed something under the file? No The reporter knew what the new president dropped the file under during the coup. Did the new president drop something under the file? No The reporter knew if the new president droppe d the file under the ca rpet during the coup. Did the reporter know if the new presid ent dropped something under the file? No 12. A, B, C, D The cashier asked what the old guard placed the money in during the burglary. Did the old guard place the money in something? Yes The cashier asked whether the old guard pla ced the money in the bag during the burglary. Did the cashier ask about the old guard placing the money in something? Yes The cashier asked what the old guard dr opped the money in during the burglary. Did the old guard drop the money into something? Yes The cashier asked whether the old guard droppe d the money in the bag during the burglary. Did the cashier ask about the old guard dropping the money in something? Yes 13. A, B, C, D The jury saw what the clever judge put th e evidence behind after the loud interruption. Did the clever judge put the ev idence behind something? Yes The jury saw if the clever judge put the evidence be hind the barrier after the loud interruption. Did the jury see if the judge put the evidence behind something? Yes The jury saw what the clever judge pushed the evidence behind afte r the loud interruption. Did the clever judge push the evidence behind something? Yes The jury saw if the clever judge pushed th e evidence behind the ba rrier after the loud interruption. Did the jury see if th e judge pushed the evidence behind something? Yes

PAGE 151

151 14. A, B, C, D The director wondered what the bad actor put the script unde r after rehearsal. Did the bad actor put something under the script after rehearsal? No The director wondered whether th e bad actor put the script unde r the costume af ter rehearsal. Did the director put something under the script after rehearsal? No The director wondered what the bad actor pushed the script under after rehearsal. Did the bad actor push something under the script after rehearsal? No The director wondered whether the bad actor pushed the script unde r the costume af ter rehearsal. Did the director push something under the script after rehearsal? No 15. A, B, C, D The lawyer saw what the new secretar y put the book under before going home. Did the new secretary put the book under something before going home? Yes The lawyer saw if the new secretary put the book under the desk before going home. Did the lawyer see if the secret ary put the book under something? Yes The lawyer saw what the new secretary pushed the book under before going home. Did the new secretary push the book unde r something before going home? Yes The lawyer saw if the new secretary pushe d the book under the desk before going home. Did the lawyer see if the secretary pushed the book under something? Yes 16. A, B, C, D The attendant asked what the busy pilot put the newspaper in before boarding the plane. Did the busy pilot put something in the newspaper before boarding the plane? No The attendant asked whether the busy pilot put the newspaper in the trash before boarding the plane. Did the attendant put something in the trash before boarding the plane? No The attendant asked what the bu sy pilot pushed the newspaper into before boarding the plane. Did the busy pilot push something into the newspaper before board ing the plane? No The attendant asked whether the busy pilot pushed the newspaper into the trash before boarding the plane. Did the attendant push something into th e trash before boarding the plane? No 17. A, B, C, D The students knew what the strong teacher put the desk behind before the school play. Did the strong teacher put the desk behind something? Yes The students knew if the strong teacher put the de sk behind the curtain before the school play. Did the students know about the teacher putting the desk behind something? Yes The students knew what the strong teacher pushe d the desk behind before the school play. Did the strong teacher push the desk behind something? Yes The students knew if the strong teacher pushed th e desk behind the curtain before the school play. Did the students know about the teacher pushing the de sk behind something? Yes 18. A, B, C, D The agent asked what the foreign officer put the documents into while leaving the country. Did the foreign officer put something into the documents? No

PAGE 152

152 The agent asked if the foreign officer put the documents into the suitcase while leaving the country. Did the agent put something into the documents? No The agent asked what the foreign officer pushed the documents into while leaving the country. Did the foreign officer push something into the documents? No The agent asked if the foreign officer pushed th e documents into the suitcase while leaving the country. Did the agent push something into the documents? No 19. A, B, C, D The dealer saw what the old player put the cards under during the poker game. Did the old player put the cards under something? Yes The dealer saw whether the ol d player put the cards under th e napkin during the poker game. Did the dealer see if the old player put the cards under something? Yes The dealer saw what the old player push ed the cards under during the poker game. Did the old player push the cards under something? Yes The dealer saw whether the old player pushed th e cards under the napkin during the poker game. Did the dealer see if the old player pushed the cards under something? Yes 20. A, B, C, D The butler knew what the quiet man put the hat into upon entering. Did the quiet man put something into the hat upon entering? No The butler knew whether the quiet ma n put the hat into a bag upon entering. Did the butler put something into the hat upon entering? No The butler knew what the quiet man pushed the hat into upon entering. Did the quiet man pushed something into the hat upon entering? No The butler knew whether the quiet man pushed the hat into a bag upon entering. Did the butler push something into the hat upon entering? No 21. A, B, C, D The driver saw what the kind policeman put the ticket into afte r issuing a warning. Did the kind policeman put the ticket into something? Yes The driver saw if the kind policeman put the ticket into the garbage after issuing a warning. Did the driver see if the policeman put the ticket into something? Yes The driver saw what the kind policeman pushe d the ticket into after issuing a warning. Did the kind policeman push the ticket into something? Yes The driver saw if the kind policeman pushed the ticket into the garbage after issuing a warning. Did the driver see if the policeman pushed the ticket into something? Yes 22. A, B, C, D The babysitter wondered what the little girl put the toys under while cleaning her room. Did the little girl put so mething under the toys? No The babysitter wondered if the li ttle girl put the toys under he r bed while cleaning her room. Did the babysitter wonder if the gi rl put something under the toys? No The babysitter wondered what the little girl pus hed the toys under while cleaning her room. Did the little girl push so mething under the toys? No

PAGE 153

153 The babysitter wondered if the little girl pushe d the toys under her bed while cleaning her room. Did the babysitter wonder if the gi rl pushed something under the toys? No 23. A, B, C, D The judge asked what the angry lawyer put the report into dur ing the interrogation. Did the angry lawyer put the report into something? Yes The judge asked if the angry lawyer put the report into the drawer during the interrogation. Did the judge ask about the lawyer pu tting the report into something? Yes The judge asked what the angry lawyer pushed the report into duri ng the interrogation. Did the angry lawyer push the report into something? Yes The judge asked if the angry lawyer pushed th e report into the drawer during the interrogation. Did the judge ask about the lawyer pu shing the report into something? Yes 24. A, B, C, D The teacher asked what the crazy teenager put the magazine behind before the inspection. Did the crazy teenager put something behind the magazine? No The teacher asked whether the crazy teenager put the magazine behind the water fountain before the inspection. Did the teacher put something behind the magazine? No The teacher asked what the crazy teenager pus hed the magazine behind before the inspection. Did the crazy teenager push something behind the magazine? No The teacher asked whether the crazy teenager pushed the magazine behind the water fountain before the inspection. Did the teacher push something behind the magazine? No 25. A, B, C, D The manager knew what the young engineer stuc k the machine into during the inspection. Did the young engineer stick some thing into the machine? No The manager knew whether the young engineer stuck the machine into the closet during the inspection. Did the manager know if the engineer st uck something into the machine? No The manager knew what the young engineer lif ted the machine onto during the inspection. Did the young engineer lift something onto the machine? No The manager knew whether the young engineer li fted the machine onto the counter during the inspection. Did the manager know if the engineer lifted something onto the machine? No 26. A, B, C, D The man wondered what the nice waiter stuck the tray onto wh ile serving the food. Did the nice waiter stick the tray onto something while serving the food? Yes The man wondered if the nice wa iter stuck the tray onto the bar while serving the food. Did the man wonder if the waiter stuck the tray onto something while serving the food?Yes The man wondered what the nice waiter lif ted the tray onto while serving the food. Did the nice waiter lift the tray onto something while serving the food? Yes The man wondered if the nice waiter lifted th e tray onto the bar while serving the food. Did the man wonder if the waiter lifted the tray onto something while serving the food?Yes

PAGE 154

154 27. A, B, C, D The customer saw what the kind manager stuck th e package into while packing the groceries. Did the kind manager stick something into the package? No The customer saw whether the kind manager stuck the package into the cart while packing the groceries. Did the customer stick something into the package? No The customer saw what the kind manager lifted the package into while packing the groceries. Did the kind manager lift some thing into the package? No The customer saw whether the kind manager lifted the package into the cart while packing the groceries. Did the customer lift something into the package? No 28. A, B, C, D The child wondered what the mean dentist stuck the tooth into after pulling it out. Did the mean dentist stick the tooth into something? Yes The child wondered if the mean dentist stuck th e tooth into a containe r after pulling it out. Did the child wonder whether the mean dentist stuck the tooth into something? Yes The child wondered what the mean dentist lifted the tooth into after pulling it out. Did the mean dentist lift the tooth into something? Yes The child wondered if the mean dentist lifted th e tooth into a containe r after pulling it out. Did the child wonder whether the mean dentis t lifted the tooth into something? Yes 29. A, B, C, D The janitor wondered what the lazy student st uck the box into while emptying the dorm room. Did the lazy student stick something into the box while emptying the dorm room? No The janitor wondered whether the lazy student st uck the box into the cabinet while emptying the dorm room. Did the janitor stick something into the box while emptying the dorm room? No The janitor wondered what the lazy student lif ted the box into while emptying the dorm room. Did the lazy student lift something into the box while emptying the dorm room? No The janitor wondered whether the lazy student li fted the box into the cabinet while emptying the dorm room. Did the janitor lift something into th e box while emptying the dorm room? No 30. A, B, C, D The customer wondered what the strange empl oyee stuck the lamp in to after it broke. Did the strange employee stick the lamp into something? Yes The customer wondered whether the strange empl oyee stuck the lamp into the dumpster after it broke. Did the customer wonder about the employee sticking the lamp into something? Yes The customer wondered what the strange empl oyee lifted the lamp in to after it broke. Did the strange employee lift the lamp into something? Yes The customer wondered whether the strange employ ee lifted the lamp into the dumpster after it broke. Did the customer wonder about the employee lifting the lamp into something? Yes

PAGE 155

155 31. A, B, C, D The librarian knew what the short bo y stuck the bag in after class. Did the short boy stick something in the bag? No The librarian knew whether the short boy stuc k the bag in the locker after class. Did the librarian stick something in the bag? No The librarian knew what the short boy lifted the bag into after class. Did the short boy lift something into the bag? No The librarian knew whether the short boy lifted the bag into the locker after class. Did the librarian lift something into the bag? No 32. A, B, C, D The coach saw what the strong athlete st uck the equipment into after the game. Did the strong athlete stick the equipment into something? Yes The coach saw if the strong athlete stuck th e equipment into the van after the game. Did the coach see if the strong athlete st uck the equipment into something? Yes The coach saw what the strong athlete lif ted the equipment into after the game. Did the strong athlete lift the equipment into something? Yes The coach saw if the strong athlete lifted the equipment into the van after the game. Did the coach see if the strong athlete lif ted the equipment into something? Yes 33. A, B, C, D The manager knew what the angry musician stuc k the guitar in before leaving the stage. Did the angry musician stick something in the guitar? No The manager knew if the angry musician stuck the guitar in the case before leaving the stage. Did the manager stick something in the guitar? No The manager knew what the angry musician lifte d the guitar into before leaving the stage. Did the angry musician lift so mething into the guitar? No The manager knew if the angry musician lifted th e guitar into the case before leaving the stage. Did the manager lift something into the guitar? No 34. A, B, C, D My mother saw what the busy worker stuck the table into during the move. Did the busy worker stick the table into something during the move? Yes My mother saw whether the busy worker stuc k the table into the truck during the move. Did my mother see if the worker stuck th e table into something during the move? Yes My mother saw what the busy worker lifted the table ont o during the move. Did the busy worker lift the table onto something during the move? Yes My mother saw whether the busy worker lifted the table onto the truck during the move. Did my mother see if the worker lifted the table onto something during the move? Yes 35. A, B, C, D The ranger saw what the quiet hunter st uck the gun into after cleaning it. Did the quiet hunter stick something into the gun? No The ranger saw if the quiet hunter stuck th e gun into the cabinet after cleaning it. Did the ranger stick something into the gun? No

PAGE 156

156 The ranger saw what the quiet hunter lif ted the gun into after cleaning it. Did the quiet hunter lift something into the gun? No The ranger saw if the quiet hunter lifted the gun into the cabinet after cleaning it. Did the ranger lift something into the gun? No 36. A, B, C, D The boy saw what the new maid stuck the blanket into while making the bed. Did the new maid stick the blanket into something? Yes The boy saw if the new maid stuck the blanke t into the toy chest while making the bed. Did the boy see if the new maid stuc k the blanket into something? Yes The boy saw what the new maid lifted th e blanket into while making the bed. Did the new maid lift the blanket into something? Yes The boy saw if the new maid lifted the blanke t into the toy chest while making the bed. Did the boy see if the new maid lifte d the blanket into something? Yes

PAGE 157

157 APPENDIX B DISTRACTER ITEMS 1. Joe asked what the talented dancer st udied on the announcem ent board after class. Did the talented dancer st udy the announcement board? Yes 2. Kate wondered what the lazy realtor found in the cabinet of the empty condo. Did the realtor wonder if Kate looked in the cabinet at something? No 3. My sister asked what my advisor said to our mother after the disagreement. Did my sister ask our moth er about my advisor? Yes 4. The patient officer asked what the angry messenge r yelled at the driver during the traffic jam. Did the officer yell during the traffic jam? No 5. The secretary saw what the busy executive bought from the representative at the conference. Did the busy executive buy something from the representative? Yes 6. The nosy clerk wondered what the shopper bo ught from the deli besides a pound of ham. Did the nosy clerk wonder if th e shopper bought something? Yes 7. The producer knew what the shy singer hid in her pocket before performing on stage. Did the shy producer know what the singer hid under her jacket? No 8. The distrustful president saw what the busy banker accepted from the customer during the transaction. Did the banker watch the president accept something? No 9. Calvin saw what the sneaky burglar stol e from the bookstore during the grand opening. Did the sneaky burglar steal so mething from the bookstore? Yes 10. The staff knew what the talented architec t planned for the unveil ing party of the new skyscraper. Did the staff know about the unveiling of the school? No 11. The angry customer wondered what the lazy butcher cut with the rusty knife. Did the angry customer cut the butcher with a rusty knife? No 12. The waiter wondered what the angry play er lost in the machine at the casino. Did the angry player lo se at the casino? Yes 13. The press asked what the clever spy found near the gun at the crime scene. Did the clever spy find so mething near the gun? Yes 14. The chef asked what the clever cook saved in the bowl next to the milk. Did the chef ask about the cook saving milk in something? No 15. The head of medicine knew what the skilled surgeon cut with the kn ife during the careful procedure. Did the head of medicine know what the su rgeon cut during the careful procedure? Yes 16. The captain wondered what the tired fisherman found on the deck after the storm. Did the captain find something on the deck? No 17. Joshua knew what the brave deputy review ed from the murder case after the arrest. Did Joshua know the deputy reviewed something from the case? Yes 18. The editor knew what the young reporte r left at the scene of the crime. Did the editor ask about the young reporter leaving the scene of the crime? No 19. The silly clown saw how the young acrobat amazed the crowd with her brave performance. Did the silly acrobat see the clown perform? No 20. The patient librarian asked us how the clever storyteller distracted the children during the ceremony. Did the librarian know who distracted the children during the ceremony? Yes

PAGE 158

158 21. The bank auditor wondered why the old publisher planned to ask for a loan. Did the bank auditor know the ol d publisher wanted a loan? Yes 22. The careless girl knew that the busy ma id had cleaned the stain from the counter. Did the girl see the stain on the counter? Yes 23. The serious diplomat knew why the reporter to ld the soldiers about the press conference concerning the war. Did the serious reporter know some thing about the soldiers? No 24. The tourists knew that the photographer had found numerous good spots for taking pictures of the canyon. Did the photographer want to take pictures of the canyon? Yes 25. Steve asked how the little raft reached the shore at th e same time that the pine logs floated past the bridge. Did Steve ask about the pine logs floating past something? Yes 26. The carpenter asked when th e foreign designer delivered th e new materials for the next phase. Did the foreign designer ask about the carpenter delivering something? No 27. The economist asked why the clever banker needed advice a bout investing in the market. Did the economist need advice about somethi ng related to investing in the market? No 28. The eager clerk asked whether the short witness described th e burglar as tall. Did the short witness see the burglar? Yes 29. The cook saw when the new hostess went into the cooler and drank a bottle of wine. Did the new hostess drink some wine in the cooler? Yes 30. The busy housewife knew when the new gardener went to shovel the s now off the sidewalk. Did the new gardener shovel snow off the driveway? No 31. The detective wondered why the quiet supervisor inspected the fingerprint with a magnifying glass. Did the detective wonder why the quiet supervisor looked at the fingerprint with a magnifying glass? Yes 32. The nice lady asked why the helpful vendor closed the store earl y during the holiday shopping season. Did the vendor close the shop early during the holiday season? Yes 33. The teachers saw when the tired principal anxiously paced around the room filled with parents. Did the tired principal pa ce around the teachers? No 34. The passenger asked me when the mean conducto r had collected all of the tickets for first class. Did the passenger ask me a bout first class tickets? Yes 35. Mary wondered why the smart student went to the principal's office during class. Did Mary know when the smart student went to the principal's office? Yes 36. The mailman saw when the strange man covered the stray cat with a blanket. Did the mailman know the man used a blanket? Yes 37. The nutritionist told the good nurse why she whispered a secret to the doctor about the patient. Did the nutritionist whisper something to the doctor? Yes

PAGE 159

159 38. The audience gossiped about the angry golfer lo sing the match after the storm disrupted the tournament. Did something cause the golfer to lose the match? Yes 39. The lazy painters laughed when the mean owners had changed their minds about the colors. Did the mean painters know when the lazy owners picked new colors? No 40. The architect criticized how the new assist ant drew the designs for the new project. Did the architect finish the draw ings for the new project? No 41. Tina scolded her children for not being at the bus stop be fore the school bus arrived. Did Tina's children make it to the bus stop late? Yes 42. Sarah wanted to know whether her angry boyfriend went to work after they fought. Did Sarah's angry boyfriend wonder about her going to work? No 43. The staff suspected that the helpful office ma nager planned the party with a limited budget. Did the helpful office mana ger plan something? Yes 44. Bill reported that the speaker planned to read several acknowledgements about his work before accepting the prize. Did Bill know about the speaker planning so mething after accepting the prize? No 45. The serious ambassador checked whether th e talented spy completed the dangerous assignment on time. Did the talented ambassador complete the dangerous assignment? No 46. The angry expert scolded the doctor for let ting the young patient stop the treatment before a complete recovery. Did the angry patient fully recover? No 47. The nun inquired as to whether the nice prie st planned his speech for the reunion while on the airplane. Did the nun wonder if the priest planned the reunion while on the plane? No 48. The nice crossing guard told us to help the old woman trying to cross the street during rush hour. Did the old woman need assi stance during rush hour? Yes 49. The skilled ballerina suspected that the mean actors laughed when she ran off the stage frightened. Did the skilled actors get frightened by something? No 50. Mark laughed when the happy drunk tried to purch ase an expensive bottle of wine with a one dollar bill. Did Mark know that the happy drunk tried to buy wine? Yes 51. The boss suspected that the sneaky employees planned a surprise party for his promotion during the coffee break. Did the sneaky boss plan a coffee break? No 52. The concerned policeman made sure that th e baby was admitted to the hospital after the accident. Did the policeman worry about the baby after the accident? Yes 53. The careful nanny wanted to know how the s hort toddler crawled on top of the table. Did the short toddler on the tabl e concern the careful nanny? Yes 54. The nervous crowd discussed how the bad shoot er got away before the police arrived on the scene. Did the bad shooter have a crowd watching him? Yes

PAGE 160

160 55. The famous drummer reported in the interview that the shy guitarist played better when standing in back. Did the famous guitarist lik e to stand in front? No 56. The wild fans discussed why the famous musici an threw bottles into the audience during the concert. Did the audience throw something at the famous musician? No 57. The janitor helped the new librarian report the extent of the damage after a shelf fell in the library. Did the janitor know something fell in the library? Yes 58. The Smiths laughed when the new neighbors had filed a complaint with the police about their noisy dog. Did the Smiths ask the neighbors about the police? No 59. The nervous passengers wanted to know whether the pilot needed a special license for flying international flights. Did the nervous pilot ask the passengers something? No 60. Phil watched when the little dog crossed the st reet causing the cat to jump over the fence behind our house. Did Phil see something go behind our house? Yes 61. Nathan told his father that a few mean ki ds rode by on bikes shouting threats at the younger kids. Did the mean kids see Nathan shout threats? No 62. Ann criticized how her strict husband saved receipts and took notes for every expenditure. Did Ann's husband know she saved something? No 63. The talented player checked th at the old coach locked the door of the stadium after practice. Did the player check if the old coach lock the stadium? Yes 64. The young couple inquired as to how the kind r ealtor could afford to spend so much time with them. Did the kind couple spend a lot time with the young realtor? No 65. The waiters gossiped about how the busy bart ender maintained the bar so well during the rush. Did the busy waiters maintain the bar during the rush? No 66. The inspector reported that th e young bus driver passe d the street of th e final destination without knowing it. Did the inspector watch the young bus driver? Yes 67. My supervisor scolded me when the new employee stole some office supplies from his desk. Did the new employee see me steal something? No 68. The sergeant wanted to know whether the mean soldier found any valuables in the abandoned museum. Did the mean sergeant find any valuables in the abandoned museum? No 69. The skilled leader criticized the crazy hi ker for demonstrating the rescue technique improperly for the campers. Did the skilled leader dist rust the crazy hiker? Yes 70. The strange dentist inquired as to why the new patient went to the pharmacy for pain medication after the procedure. Did the strange dentist do a procedure on the patient? Yes

PAGE 161

161 71. My nice mother checked whether the plumbe r needed a deposit before fixing the leak. Did nice plumber fix the leak? No 72. The crazy fan helped the famous athlet e buy a new sports car after winning the championship. Did the crazy athlete help the fan buy a sports car? No

PAGE 162

162 APPENDIX C SUBCATEGORIZATION FRAME S ELF-COMPLETION QUESTIONNAIRE Hi, The goal of the present questionnaire study is to investigate how nativ e speakers of English com plete certain sentence fragments. We would really appreciate your help!! On the following pages, you will find sentence fragments such as: The tired worker I would like to ask you to complete each fragment with the first thing that crosses your mind. Please do not think too long about it; what counts is the first th ing. You may use the same words more than once, but please do not look back at previous sentences. Please note that your particip ation is voluntary, and that you can stop whenever you wish. Before you start, please complete your name, which languages (including English) you speak, and at which age you started learning them. We will separate this sheet from the rest of this questionnaire before we process the data. Ho wever if you are uncomfortable completing the requested info, you can either use a separate sheet of paper and give this to the experimenter, or leave these lines blank. Name ___________________________ Languages spoken: language when l earned (e.g. from birth, high school) __________ __________ __________ __________ __________ __________ Thank you very much for your help!!! Andrea Dallas Program in Linguistics University of Florida

PAGE 163

163 The careful engineer poured The smart shopper chose The cheap customer positioned The generous executive bought The bad student placed The old man tacked The fit athlete lost The handsome bartender spilled The new manager handed The angry senator stuck The nervous lawyer lifted The busy realtor confirmed The young mother moved The capable teacher situated The tired farmer lowered The sneaky burglar stole

PAGE 164

164 The quiet prisoner strapped The nice host accepted Her strong husband arranged The tired fisherman mounted The careful butcher grasped The helpful driver dumped The strong soldier loaded The beautiful woman found The happy grandfather saved The crazy director dropped Her good friend laid The skilled surgeon hit The young technician dipped The foreign officer lodged The practical plumber raised The busy employee hurled

PAGE 165

165 The clever spy revealed Her messy daughter threw The angry messenger attacked The famous actor draped The tall king hung The new judge glued The shy singer hid The busy worker set The angry boyfriend drove The brave deputy reviewed The helpful professor copied The little child put The strict father returned His tidy uncle pushed Her happy son rested The talented dancer studied

PAGE 166

166

PAGE 167

167 APPENDIX D SUBCATEGORIZATION FRAME GRAMMATICALITY JUDGMENT TASK FOR NORMING Hi, The goal of the present questionn aire study is to investigate ho w native speakers judge certain English sentences. W e would re ally appreciate your help!! Instructions : On the following pages, you will read sentences as: On Sunday, the tired workers relaxed. Some are grammatically incorrect while others are grammatically co rrect. Please judge the sentences based on how grammatical or ungramma tical you feel they are, using the following category: 1 Completely ungrammatical : it's really bad, I would never say it or write it 2 Ungrammatical : it's still grammatica lly incorrect but it doesn't sound as bad 3 I don't know : sorry, I can't tell whether it is grammatically correct or incorrect 4 Grammatical : I would accept it as a grammatical sentence 5 Completely grammatical : it is perfectly correct, I would say it exactly like that Please ask the proctor if you ha ve any questions about this. Before you start, please complete your name, which languages (including English) you speak, and at which age you started learning them. We will separate this sheet from the rest of this questionnaire before we process the data. Ho wever if you are uncomfortable completing the requested info, you can either use a separate sheet of paper and give this to the experimenter, or leave these lines blank. Name ___________________________ Languages spoken: language when l earned (e.g. from birth, high school) __________ __________ __________ __________ __________ __________ Please note that your particip ation is voluntary, and that you can stop whenever you wish. Thank you very much for your help!!! Andrea Dallas Program in Linguistics

PAGE 168

168 Grammaticality Judgment Questionnaire 1 Completely Ungrammatical : It is really bad; I w ould never say it or write it. 2 Ungrammatical : It is still ungrammatical, but it doesnt sound as bad. 3 I dont know : I cant tell whether it is grammatical or ungrammatical. 4 Grammatical : I would accept it as grammatical. 5 Completely Grammatical : It is completely grammatical; I w ould say it or write it like that. The confused experts studied the issue for weeks before finding a solution. 1 2 3 4 5 BAD GOOD The manager saw the cheap customer slowly pushing the cart. 1 2 3 4 5 BAD GOOD As a single mother, Tina struggled every month. 1 2 3 4 5 BAD GOOD When we were camping, we drifted the raft downstream. 1 2 3 4 5 BAD GOOD The crowd wondered if the musician chose for the set. 1 2 3 4 5 BAD GOOD The attentive teacher saw the student place the crayon. 1 2 3 4 5 BAD GOOD The ballerina hid her sh oes with the table. 1 2 3 4 5 BAD GOOD The long branches of the willo w tree swayed in the wind. 1 2 3 4 5 BAD GOOD The clumsy boy spilled the milk. 1 2 3 4 5 BAD GOOD The friendly cashier saw the woman stick the cereal. 1 2 3 4 5 BAD GOOD The janitor bought a new broom. 1 2 3 4 5 BAD GOOD The nurse carelessly yelled words at the patient. 1 2 3 4 5 BAD GOOD

PAGE 169

169 1 Completely Ungrammatical : It is really bad; I w ould never say it or write it. 2 Ungrammatical : It is still ungrammatical, but it doesnt sound as bad. 3 I dont know : I cant tell whether it is grammatical or ungrammatical. 4 Grammatical : I would accept it as grammatical. 5 Completely Grammatical : It is completely grammatical; I w ould say it or write it like that. The lazy clerk strapped the books. 1 2 3 4 5 BAD GOOD The medical students knew that th e famous scientist confirmed. 1 2 3 4 5 BAD GOOD When the alarm went off, the anxiou s technician dropped the test tube. 1 2 3 4 5 BAD GOOD The politician talked for hours. 1 2 3 4 5 BAD GOOD At the party, we danced the floor again and again. 1 2 3 4 5 BAD GOOD The lazy dentist reviewed th e technique after making a mist ake with the procedure. 1 2 3 4 5 BAD GOOD When he entered the doctors office, Harold put his hat. 1 2 3 4 5 BAD GOOD The children marveled at how the chimpan zees jumped from branch to branch. 1 2 3 4 5 BAD GOOD The fantastic magician disappeared the rabbit right before our eyes. 1 2 3 4 5 BAD GOOD When ordered, the soldier lifted his gun. 1 2 3 4 5 BAD GOOD Phil lost his dog at the park while he was watching the baseball game. 1 2 3 4 5 BAD GOOD When the warden was doing rounds, th e unruly prisoner attacked wildly. 1 2 3 4 5 BAD GOOD I saw the baby fell the bottle during breakfast. 1 2 3 4 5 BAD GOOD

PAGE 170

170 1 Completely Ungrammatical : It is really bad; I w ould never say it or write it. 2 Ungrammatical : It is still ungrammatical, but it doesnt sound as bad. 3 I dont know : I cant tell whether it is grammatical or ungrammatical. 4 Grammatical : I would accept it as grammatical. 5 Completely Grammatical : It is completely grammatical; I w ould say it or write it like that. The alley cat found a pile of bones in the gutter. 1 2 3 4 5 BAD GOOD The attentive teacher saw the rude child push. 1 2 3 4 5 BAD GOOD The painters agreed about how bad the colors looked. 1 2 3 4 5 BAD GOOD Ann accepted the reading from her mother. 1 2 3 4 5 BAD GOOD The thoughtful host placed the door stop under the door to prevent it from closing. 1 2 3 4 5 BAD GOOD The nanny stole a piece of expensive jewelry from her employers. 1 2 3 4 5 BAD GOOD The crazy fans cheered loudly for the team. 1 2 3 4 5 BAD GOOD The glamorous event happened with ample time. 1 2 3 4 5 BAD GOOD Every Sunday, the frugal man saved coupons on the paper. 1 2 3 4 5 BAD GOOD After the rush, the cook spilled. 1 2 3 4 5 BAD GOOD The busy director stuck the memo to the bulletin board. 1 2 3 4 5 BAD GOOD The bitter maid dropped. 1 2 3 4 5 BAD GOOD The architect knew when his new assistant grasped the new skyscraper project. 1 2 3 4 5 BAD GOOD

PAGE 171

171 1 Completely Ungrammatical : It is really bad; I w ould never say it or write it. 2 Ungrammatical : It is still ungrammatical, but it doesnt sound as bad. 3 I dont know : I cant tell whether it is grammatical or ungrammatical. 4 Grammatical : I would accept it as grammatical. 5 Completely Grammatical : It is completely grammatical; I w ould say it or write it like that. The police saw shooter hit the cler k on the head with his gun. 1 2 3 4 5 BAD GOOD The surfer strapped his surfboard to the roof of his car. 1 2 3 4 5 BAD GOOD The camp director warned the swimme rs not to hesitate the dive. 1 2 3 4 5 BAD GOOD The lonely dog barked through out the night. 1 2 3 4 5 BAD GOOD Whenever Mary had to work late, her patient husband waited her. 1 2 3 4 5 BAD GOOD The noisy flight attendant saw the passenger in aisle 26 lift. 1 2 3 4 5 BAD GOOD Mark put his coat on the tabl e instead of hanging it up. 1 2 3 4 5 BAD GOOD My mother wondered if my friend returned my sweater 1 2 3 4 5 BAD GOOD At the wedding, there were beautiful flowers floating on the pool. 1 2 3 4 5 BAD GOOD The musician revealed on stage. 1 2 3 4 5 BAD GOOD

PAGE 172

172 APPENDIX E LANGUAGE BACKGROUND QUESTIONNAIRE Language Background Survey Participant ID _________ 1. Have you at any time had a speech, reading or la nguage disorder? YES NO 1b. If so, please describe._________________________________________________________ _____________________________________________________________________________ _____________________________________________________________________________ 3. If decided, what is your major? _________________________________________________ 4. How many years of education have you had (hi gh school, # of years of college, # of years of graduate school, # of years of vocational school, highest degree earned) _____________________________________________________________________________ 5. Have you ever taken a linguistics course? YES NO 6. If you've taken the TOEFL, what was your score? ____________ 7. Mother's Native Language ______________________ 8. Father's Native Language ______________________ # of Years Resided Place of Birth: _________________________________ ______ Places you've lived: _________________________________ ______ _________________________________ ______ _________________________________ ______ _________________________________ ______ _________________________________ ______

PAGE 173

173 Native Language # of years spoken Proficiency* Proficiency* Languages Studied/Spoken # of years spoken age began acquiring age & country of immersion age & place of significant exposure reading writing listening speaking *Please indicate degree of proficiency: basic, intermediate, high, fluent The age & country when you first entered a country where the target language was the primary language spoken. The age & place when you first used the target language on a daily basis as a form of communication with native speakers (this may be the same as age of immersion, or it may not be).

PAGE 174

174Self-rating of Language Skills (this component was adapted from the EUs European Language Portfolio ) Please use the empty grid below and the guidelines on the following page to assess the various sk ills represented in the first column by the icons. means Listening means Reading means spoken interaction means spoken production means writing Directions: Start by reading the guidelines in A1 to assess your listening skills in English. If you can do the tasks describ ed in A1, shade in the box, and go on to A2. Work your way across the grid from A1 to C2 for each skill. Do NOT shade in non-consecutive blocks, e.g. do not shade in A1 and B1, but not A2. The guidelines present increa sing proficiency in each skill, so if you can accomplish the task in B1, this means you can also accomplish A2. Please ask the experi menter if you have any ques tions about this task. Language: English A1 A2 B1 B2 C1 C2 Language: English A1 A2 B1 B2 C1 C2 EXAMPLE

PAGE 175

175 Self-assessment grid A1 A2 B1 B2 C1 C2 Understanding Listening I can understand familiar words and very basic phrases concerning myself, my family and immediate concrete surroundings when people speak slowly and clearly. I can understand phrases and the highest frequency vocabulary related to areas of most immediate personal relevance (e.g. very basic personal and family information, shopping, local area, employment). I can catch the main point in short, clear, simple messages and announcements. I can understand the main points of clear standard speech on familiar matters regularly encountered in work, school, leisure, etc. I can understand the main point of many radio or TV programs on current affairs or topics of personal or professional interest when the delivery is relatively slow and clear. I can understand extended speech and lectures and follow even complex lines of argument provided the topic is reasonably familiar. I can understand most TV news and current affairs programs. I can understand the majority of films in standard dialect. I can understand extended speech even when it is not clearly structured and when relationships are only implied and not signaled explicitly. I can understand television programs and films without too much effort. I have no difficulty in understanding any kind of spoken language, whether live or broadcast, even when delivered at fast native speed, provided I have some time to get familiar with the accent. Understanding Reading I can understand familiar names, words and very simple sentences, for example on notices and posters or in catalogues. I can read very short, simple texts. I can find specific, predictable information in simple everyday material such as advertisements, prospectuses, menus and timetables and I can understand short simple personal letters. I can understand texts that consist mainly of high frequency everyday or job-related language. I can understand the description of events, feelings and wishes in personal letters. I can read articles and reports concerned with contemporary problems in which the writers adopt particular attitudes or viewpoints. I can understand contemporary literary prose. I can understand long and complex factual and literary texts, appreciating distinctions of style. I can understand specialized articles and longer technical instructions, even when they do not relate to my field. I can read with ease virtually all forms of the written language, including abstract, structurally or linguistically complex texts such as manuals, specialized articles and literary works. Speaking Spoken interaction I can interact in a simple way provided the other person is prepared to repeat or rephrase things at a slower rate of speech and help me formulate what Im trying to say. I can ask and answer simple questions in areas of immediate need or on very familiar topics. I can communicate in simple and routine tasks requiring a simple and direct exchange of information on familiar topics and activities. I can handle very short social exchanges, even though I cant usually understand enough to keep the conversation going myself. I can deal with most situations likely to arise whilst traveling in an area where the language is spoken. I can enter unprepared into conversation on topics that are familiar, of personal interest or pertinent to everyday life (e.g. family, hobbies, work, travel and current events). I can interact with a degree of fluency and spontaneity that makes regular interaction with native speakers quite possible. I can take an active part in discussion in familiar contexts, accounting for and sustaining my views. I can express myself fluently and spontaneously without much obvious searching for expressions. I can use language flexibly and effectively for social and professional purposes. I can formulate ideas and opinions with precision and relate my contribution skillfully to those of other speakers. I can take part effortlessly in any conversation or discussion and have a good familiarity with idiomatic expressions and colloquialisms. I can express myself fluently and convey finer shades of meaning precisely. If I do have a problem I can backtrack and restructure around the difficulty so smoothly that other people are hardly aware of it. Speaking Spoken production I can use simple phrases and sentences to describe where I live and people I know. I can use a series of phrases and sentences to describe in simple terms my family and other people, living conditions, my educational background and my present or most recent job. I can connect phrases in a simple way in order to describe experiences and events, my dreams, hopes and ambitions. I can briefly give reasons and explanations for opinions and plans. I can narrate a story or relate the plot of a book or film and describe my reactions. I can present clear, detailed descriptions on a wide range of subjects related to my field of interest. I can explain a viewpoint on a topical issue giving the advantages and disadvantages of various options. I can present clear, detailed descriptions of complex subjects integrating sub-themes, developing particular points and rounding off with an appropriate conclusion. I can present a clear, smoothlyflowing description or argument in a style appropriate to the context and with an effective logical structure which helps the recipient to notice and remember significant points. Writing Writing I can write a short, simple postcard, for example sending holiday greetings. I can fill in forms with personal details, for example entering my name, nationality and address on a hotel registration form. I can write short, simple notes and messages. I can write a very simple personal letter, for example thanking someone for something. I can write simple connected text on topics which are familiar or of personal interest. I can write personal letters describing experiences and impressions. I can write clear, detailed text on a wide range of subjects related to my interests. I can write an essay or report, passing on information or giving reasons in support of or against a particular point of view. I can write letters highlighting the personal significance of events and experiences. I can express myself in clear, well-structured text, expressing points of view at some length. I can write about complex subjects in a letter, an essay or a report, underlining what I consider to be the salient issues. I can select a style appropriate to the reader in mind. I can write clear, smoothlyflowing text in an appropriate style. I can write complex letters, reports or articles which present a case with an effective logical structure which helps the recipient to notice and remember significant points. I can write summaries and reviews of professional or literary works.

PAGE 176

176 APPENDIX F SHIPLEY VOCABULARY TEST On the test below, the first word o f each li ne is printed in CAPITAL LETTERS. Opposite it are four other words. CIRCLE the one wo rd which means the same thing, or most nearly the same thing, as the first word. A sample has been worked out for you. If you don't know, guess. Be sure to circle the one word on each line which means the same thing as the first word. sample: LARGE red big silent wet 1. TALK draw eat speak sleep 2. PERMIT allow sew cut drive 3. PARDON forgive pound divide tell 4. COUCH pin eraser sofa glass 5. REMEMBER swim recall number defy 6. TUMBLE drink dress fall think 7. HIDEOUS silvery tilted young dreadful 8. CORDIAL swift muddy leafy hearty 9. EVIDENT green obv ious skeptical afraid 10. IMPOSTER conductor officer book pretender 11. MERIT deserve distrust fight separate 12. FASCINATE welcome fix stir enchant 13. INDICATE defy excite signify bicker 14. IGNORANT red sharp uninformed precise 15. FORTIFY submer ge strengthen vent deaden 16. RENOWN length head fame loyalty 17. NARRATE yield buy associate tell

PAGE 177

177 18. MASSIVE bright large speedy low 19. HILARITY laughter speed grace malice 20. SMIRCHED stolen pointed remade soiled 21. SQUANDER tease belittle cut waste 22. CAPTION drum ballast heading ape 23. FACILITATE help turn strip bewilder 24. JOCOSE humorous paltry fervid plain 25. APPRISE reduce scr ew inform delight 26. RUE eat lament dominate cure 27. DENIZEN senator i nhabitant fish atom 28. DIVEST dispossess intrude rally pledge 29. AMULET charm orphan dingo pond 30. INEXORABLE untidy inviolate rigid sparse 31. SERRATED dried notched armed blunt 32. LISSOM moldy loose supple convex 33. MOLLIFY mitigate direct pertain abuse 34. PLAGIARIZE intent revo ke maintain misappropriate 35. ORIFICE brush hole building lute 36. QUERULOUS maniacal curious devote complaining 37. PARIAH outcast priest lentil locker 38. ABET waken ensue incite placate 39. TEMERITY rashness timidity desire kindness 40. PRISTINE vain sound first level

PAGE 178

178 APPENDIX G WAIS VOCABULARY TEST I w ant you to briefly define some words. Let's start with __________; can you define _________? For subjects who seem to have poor verbal ability, start with BED. For all others start with WINTER. Give fu ll credit for BED SHIP PENNY if the subject passes (1 or 2 point answer) WINTER BREAKF AST REPAIR FABRIC ASSEMBLE. If any one of these is failed, administer BED SHIP PENNY. If it is difficult to determine whether the subjec t does or does not know the meaning of a word, prompt with Tell me more about it or Explain what you mean. Record a 2, 1, 0 after each word to record whether the subject passe d (2 or 1 points) or failed the word. Scoring: score each word as 2 points for: good synonym, a major use, a definite or primary feature, a general classification, a correct figurativ e use, a good example of an action or causal relation score each word as 1 point for: vague or less pertinent synonym, a minor use, an attribute which is not definitive, an example that merely uses the word, a concrete instance score each word as 0 for: wrong answers, vague or trivial answers. maximum score: 70

PAGE 179

179 WAIS Vocabulary Response Sheet Participant ID ______ ___BED: an article of furniture to sleep on/plot of ground prepared for plants ___SHIP: a vehicle which travels over water/to transport ___PENNY: money ___WINTER: the cold season ___BREAKFAST: the first meal of the day ___REPAIR: to fix or correct something damaged ___FABRIC: woven material ___ASSEMBLE: to bring, put, or come together ___ENORMOUS: exceeding the usual size, number, or degree ___CONCEAL: to hide ___SENTENCE: a group of connected words expressing a complete idea/a judgment imposed by a court ___CONSUME: to use up/to destroy/to engage fully ___REGULATE: to control/to bring order and uniformity ___TERMINATE: to end ___COMMENCE: to begin ___DOMESTIC: pertaining to home or family/a household servant ___TRANQUIL: calm ___PONDER: to think ___DESIGNATE: to pick out/someone chosen for office but not yet installed ___RELUCTANT: resisting or hesitating ___OBSTRUCT: to impede ___SANCTUARY: a place consecrat ed to worship/place of refuge

PAGE 180

180 WAIS Vocabulary Response Sheet contd. Participant ID _______ ___COMPASSION: sorrow or pity aroused by another's distress ___EVASIVE: tending to avoid something by trickery ___REMORSE: feeling of distress or regret caused by a sense of guilt ___PERIMETER: the distance around a figure/a boundary/ outer limits ___GENERATE: to bring something into existence/to cause to be ___MATCHLESS: wit hout equal/peerless ___FORTITUDE: resoluteness, endurance, courage ___TANGIBLE: capable of being touched/real ___PLAGIARIZE: to steal another' s ideas and pass them off as one's own ___OMINOUS: foreboding evil ___ENCUMBER: to impede the motion/to burden/to weigh down ___AUDACIOUS: showing presumptuous daring ___TIRADE: a long emotional speech marked by anger _____ Total score Participant ID ______

PAGE 181

181 APPENDIX H EXPERIMENT 1: GRAMMATICALITY JUDGMENT TAS K POST TESTING The goal of the present questionn aire study is to investigate ho w speakers judge certain English sentences. Instructions : On the following pages, you will read sentences as: On Sunday, the tired workers relaxed. Some of the sentences are grammatically incorr ect, while others are grammatically correct. Please judge the sentences based on how grammatical or ungrammatical you feel they are, using the following categories: 1 Completely ungrammatical : It is really bad; I woul d never say it or write it e.g., *The meeting will take plac e the conference room in. 2 Ungrammatical : It is still ungrammatical, but it doesn't sound as bad. 3 I don't know : I can't tell whether it is grammatical or ungrammatical. 4 Grammatical : I would accept it as a grammatical sentence. 5 Completely grammatical : It is completely grammatical; I w ould say it or write it like this. Please do not ponder excessively when making your deci sions; what counts is your first reaction. If you rate something as 1, 2, or 3, please circle the word or phra se where you think the sentence first begins being ungrammatical; e.g., *The meeting will take place the conference room in. Please ask the experimenter if you have que stions about any of these instructions. Thank you very much for your help!!!

PAGE 182

182 6 Completely Ungrammatical : It is really bad; I w ould never say it or write it. 7 Ungrammatical : It is still ungrammatical, but it doesnt sound as bad. 8 I dont know : I cant tell whether it is grammatical or ungrammatical. 9 Grammatical : I would accept it as grammatical. 10 Completely Grammatical : It is completely grammatical; I w ould say it or write it like that. The lazy dentist told us that the busy student needed to review the technique after making a mistake with the procedure. 1 2 3 4 5 BAD GOOD The old architect discussed his new assist ant grasped the new skyscraper project. 1 2 3 4 5 BAD GOOD The minister wondered if politician saved money before the campaign began. 1 2 3 4 5 BAD GOOD The strict general asked if the soldier lifted his gun. 1 2 3 4 5 BAD GOOD Mary asked that her patient husband waited her. 1 2 3 4 5 BAD GOOD The medical students found the cure that the famous scientist had worked to find for many years. 1 2 3 4 5 BAD GOOD The wild teenagers asked what their strict chaperone planned the party. 1 2 3 4 5 BAD GOOD The supervisor saw what the lazy worker put the money into before taking a break. 1 2 3 4 5 BAD GOOD The tourist suspected that the mean guide planned. 1 2 3 4 5 BAD GOOD The crowd criticized how the shy singer went for the set. 1 2 3 4 5 BAD GOOD

PAGE 183

183 The clever therapist knew when t he angry couple went see the doctor. 1 2 3 4 5 BAD GOOD The nice doctor noticed if Harold put his hat. 1 2 3 4 5 BAD GOOD The lazy man knew what the little child found on the top shelf at the library. 1 2 3 4 5 BAD GOOD Carrie saw what her old uncle dropp ed the tea on after the quick lunch. 1 2 3 4 5 BAD GOOD The boss know if the short employ ee went to the training session. 1 2 3 4 5 BAD GOOD My mother wondered if my friend found my sweater after the winter holiday. 1 2 3 4 5 BAD GOOD The confused scientist asked what the student stuck the wire into be fore checking the power supply. 1 2 3 4 5 BAD GOOD The soldiers saw that the angry prisoner distracted after he was put in his cell. 1 2 3 4 5 BAD GOOD We described to our friends how the angry musician revealed on stage. 1 2 3 4 5 BAD GOOD The crazy salesman ask how the new customer described the cleaning product. 1 2 3 4 5 BAD GOOD The busy manager saw if the cheap customer pushed the cart. 1 2 3 4 5 BAD GOOD

PAGE 184

184 Her brother suspected whether Ann stopped from reading their clever mother. 1 2 3 4 5 BAD GOOD The fantastic magician asked what the assist ant found the rabbit right before our eyes. 1 2 3 4 5 BAD GOOD The brave leader knew what we planned after the canoe trip. 1 2 3 4 5 BAD GOOD The attentive teacher saw if t he lazy student placed the book. 1 2 3 4 5 BAD GOOD The detective asked when the calm farmer found the skull in the field. 1 2 3 4 5 BAD GOOD Phil wanted to ask us why his crazy dog went into the lake. 1 2 3 4 5 BAD GOOD The strong ballerina checked if the other dancer hid her shoes under the table. 1 2 3 4 5 BAD GOOD Our great grandmother wondered if we had st opped at the store before going home. 1 2 3 4 5 BAD GOOD The florist asked what the beautiful woman li fted the centerpiece ont o before the dinner. 1 2 3 4 5 BAD GOOD The nanny scolded the children for planning to jump at the bed. 1 2 3 4 5 BAD GOOD The foreign designer knew if glamorous event planned with ample time. 1 2 3 4 5 BAD GOOD

PAGE 185

185 The coach saw how the champion skier strapped the skis t op of the truck. 1 2 3 4 5 BAD GOOD The agent told us whether we needed passports during entering. 1 2 3 4 5 BAD GOOD The queen wondered what the short king pushed the chest behind before the battle. 1 2 3 4 5 BAD GOOD The busy man gossiped about the nanny st ealing a piece of expensive jewelry. 1 2 3 4 5 BAD GOOD My mother asked what the nice guests found under the tables at the wedding. 1 2 3 4 5 BAD GOOD The clever professor laughed at why the students needed a grade. 1 2 3 4 5 BAD GOOD The entire neighborhood discussed how the new mayor went to the state capitol to ask for a larger budget. 1 2 3 4 5 BAD GOOD The lab teacher saw whether the anxio us technician dropped the test tube. 1 2 3 4 5 BAD GOOD The camp director wondered whether the swimmers played the pool after eating. 1 2 3 4 5 BAD GOOD The calm nurse knew whether the patient needed bandages against the wound. 1 2 3 4 5 BAD GOOD Tammy knew that her old friend stopped therapy after winning the lottery. 1 2 3 4 5 BAD GOOD

PAGE 186

186 The friendly cashier saw whether the old woman stuck the cereal. 1 2 3 4 5 BAD GOOD His serious wife complained that the frugal man saved coupons on the paper every Sunday. 1 2 3 4 5 BAD GOOD I saw what the quiet baby lost while at the table during breakfast after the bath. 1 2 3 4 5 BAD GOOD Mrs. Davis knew what her quiet so n placed the bowl on after breakfast. 1 2 3 4 5 BAD GOOD The mean warden inquired if the unruly prisoner found the other inmate dead. 1 2 3 4 5 BAD GOOD

PAGE 187

187 APPENDIX I EXPERIMENT 2: GRAMMATICALITY JUDGMENT TAS K POST TESTING The goal of this questionnaire is to investigat e how different speakers of English judge certain sentences. On the following pages, you will find sentences such as: The man slept soundly. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical I would like to ask you to rate how grammatical you find the sentence to be on a scale of 1 to 7, with 1 being completely UNgrammatical and 7 being complet ely grammatical. The above example is grammatical, so you might rate it as a 6 or 7. The following example is ungrammatical, so you might rate it as a 1 or 2: Man sleeping. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical Finally, below is an example th at is not completely grammati cal or completely ungrammatical, so you would probably rate it as a 3, 4, or 5. The man imitated accurately. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical When you are deciding your rating, do so with the first thing that crosses your mind. Please do not think too long about it; what counts is the first thing. Before you begin, please ask the expe rimenter if you have any questions.

PAGE 188

188 The frugal couple saved for many years. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The teacher emphasized. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strict judge listened closely. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The crowd cheered the team. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical We anticipating gladly for the weekend. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The editor reviewed before the meeting. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The bank ignored the angry woman's request. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The teenager crashed the car. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The nurse carefully helped the doctor with the patient. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The child dropped the plate on the floor. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The criminal escaped the jail cell. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The librarian read slowly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strong boy lifted the box. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The trainee flew the airplane. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The busy diplomat entered the room unexpectedly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The artist painted. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clerk lifted the cans onto the counter. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 189

189 The small child floated the boat in the bathtub. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical She gotten a bad grade. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The crowd watched the movie. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The team lost. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strict librarian placed the book. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The mailman carefully sealed the envelope. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical Tree swaying in the wind. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The printer hurried the job. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The man killed because he was angry. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The little boy ate pizza with pepperoni. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The army marched the hill towards the battle. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical His wife waited the late bus. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The child heard clearly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The hunter pointed the gun at the deer. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The policeman guarded all night long. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The angry wife pushed the chair. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The professor talked quickly the lecture. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 190

190 The student copied from the board. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The tired worker pushed the cart down the aisle. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The cowboy raced the horse. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The president put the file. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The maid cleaned for the party. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The bus driver hesitated the door. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The nurse rested her legs. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The secretary called about the problem. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The opponents agreed the problem. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The messenger rushed the package. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The busy director put the computer on the desk. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The rich executive stuck the money. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The captain sunk the boat. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The mean chef been given an award. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The chef baked for the festival. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The manager dusted the shelves in the store. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The reporter stuck the camera in the large bag. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 191

191 The rich lady liked the diamond earrings. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The machine maintain daily. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The director stood the camera on the tripod. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The dog attacked viciously. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The file included. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clever vender sold many hotdogs during the concert. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The kind student placed the ruler on the table. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The children asked about the song. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The girl walked the dog. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clown dropped the ball. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The traveler visited for a while. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical She reacted to the accident quickly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical Her mother yelled her son's name. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The counselor advised all day. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 192

192 APPENDIX J EXPERIMENT 3: EXPERIMENTAL ITEMS 1. A, B, C, D The lifeguard asked which match the swimmer lost carelessly in before the storm began. Did the swimmer lose in a match? Yes The lifeguard asked which team the swimmer lo st carelessly against before the storm began. Did the swimmer lose against a team? Yes The lifeguard asked which raft the swimmer fl oated carelessly on before the storm began. Did the swimmer float on a raft? Yes The lifeguard asked which bridge the swimmer floated carelessl y under before the storm began. Did the swimmer float under a bridge? Yes 2. A, B, C, D The captain knew which boat the sailor studied quickly over before th e sailing certification exam. Did the sailor study an exam? No The captain knew which library the sailor studied quickly at before the sailing certification exam. Did the sailor study an exam? No The captain knew which boat the sailor sunk qui ckly under during the sailing certification exam. Did the sailor sink an anchor? No The captain knew which desk the sailor sunk quickly under during the sailing certification exam. Did the sailor sink an anchor? No 3. A, B, C, D The manager wondered which waiter the customer asked rudely about during the lunch rush. Did the manager wonder about the customer? No The manager wondered which plate the customer asked rudely about during the lunch rush. Did the manager wonder about the customer? No The manager wondered which waiter the customer rushed rudely around during the lunch rush. Did the manager wonder about the customer? No The manager wondered which mop the customer rushed rudely around during the lunch rush. Did the manager wonder about the customer? No 4. A, B, C, D The agent saw which enemy the soldier att acked bravely against during the uprising. Did the agent see the solder attack? Yes The agent saw which gun the soldier attack ed bravely with during the uprising. Did the agent see the solder attack? Yes The agent saw which gun the soldier point ed bravely at during the uprising. Did the agent see the solder point? Yes The agent saw which building the soldier pointed bravely at during the uprising. Did the agent see the solder point? Yes

PAGE 193

193 5. A, B, C, D The journalist reported which leader the crowd heard much about during the protest. Did the crowd hear about a leader? Yes The journalist reported whic h store the crowd heard much about during the protest. Did the crowd hear about a store? Yes The journalist reported which leader the cr owd cheered much for behind during the protest. Did the crowd cheer for a leader? Yes The journalist reported which sidewalk the crowd cheered much from during the protest. Did the crowd cheer from a sidewalk? Yes 6. A, B, C, D The teacher found out which answer the studen t copied stupidly from during the test. Did the student copy from the blackboard? No The teacher found out which seat the student copied stupidly from during the test. Did the student copy during lunch time? No The teacher found out which answer the stud ent hurried stupidly over during the test. Did the student hurry to lunch? No The teacher found out which seat the student hurried stupidly to during the test. Did the student hurry to lunch? No 7. A, B, C, D The neighbor wondered which teenager the wo man advised loudly about after the party. Did the neighbor advise a teenager? No The neighbor wondered which accident the woma n advised loudly about after the party. Did the neighbor advise the women? No The neighbor wondered which vehicle the woma n crashed loudly into after the party. Did the neighbor crash? No The neighbor wondered which teenager the wo man crashed loudly into after the party. Did the neighbor crash? No 8. A, B, C, D The naval commander knew which ship the pilo t attacked violently over during the battle. Did the naval commander know about the pilot? Yes The naval commander knew which ocean the pilo t attacked violently over during the battle. Did the naval commander know about the pilot? Yes The naval commander knew which jet the pilo t flew violently over during the battle. Did the naval commander know about the pilot? Yes The naval commander knew which ocean the p ilot flew violently over during the battle. Did the naval commander know about the pilot? Yes 9. A, B, C, D The coach saw which test the athlete review ed carefully for while at the library. Did the athlete see the coach review? No The coach saw which rest the athlete reviewed carefully during in the middle of the game. Did the athlete see the coach review? No The coach saw which leg the athlete rested carefully on in the middle of the game. Did the athlete see the coach rest? No

PAGE 194

194 The coach saw which test the athlete re sted calmly for while in the library. Did the athlete see the coach rest? No 10. A, B, C, D The attorney found out which witn ess the reporter asked unexpect edly about after the murder. Did the attorney find out about a witness? Yes The attorney found out which docum ent the reporter asked unexpect edly about before the trial began. Did the attorney find out about a document? Yes The attorney found out which name the reporter yelled unexpectedly a bout before the trial began. Did the attorney find out what th e reporter yell ed about? Yes The attorney found out which documen t the reporter yelled un expectedly about before the trial began. Did the attorney find out what th e reporter yell ed about? Yes 11. A, B, C, D The journalist reported which fr iend the senator visited cheerfu lly with before the rally. Did the senator visit with a friend? Yes The journalist reported which coat the sena tor visited cheerfully in before the rally. Did the senator visit in a coat? Yes The journalist reported whic h plane the senator flew cheer fully in before the rally. Did the senator fly in a plane? Yes The journalist reported whic h seat the senator flew cheerfully in before the rally. Did the senator fly in a seat? Yes 12. A, B, C, D The writer wondered which book th e editor copied angrily from before getting fired. Did the editor copy into a book? No The writer wondered which pen the editor copied angrily wi th before getting fired. Did the editor copy with a pencil? No The writer wondered which printe r the editor hurried angrily to before getting fired. Did the editor hurry to a bookstore? No The writer wondered which bag the editor hurri ed angrily with before getting fired. Did the editor hurry to a bookstore? No 13. A, B, C, D The coach knew which match the player lost unexpectedly in during the championship game. Did the coach know the player lost? Yes The coach knew which stadium the player lost unexpectedly in during the championship game. Did the coach know the player lost? Yes The coach knew which team the player cheered unexpectedly for during the championship game. Did the coach know the player cheered? Yes The coach knew which towel the player chee red unexpectedly with during the championship game. Did the coach know the player cheered? Yes

PAGE 195

195 14. A, B, C, D The food critic wasn't sure which crust the chef baked recklessly with during the pie competition. Did the food critic bake with a crust? No The food critic wasn't sure which cook the chef baked recklessly with during the pie competition. Did the food critic bake with a cook? No The food critic wasn't sure which cook the chef rushed recklessl y with during the pie competition. Did the food critic rush the chef? No The food critic wasn't sure which window the ch ef rushed recklessly towards during the fire. Did the food critic rush the chef? No 15. A, B, C, D The rancher knew which horse the cowboy sa ved carefully for during the last year. Did the cowboy save for a horse? Yes The rancher knew which truck the cowboy sa ved carefully for during the last year. Did the cowboy save for a truck? Yes The rancher knew which horse the cowboy raced car efully past on the night of the incident. Did the cowboy race past a horse? Yes The rancher knew which hill the cowboy raced car efully past on the night of the incident. Did the cowboy race past a hill? Yes 16. A, B, C, D The host knew which man the maid heard cu riously about before the dinner party. Did the maid hear a man? No The host knew which room the maid heard cu riously about before the dinner party. Did the maid hear a room? No The host knew which flowers the maid stood curiously near before the dinner party. Did the maid stand some flowers? No The host knew which room the maid stood cu riously near before the dinner party. Did the maid stand some flowers? No 17. A, B, C, D The assistant wondered which client the lawyer visited quickly with after trial preparation. Did the lawyer visit with a client? Yes The assistant wondered which car the lawyer visited quickly in after trial preparation. Did the lawyer visit in a car? Yes The assistant wondered which client the lawyer rushed quickly to during trial preparation. Did the lawyer rush to a client? Yes The assistant wondered which room the lawyer rushed quickly into during trial preparation. Did the lawyer rush into a room? Yes 18. A, B, C, D The newscaster reported which player the coach advised rudely about during the final round. Did the coach advise nicely? No

PAGE 196

196 The newscaster reported which diagram the coach advised rudely with during the final round. Did the coach advise nicely? No The newscaster reported which player the coach cheered rudely for during the final round. Did the coach cheer nicely? No The newscaster reported which bench the coach cheered rudely from during the final round. Did the coach cheer nicely? No 19. A, B, C, D The police knew which leader the rioters killed foolishly for during the massacre. Did the police know about the rioters? Yes The police knew which cause the rioters killed foolishly for during the massacre. Did the police know about the rioters? Yes The police knew which building the rioters escaped foolishly into during the massacre. Did the police know about the rioters? Yes The police knew which computer the rioters escaped foolishly with during the massacre. Did the police know about the rioters? Yes 20. A, B, C, D The doctor wondered which bed the nurse cl eaned carefully near before going home. Did the doctor clean a bed? No The doctor wondered which medicine the nurse cleaned carefully near before going home. Did the doctor clean near some medicine? No The doctor wondered which patient the nurse walked carefully past before going home. Did the doctor walk past a patient? No The doctor wondered which bed the nurse walked carefully past before going home. Did the doctor walk past a bed? No 21. A, B, C, D The assistant wasn't sure which actor the di rector called angrily a bout during the party. Did the director call ab out an actor? Yes The assistant wasn't sure which camera the director called angrily about during the party. Did the director call about a camera? Yes The assistant wasn't sure which line the dire ctor yelled angrily a bout during the party. Did the director yell about a line? Yes The assistant wasn't sure which actor the di rector yelled angrily a bout during the party. Did the director yell about an actor? Yes 22. A, B, C, D The babysitter asked which fence the boy painte d carelessly near as the storm approached. Did the boy paint carefully? No The babysitter asked which dog the boy painted carelessly near as the storm approached. Did the boy paint carefully? No The babysitter asked which boat the boy floated carelessly near as the wave approached. Did the boy float carefully? No The babysitter asked which fish the boy floate d carelessly near as the wave approached. Did the boy float carefully? No

PAGE 197

197 23. A, B, C, D The farmer wondered which animal the worker heard often about at the town meeting. Did the farmer wonder about what the worker heard about? Yes The farmer wondered which weed the worker heard often about at the town meeting. Did the farmer wonder about what the worker heard about? Yes The farmer wondered which tractor the worker raced often around before coming in for lunch. Did the farmer wonder about what the worker raced around? Yes The farmer wondered which fence the worker raced often around before coming in for lunch. Did the farmer wonder about what the worker raced around? Yes 24. A, B, C, D The captain wanted to know which prisoners the army attacked carelessly near during the drill. Did the army attack the captain? No The captain wanted to know wh ich night the army attacked carelessly on during the drill. Did the army attack the captain? No The captain wanted to know which prisoners the army marched car elessly around during the drill. Did the army march ar ound the captain? No The captain wanted to know which tank the ar my marched carelessly around during the drill. Did the army march ar ound the captain? No 25. A, B, C, D The principal wondered which book the child read dramatically from while trying out for the play. Did the principal wonder about the child? Yes The principal wondered which stage the child read dramatically on while tyring out for the play. Did the principal wonder about the child? Yes The principal wondered which teacher the child cheered dramatically fo r while trying out for the play. Did the principal wonder about the child? Yes The principal wondered which stage the child cheere d dramatically on while trying out for the play. Did the principal wonder about the child? Yes 26. A, B, C, D The news announced which gun the policeman hear d clearly about during the undercover operation. Did the policeman hear the news? No The news announced which jail the policeman hear d clearly about during the undercover operation. Did the policeman hear the news? No The news announced which gun the policeman pointed clearly to during the undercover operation. Did the policeman point to the news? No The news announced which criminal the policem an pointed clearly to during the undercover operation. Did the policeman point to the news? No 27. A, B, C, D The homeowner wondered which door the painter painted carelessly around before being fired. Did the painter paint carelessly? Yes

PAGE 198

198 The homeowner wondered which company the painte r painted carelessly for before being fired. Did the painter paint carelessly? Yes The homeowner wondered which brush the painter rested carelessly near before being fired. Did the painter rest near a brush? Yes The homeowner wondered which bed the painte r rested carelessly on before being fired. Did the painter rest near a bed? Yes 28. A, B, C, D The musician saw which register the bartende r guarded closely over as the crowd got rowdy. Did the bartender guard the crowd? No The musician saw which song the bartender gua rded closely during as the crowd got rowdy. Did the bartender guard a song? No The musician saw which song the bartender yelled loudly over as the crowd got rowdy. Did the bartender yell quietly? No The musician saw which customers the barte nder yelled loudly at as the crowd got rowdy. Did the bartender yell quietly? No 29. A, B, C, D The architect wondered which law the builder read directly about after the in spection. Did the architect wonder what the builder read about? Yes The architect wondered which pipe the builder read directly about after the in spection. Did the architect wonder what the builder read about? Yes The architect wondered which pipe the build er stood directly over during the inspection. Did the architect wonder who the builder stood over? Yes The architect wondered which employee the buil der stood directly over during the inspection. Did the architect wonder what the builder stood over? Yes 30. A, B, C, D The public wanted to know which committee the mayor advised angrily over duri ng the financial crisis. Did the public angrily advise the mayor? No The public wanted to know which file the ma yor advised angrily about during the financial crisis. Did the public angrily advise the mayor? No The public wanted to know which meeting the mayor hurried angrily to during the financial crisis. Did the public angrily hurry the mayor? No The public wanted to know which bu ilding the mayor hurried angrily to during after the financial crisis. Did the public angrily hurry the mayor? No 31. A, B, C, D The cashier asked which mechanic the driver as ked angrily about before arguing over the bill. Did the driver ask about a mechanic? Yes The cashier asked which tire the driver asked angrily about before arguing over the bill. Did the driver ask about a tire? Yes The cashier asked which truck the driver cras hed angrily into before arguing over the bill. Did the driver crash into a truck? Yes

PAGE 199

199 The cashier asked which mechanic the driver cr ashed angrily into before arguing over the bill. Did the driver crash into a mechanic? Yes 32. A, B, C, D The investigator didn't know which dog the ne ighbor heard regularly about before being attacked. Did the neighbor hear about the attack? No The investigator didn't know which money the nei ghbor heard regularly about before being robbed. Did the neighbor hear ab out being robbed? No The investigator didn't know which dog the neighbor walked regularly around before being attacked. Did the neighbor walk a dog? No The investigator didn't know which fence the neighbor walked regularly around before being attacked. Did the neighbor walk around a dog? No

PAGE 200

200 APPENDIX K EXPERIMENT 3: DISTRACTER I TEMS 1. The mailman saw when the strange man gently covered the stray cat with a blanket. Did the mailman gently cover the cat? Yes 2. The silly clown saw how the young acrobat amazed the crowd with her brave performance. Did the silly acrobat see th e clown perform? No 3. Mary wondered why the smart student went to the principal's of fice during class. Did Mary know when the smart student went to the principal's office? Yes 4. The patient librarian asked us how the clever storyteller distracted the children during the ceremony. Did the librarian know who distracted the children during the ceremony? Yes 5. The passenger asked me when the mean conducto r had collected all of the tickets for first class. Did the passenger ask me about first class tickets? Yes 6. The bank auditor wondered why the old publishe r planned stupidly to ask for a loan. Did the old publisher plan cautiously? No 7. The teachers saw when the tired principal anxiously paced around the room filled with parents. Did the tired principal pace around the room? Yes 8. The careless girl knew that the busy maid had cleaned the stain from the counter. Did the girl see the stain on the counter? Yes 9. The serious diplomat knew why the reporter to ld the soldiers about the press conference concerning the war. Did the serious reporter know some thing about the soldiers? No 10. The nice lady asked why the helpful vendor closed the store earl y during the holiday shopping season. Did the vendor close the shop early during the holiday season? Yes 11. The detective wondered why the supervisor quietly inspected the fingerprint with a magnifying glass. Did the detective wonder why the quiet supervisor looked at the fingerprint with a magnifying glass? Yes 12. The tourists knew that the photographer had found numerous good spots for taking pictures of the canyon. Did the photographer want to take pictures of the canyon? Yes 13. Steve asked how the little raft reached the shore at th e same time that the pine logs floated past the bridge. Did Steve ask about the pine logs floating past something? Yes 14. The busy housewife knew when the new gardener we nt to shovel the snow off the sidewalk. Did the new gardener shovel snow off the driveway? No 15. The carpenter asked when th e foreign designer delivered th e new materials for the next phase. Did the foreign designer ask about the carpenter delivering something? No 16. The cook saw when the new hostess went secretly into the cooler and dra nk a bottle of wine. Did the new hostess drink some wine in the cooler? Yes 17. The eager clerk asked whether the short wi tness described the bur glar as tall. Did the short witness see the burglar? Yes

PAGE 201

201 18. The economist asked why the cl ever banker needed advice about investing in the market. Did the economist need advice about something related to investing in the market? No 19. The nervous crowd discussed how the bad shooter quickly got away before the police arrived on the scene. Did the bad shooter get away quickly? Yes 20. The nutritionist told the good nurse why she whispered a secret to the doctor about the patient. Did the nutritionist whisper something to the doctor? Yes 21. The audience gossiped about the angry golfer lo sing the match after the storm disrupted the tournament. Did something cause the golfer to lose the match? Yes 22. The careful nanny wanted to know how the shor t toddler crawled on t op of the table. Did the short toddler on the table concern the careful nanny? Yes 23. The lazy painters laughed when the mean owne rs had changed their minds about the colors. Did the mean painters know when the lazy owners picked new colors? No 24. The concerned policeman made sure that the baby was admitted quickly to the hospital after the accident. Did the policeman worry about th e baby after the accident? Yes 25. The boss suspected that the sneaky employees planned a surprise party for his promotion during the coffee break. Did the sneaky boss plan a coffee break? No 26. The architect criticized how the new assistant drew the desi gns for the new project. Did the architect finish the draw ings for the new project? No 27. Mark laughed when the happy drunk tried to purch ase an expensive bottle of wine with a one dollar bill. Did Mark know that the happy dr unk tried to buy wine? Yes 28. Tina scolded her children for not being at the bus stop when the school bus arrived. Did Tina's children make it to the bus stop late? Yes 29. Sarah wanted to know whether her angry boy friend went to work after they fought. Did Sarah's angry boyfriend wonder about her going to work? No 30. The skilled ballerina suspected that the mean actors laughed loudly when she ran off the stage frightened. Did the skilled actors laugh quietly? No 31. The nice crossing guard told us to help the old woman trying to cross the street during rush hour. Did the old woman need assist ance during rush hour? Yes 32. The staff suspected that the helpful office mana ger planned the party with a limited budget. Did the helpful office manage r plan something? Yes 33. The nun inquired as to whether the nice prie st planned his speech for the reunion while on the airplane. Did the nun wonder if the priest planned the reunion while on the plane? No 34. Bill reported that the speaker planned to read several acknowledgements about his work before accepting the prize. Did Bill know about the speaker planning some thing after accepting the prize? No 35. The ambassador reported that the spy su ccessfully completed the mission on time. Did the ambassador complete the mission on time? No

PAGE 202

202 The angry expert scolded the doctor for letting th e young patient stop the treatment before a complete recovery. Did the angry patient fully recover? No

PAGE 203

203 APPENDIX L EXPERIMENT 3: PLAUSIBILITY JUDGME NT QUESTIONNAIRE 1 The goal of the present questionnaire study is to investigate how nativ e speakers of English judge certain events. We would really apprecia te your help!! On the following pages, you will find sentence fragments such as: How plausible is it for. a man to wreck a car 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible I would like to ask you to rate the plausibility of such an event on a scale of 1 to 7, with 1 being really implausible and 7 being completely plausible. The above example is plausible, so you might rate it as a 5, 6, or 7. The following example is very odd, so you migh t rate it as a 1, 2, or 3, depending on how implausible you find it: a man to eat a car 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible Finally, below is an example that is not very plausible or really implausible, so you would probably rate it as a 3, 4, or 5. a dog to eat ice cream 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible When you are deciding your rating, do so with the first thing that crosses your mind Please do not think too long about it; what counts is the first thing. Please note that your particip ation is voluntary, and that you can stop whenever you wish. Before you start, please complete your name, which languages (including English) you speak, and at which age you started learning them. We will separate this sheet from the rest of this questionnaire before we process the data. Ho wever if you are uncomfortable completing the requested info, you can either use a separate sheet of paper and give this to the experimenter, or leave these lines blank. Name ___________________________ Languages spoken: language when l earned (e.g. from birth, high school) __________ __________ __________ __________ __________ __________ Please note that your particip ation is voluntary, and that you can stop whenever you wish. Thank you very much for your help!!!

PAGE 204

204 How plausible is it for rioters to escape a build ing 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an economist to notice a flaw 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a victim to see some doors 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an athlete to review a test 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a reporter to yell a name 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a gardener to cover a field 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a dog to jump a gate 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a senator to visit a friend 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a minister to hang a flower 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a man to save some coins 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a secretary to rest a package 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a treasurer to hang a sign 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a dentist to freeze a cake 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a driver to crash a truck 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a crowd to hear a leader 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an coach to cheer a player 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a realtor to watch a cont ract 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a child to drink a pudding 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a maid to hear a man 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a basketball player to force a fire 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a surgeon to shrink a cyst 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 205

205 a violinist to rush a sessi on 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a swimmer to lose a matc h 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a farm worker to race a tractor 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a photographer to kick a stand 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a golfer to praise a cat 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a painter to rest a door 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a king to follow a bear 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a cowboy to save a horse 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a man to save a trip 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an army to march prisoners 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a fisherman to watch a f ence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a guide to anticipate an accident 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a player to lose a match 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a cat to clean a bowl 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a policeman to point a gun 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a builder to stand a pipe 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a squirrel to drink some paint 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a senator to stop a law 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a mayor to hurry a meeting 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a boy to paint a friend 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a president to pay a parking ticket 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a neighbor to walk a dog 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 206

206 a chef to grow a bamboo 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a student to study a boat 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a boy to paint a fence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a nurse to walk a patient 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a detective to move an o ffice 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a tourist to study a map 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a customer to ask a waiter 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a lawyer to rush a client 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a diplomat to praise a meeting 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a cook to drip some fat 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a woman to advise a teenager 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a pilot to attack a jet 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a chef to bake a crust 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a student to copy an answe r 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a soldier to attack a general 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a janitor to swing a mop 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an editor to copy a book 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a bartender to yell a song 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a child to cheer a teacher 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a general to advise a prisoner 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a director to yell a line 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a grandmother to fill a trunk 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 207

207 a man to save a vacation 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a policeman to point a cr iminal 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a scientist to shrink a sponge 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a general to advise a ship 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a girl to eat caviar 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a butcher to eat a cherry 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e rioters to kill a cause 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a nurse to walk a bed 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a violinist to rush a song 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a pig to dance a jig 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a mayor to bury a building 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a student to copy a seat 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a mother to dance salsa 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a player to lose a stadium 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a lawyer to rush a room 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a cowboy to save a truck 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a neighbor to hear a fence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a wife to swing a towel 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a boy to paint a dog 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a pilot to attack an ocean 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a stranger to recall a beach 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a reporter to yell a docum ent 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 208

208 a dog to jump a training 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a crowd to hear a store 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a director to yell an actor 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a child to cheer a stage 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a florist to rotate a vase 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a swimmer to lose a team 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a judge to follow a clock 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a victim to see a lie 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a soldier to attack a gun 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a driver to crash a mechani c 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a teenager to cover a co mputer 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a violinist to rush a stage 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a bartender to yell customer s 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a hunter to chase a cloud 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a detective to move evi dence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a customer to ask a plate 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a tourist to study a hotel 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a child to burst a ball 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an editor to copy a pen 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a father to rotate a lamp 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a senator to visit a coat 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a spy to hunt a scholar 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 209

209 a maid to hear a room 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a messenger to drive a skateboard 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a cat to move a wall 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a dancer to stop a sink 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a professor to anticipate evidence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a student to study a desk 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a clown to burst a smile 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an coach to cheer a bench 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a builder to stand an employee 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a painter to rest a bed 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a chef to bake a cook 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a farm worker to race a fence 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an army to march a hill 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a woman to advise an accident 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an expert to freeze a so lution 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a secretary to rest an engineer 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a deputy to find a broom 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e a player to lose a round 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e an athlete to review a tabl e 1 2 3 4 5 6 7 Implausible-------------------------------------------------Plausibl e

PAGE 210

210 APPENDIX M EXPERIMENT 3: PLAUSIBILITY JUDGME NT QUESTIONNAIRE 2 The goal of the present questionnaire study is to investigate how nativ e speakers of English judge certain events. We would really apprecia te your help!! On the following pages, you will find sentence fragments such as: How plausible is it for. a man to wreck a car 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible I would like to ask you to rate the plausibility of such an event on a scale of 1 to 7, with 1 being really implausible and 7 being completely plausible. The above example is plausible, so you might rate it as a 5, 6, or 7. The following example is very odd, so you migh t rate it as a 1, 2, or 3, depending on how implausible you find it: a man to eat a car 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible Finally, below is an example that is not very plausible or really implausible, so you would probably rate it as a 3, 4, or 5. a dog to eat ice cream 1 2 3 4 5 6 7 Implausible----------------------------------------------Plausible When you are deciding your rating, do so with the first thing that crosses your mind Please do not think too long about it; what counts is the first thing. Please note that your particip ation is voluntary, and that you can stop whenever you wish. Before you start, please complete your name, which languages (including English) you speak, and at which age you started learning them. We will separate this sheet from the rest of this questionnaire before we process the data. Ho wever if you are uncomfortable completing the requested info, you can either use a separate sheet of paper and give this to the experimenter, or leave these lines blank. Name ___________________________ Languages spoken: language when l earned (e.g. from birth, high school) __________ __________ __________ __________ __________ __________ Please note that your particip ation is voluntary, and that you can stop whenever you wish. Thank you very much for your help!!!

PAGE 211

211 How plausible is it for a builder to stand a pipe 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a neighbor to hear a house 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a coach to advise a diagram 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a policeman to point a criminal 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a crowd to cheer a sidewalk 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a florist to rotate a vase 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a secretary to rest a phone 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a mayor to advise a building 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible rioters to kill a cause 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an army to march a tank 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a bartender to yell a song 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a scientist to shrink a sponge 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to study a bed 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a chef to grow some bamboo 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a clown to burst a smile 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a dog to guard a tree 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a king to follow a bear 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a grandmother to fill a trunk 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a reporter to ask a card 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a navy pilot to attack a ship 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a president to pay a parking ticket 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible

PAGE 212

212 a boy to float a toy 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a child to burst a ball 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a coach to advise a seat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a swimmer to float a surfboard 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an army to attack an order 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible rioters to escape a document 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to study a hotel 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a sailor to study a library 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a minister to hang a flower 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a chef to rush an oven 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a senator to fly an aisle 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a sailor to study a desk 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a crowd to cheer a tent 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a maid to stand some flowers 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a painter to paint a company 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a treasurer to hang a sign 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a director to call a movie 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a pig to dance a jig 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to study a restaurant 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a dog to guard a bush 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a girl to eat caviar 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a swimmer to float a raft 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible

PAGE 213

213 a navy pilot to attack a jet 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a spy to hunt a scholar 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a realtor to watch a contract 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to crash a crowd 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a secretary to rest a box 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a messenger to drive a skateboard 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a teenager to cover a computer 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a senator to fly a seat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a mayor to advise a file 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a cat to clean a kitten 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a secretary to call a client 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a coach to advise a clipboard 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a sailor to sink a desk 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a father to rotate a lamp 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a customer to rush a mop 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a butcher to eat a cherry 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an army to march a wall 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a deputy to find a broom 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a soldier to attack a county 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a mother to dance salsa 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an expert to freeze a solution 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to crash a street 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible

PAGE 214

214 a cat to move a ball 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a boy to float a boat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a player to cheer a door 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a cat to clean a paw 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a sailor to study a boat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a customer to rush a bathroom 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a painter to rest a brush 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a squirrel to drink some paint 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a soldier to attack an enemy 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a basketball player to force a fire 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a navy pilot to attack a town 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a nurse to clean medicine 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a maid to stand a statue 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a chef to rush a window 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a victim to see a lie 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible rioters to escape a computer 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a neighbor to hear some money 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a secretary to call a buyer 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an athlete to review a rest 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a cook to drip some fat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a customer to rush an aisle 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a diplomat to praise a meeting 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible

PAGE 215

215 an athlete to review a break 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a sailor to sink a boat 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a tourist to crash a statue 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a chef to rush a stove 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a boy to paint a friend 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a director to call a camera 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a farm worker to race a tractor 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a victim to point a person 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a reporter to ask a document 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible an army to attack a night 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a player to cheer a towel 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a navy pilot to attack a bomb 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible a soldier to attack an army 1 2 3 4 5 6 7 Implausible------------------------------------------Plausible

PAGE 216

216 APPENDIX N EXPERIMENT 3: GRAMMATICALITY JUDGMENT TAS K POST TESTING The goal of this questionnaire is to investigat e how different speakers of English judge certain sentences. On the following pages, you will find sentences such as: The man slept soundly. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical I would like to ask you to rate how grammatical you find the sentence to be on a scale of 1 to 7, with 1 being completely UNgrammatical and 7 being complet ely grammatical. The above example is grammatical, so you might rate it as a 6 or 7. The following example is ungrammatical, so you might rate it as a 1 or 2: Man sleeping. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical Finally, below is an example th at is not completely grammati cal or completely ungrammatical, so you would probably rate it as a 3, 4, or 5. The man imitated accurately. 1 2 3 4 5 6 7 Ungrammatical------------------------------------------Grammatical When you are deciding your rating, do so with the first thing that crosses your mind. Please do not think too long about it; what counts is the first thing. Before you begin, please ask the expe rimenter if you have any questions.

PAGE 217

217 The frugal couple saved for many years. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The teacher emphasized. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strict judge listened closely. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The crowd cheered the team. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical We anticipating gladly for the weekend. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The editor reviewed before the meeting. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The bank ignored the angry woman's request. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The teenager crashed the car. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The nurse carefully helped the doctor with the patient. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The child dropped the plate on the floor. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The criminal escaped the jail cell. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The librarian read slowly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strong boy lifted the box. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The trainee flew the airplane. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The busy diplomat entered the room unexpectedly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The artist painted. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clerk lifted the cans onto the counter. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 218

218 The small child floated the boat in the bathtub. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical She gotten a bad grade. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The crowd watched the movie. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The team lost. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The strict librarian placed the book. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The mailman carefully sealed the envelope. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical Tree swaying in the wind. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The printer hurried the job. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The man killed because he was angry. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The little boy ate pizza with pepperoni. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The army marched the hill towards the battle. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical His wife waited the late bus. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The child heard clearly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The hunter pointed the gun at the deer. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The policeman guarded all night long. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The angry wife pushed the chair. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The professor talked quickly the lecture. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 219

219 The student copied from the board. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The tired worker pushed the cart down the aisle. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The cowboy raced the horse. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The president put the file. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The maid cleaned for the party. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The bus driver hesitated the door. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The nurse rested her legs. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The secretary called about the problem. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The opponents agreed the problem. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The messenger rushed the package. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The busy director put the computer on the desk. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The rich executive stuck the money. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The captain sunk the boat. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The mean chef been given an award. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The chef baked for the festival. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The manager dusted the shelves in the store. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The reporter stuck the camera in the large bag. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 220

220 The rich lady liked the diamond earrings. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The machine maintain daily. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The director stood the camera on the tripod. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The dog attacked viciously. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The file included. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clever vender sold many hotdogs during the concert. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The kind student placed the ruler on the table. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The children asked about the song. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The girl walked the dog. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The clown dropped the ball. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The traveler visited for a while. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical She reacted to the accident quickly. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical Her mother yelled her son's name. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical The counselor advised all day. 1 2 3 4 5 6 7 UNgrammatical----------------------------------Grammatical

PAGE 221

221 LIST OF REFERENCES Altarriba, J., Kroll, Judith F., Scholl, A., and Rayner, K. 1996. The influence of lexical and conceptual constraints on reading m ixed-langu age sentences: evidence from eye-fixations and reading times. Memory & Cognition 24:477-492. Aoshima, Sachiko, Philips, Colin, and Weinbe rg, Amy. 2004. Processing filler-gap dependencies in a head-final language. Journal of Memory and Language 51:23-54. Ardila, Alfredo. 2003. Language representation and working memory with bilinguals. Journal of Communication Disorders 36:233-240. Birdsong, David ed. 1999. Second Language Acquisition and the Critical Period Hypothesis Mahwah: Lawrence Erlbaum Associates. Birdsong, David. 2006. Dominance, proficiency, and second language grammatical processing. Applied Psycholinguistics 27:46-49. Bley-Vroman, R. 1990. The logical probl em of second language learning. Linguistic Analysis 20:2-49. Blodgett, Allison, and Boland, Julie E. 2004. Differ ences in the timing of implausibility detection for recipient and inst rument prepositional phrases. Journal of Psycholinguistic Research 33:1-24. Boland, Julie E. 1993. The role of verb argument st ructure in sentence pr ocessing: distinguishing between syntactic and semantic effects. Journal of Psycholinguistic Research 22:133152. Boland, Julie E. 1997. The relationship between syntactic and semantic processes in sentence comprehension. Language and Cognitive Processes 12: 423-484. Boland, Julie. 2004. Linking eye movement to se ntence comprehension in reading and listening. In The On-line Study of Sentence Compre hension: Eye-tracking, ERPs and Beyond eds. Manual Carreiras and Charles Clifto n Jr. Hove, UK: Psychology Press. Clahsen, Harald, and Felser, Claudia. 2006a. Con tinuity and shallow structures in language processing. Applied Psycholinguistics 27:107-126. Clahsen, Harald, and Felser, Claudia. 2006b. How native-like is non-native language processing? Trends in Cognitive Sciences 10:564-570. Clifton, Charles, Traxler, Matthew J., Mohame d, Mohamed Taha, Williams, Rihana S., Morris, Robin K., and Raynera, Keith. 2003. The us e of thematic role information in parsing:Syntactic processing autonomy revisited. Journal of Memory and Language 49:317-334.

PAGE 222

222 Clifton, Charles, Jr., Speer, Shari, and Abne y, Steven P. 1991. Parsing arguments: phrase structure and argument structure as dete rminants of initial parsing decisions. Journal of Memory and Language 30:251-271. Crain, S., and Fodor, J. D. 1985. How can grammars help parsers? In Natural Language Parsing: Psychology, Computational and Theoretical Perspectives eds. D. Dowty, L. Kartuunen and A. Zwicky. Cambridge: Cambridge University Press. Daneman, Meredyth and Carpenter, Patricia A. 1980. Individual differences in working memory capacity. Journal of Verbal Lear ning and Verbal Behavior 19:450-466. DeKeyser, Robert. 2000. The robustness of critical period effects in second language acquisition. Studies in Second Language Acquisition 22:499-533. Dowens, Margaret Gillon, and Carreiras, Manue l. 2006. The shallow structure hypothesis of second language sentence processing : What is restricted and why? Applied Psycholinguistics 27:49-52. Dussias, P. E., and Cramer, T. R. 2006. The role of L1 verb bias on L2 sentence parsing. In Proceedings of the 30th annual Boston Uni versity Conference of Language Development eds. D. Bamman and T. Magnitskaia, 166-177. Featherston, Samuel. 2001. Empty Categories in Sentence Processing : Linguistics Today. Amsterdam: Johns Benjamin Publishing Company. Fodor, Janet Dean. 1989. Empty cate gories in sentence processing. Language and Cognitive Processes 4:S1 155-209. Felser, Claudia, Clahsen, Harald, and Munte, Thomas F. 2003. Storage and integration in the processing of filler-gap dependencies: An ERP study of topicalization and wh-movement in German. Brain and Language 87:345-354. Felser, Claudia, and Roberts, Leah. 2007. Pro cessing wh-dependencies in a second language: a cross-modal priming study. Second Language Research 23:9-36. Fernandez, Eva. 2002. Relative clause attachment in bilinguals and monolinguals. In Bilingual Sentence Processing eds. R.R. Heredia and J. Altarr iba, 187-215: Elsevier Science. Fernandez, Eva. 2003. Bilingual Sentence Processi ng: Relative Clause A ttachment in English and Spanish .vol. 29: Language Acquisition & Lan guage Disorders. Amsterdam: John Benjamins Publishing Company. Ferreira, Fernanda, and Clifton Jr., Charles. 1986. The independence of syntactic processing. Journal of Memory and Language 25:348-368.

PAGE 223

223 Ferreira, Fernanda, and Henderson, John M. 1990. Use of verb information in syntactic parsing: evidence from eye movements and word-by-word self-paced reading. Journal of Experimental Psychology: Learning, Memory, and Cognition 16:555-568. Ferreira, Fernanda, Bailey, K., and Ferraro, V. 2002. Good-enough representations in language comprehension. Current directions in psychological science 11:11-15. Ferreira, Fernanda. 2003. The misinterpr etation of noncanonical sentences. Cognitive Psychology 47:164-203. Frazier, Lyn. 1979. On comprehending sentences: s yntactic parsing strategies, University of Connecticut: Ph.D. Thesis. Frazier, Lyn, and Rayner, K. 1982. Making and correcting errors during sentence comprehension: Eye movements in the analys is of structurally ambiguous sentences. Cognitive Psychology 14:178-210. Frazier, Lyn, and Flores D'Arcai s, Giovanni B. 1989. Filler driven parsing: a study of gap filling in dutch. In Journal of Memory and Language 331-344. Frenck-Mestre, Cheryl, and Pynte, J. 1997. Synt actic ambiguity resoluti on while reading in second and native languages. The Quarterly Journal of Experimental Psychology 50A:119-148. Frenck-Mestre, Cheryl. 2002. An on-line look at sentence processing in the second language. In Bilingual Sentence Processing eds. R.R. Heredia and J Altarriba, 217-236. New York: Elsevier Science B.V. Frenck-Mestre, Cheryl. 2005. Eye-movement r ecording as a tool for studying syntactic processing in a second language: a review of methodologies and experimental findings. Second Language Research 21:175-198. Frenck-Mestre, Cheryl. 2006. Co mmentary on Clahsen and Felser. Applied Psycholinguistics 27:64-65. Friederici, Angela D., and Frisch, Stefan. 2000. Ve rb argument structure pr ocessing: the role of verb-specific and argument-specific information. Journal of Memory and Language 43:476-507. Gahl, Susanne, Jurafsky, Dan, and Roland, Douglas. 2004. Verb subcategorization frequencies: American English corpus data, methodologica l studies, and cross-corpus comparisons. Behavior Research Methods, Instruments & Computers 36:432-443. Garnsey, Susan M, Pearlmutter, Neal J, Myer s, Elizabeth, and Lotocky, Melanie A. 1997. The Contributions of Verb Bias and Plausibil ity to the Comprehension of Temporarily Ambiguous Sentences. Journal of Memory and Language 37:58-93.

PAGE 224

224 Gass, Susan. 1996. Second language acquisition and linguistic theory: the role of language transfer. In Handbook of Second Language Acquisition eds. William C. Ritchie and Tej K. Bhatia, 317-345. San Diego: Academic Press. Goodluck, Helen, Finney, Malcolm, and Sedivy, Julie. 1991. Sentence completeness and fillergap dependency parsing. In OTS Yearbook 1991 eds. Peter Coopmans, Bert Schouten and Wim Zonneveld, 19-32. Utrecht: University of Utrecht. Goodluck, Helen, Finney, Malcolm, and Ling, Ma rgarette. 1995. Gap accessibility and children's sentence processing. Cahiers Linguistiques D'Ottawa 23 Supplement:1-32. Hahne, A. 2001. What's different in second languag e processing? Eviden ce from event-related brain potentials. Journal of Psychol inguistic Research 30:251-266. Hahne, A., and Friederici, Angela. 2001. Pro cessing a second language: late learners' comprehension mechanisms as revealed by event-related brain potentials. Bilingualism: Language and Cognition 4:123-141. Heredia, R.R., and Altarriba, J eds. 2002. Bilingual Sentence Processing vol. 134. Advances in Psychology. Amsterdam: Elsevier. Hoover, M. L., and Dwivedi, V.D. 1998. Synt actic processing by skilled bilinguals. Language Learning 48:1-29. Ioup, Georgette, Boustagui, Elizabeth, El Tigi, Manal, and Moselle, Martha. 1994. Reexamining the critical period hypothesis. Studies in Second Language Acquisition 16:73-98. Johnson, J.S., and Newport, E.L. 1989. Critical pe riod effects in second language learning: The influence of maturational state on the ac quisition of English as a second language. Cognitive Psychology 21:60-99. Juffs, Alan, and Harrington, Mi chael. 1995. Parsing effects in second language sentence processing: subject and object as ymmetries in wh-extraction. Studies in Second Language Acquisition 17:483-516. Juffs, Alan, and Harrington, Mich eal. 1996. Garden path sentences and error data in second language sentence processing. Language Learning 46:283-326. Juffs, Alan. 1998. Some effects of first language argument st ructure and morphosyntax on second language sentence processing. Second Language Research 14:406-424. Juffs, Alan. 2004. Representation, processing an d working memory in a second language. Transactions of the Philological Society 102:199-225. Juffs, Alan. 2005. The Influence of first language on the processing of wh-movement in English as a second language. Second Language Research 21:121-151.

PAGE 225

225 Just, Marcel, and Carpenter, Patricia A. 1992. A capacity theory of comprehension: individual differences in working memory. Psychological Review 99:122-149 Just, Marcel, Carpenter, Patric ia A., and Wooley, Jacqueline D. 1982. Paradigms and processes in reading comprehension. Journal of Experiment al Psychology: General 111:228-238. Kiaer, Jieun, and Kempson, Ruth. 2005. Pro-ac tive parsing of korean scrambling. Paper presented at Proceedings of the 24th West Coast Conference on Formal Linguistic. Knoeferle, Pia, Crocker, Matthew W., Scheepers, Christoph, and Pickering, Martin J. 2005. The influence of the immediate visual contex t on incremental thematic role-assignment: evidence from eye-movements in depicted events. 95:95-127. Kroll, Judith F., and De Groot, Annette M.B. eds. 2005. Handbook of bilingualism: psycholinguistic approaches New York: Oxford University Press. Lenneberg, E. H. 1967. Biological Foundations of Language New York: Wiley. Long, Michael. 1990. Maturational constraints on language development. SSLA 12:251-285. Love, Tracy, Maas, Edwin, and Swinney, David. 2003. The influence of language exposure on lexical and syntactic language processing. Experimental Psychology 50:204-216 MacDonald, Just, M. A., and Carpenter, P. A. 1992. Working memory constraints on the processing of syntactic ambiguity. Cognitive Psychology 24:56-98. Marinis, Theodore, Roberts, Leah, Felser, Cl audia, and Clahsen, Harald. 2005. Gaps in second language sentence processing. Essex Research Reports in Linguistics 45:43-79. Mitchell, D. 2004. On-line methods in language pr ocessing: Introduction and historical review. In The On-line Study of Sentence Compre hension: Eye-tracking, ERPs and Beyond eds. Manual Carreiras and Charles Clifton Jr., 15-32. Hove, UK: Psychology Press. Miyake, A., and Friedman, Naomi P. 1998. I ndividual differences in second language proficiency: Working memory as language aptitude. In Foreign Language Learning: Psycholinguistic Studies on Training and Retention ed. A. & L. Bourne Healy, 339-365. Mahwah, NJ: Lawrence Erlbaum Associates. Mueller, Jutta L. 2005. Electrophysiological correlates of second language processing. Second Language Research 21:152-174. Nakano, Y., Felser, Claudia, and Clahsen, Harald. 2002. Antecedent priming at trace positions in Japanese long-distance scrambling. Journal of Psycholinguistic Research 31:531-571. Nicol, J., and Swinney, David. 1989. The role of structure and co-refere nce asssignment during sentence comprehension. Journal of Psycholinguistic Research 18:5-19.

PAGE 226

226 Ojima, Shiro. 2005. An ERP study of second language learning after childhood: effects of proficiency. Journal of Cognitive Neuroscience 17:1212-1228. Papadopoulou, Despina. 2005. Reading-time studies of second language ambiguity resolution. Second Language Research 21:98-120. Paradis, Michel. 1994. Neurolinguist ic aspects of implicit and explicit memory: implications for bilingualism and SLA, 393-419: Academic Press. Penfield, W., and Roberts, L. 1959. Speech and Brain-mechanisms Princeton: Princeton University Press. Pickering, M., and Barry, G. 1991. Sentence processing without empty categories. Language and Cognitive Processes 6:229-259. Pickering, Martin J., and Traxler, Matthew J. 2003. Evidence against the Use of Subcategorisation Frequency in the Processing of Unbounded Dependencies. Language and Cognitive Processes 18:469-503 Rayner, K., and Sereno, Sara C. 1994. Eye movement s in reading: psycho linguistics studies. In Handbook of psycholinguistics ed. M. A. Gernsbacher: Academic Press. Reichle, Erik D., Rayner, Keith, and Pollatsek, Alexander. 2003. The E-Z reader model of eyemovement control in reading: comparisons to other models. Behavioral and Brain Sciences 26: 445-526. Rossi, Sonja, Gugler, Manfred F, Friederici, Angela D., and Hahne, Anja. 2006. The impact of proficiency on syntactic secondlanguage processing of German and Italian: Evidence from event-related potentials. Journal of Cognitive Neuroscience 18:2030-2048. Sabourin, Laura. 2003. Grammatical gender and second language processing: an ERP study, Linguistics, University of Groningen. Sabourin, Laura, and Haverkort. 2003. Neural substrates of representation and processing of a second language. In The lexicon-syntax interfa ce in second language acquisition ed. Roeland van Hout. Philadelphia: John Benjamins Publisher. Sabourin, Laura. 2006. Does the shallow structures proposal account for qualitative differences in first and second language processing? Applied Psycholinguistics 27:81-84. Schmidt, R. 1990. The role of consci ousness in second language learning. Applied Linguistics 11:17-46. Schmidt, R. 1992. Psychological mechanisms underlying second language learning. Studies in Second Language Acquisition 14:357-385.

PAGE 227

227 Sedivy, Julie C., Tanenhaus, Michael K., Chambe rs, Craig G., and Carlson, Gregory N. 1999. Achieving incremental semantic interpre tation through contextu al representation. Cognition 71:109-147. Segalowitz, Norman. 1986. Skilled r eading in second language. In Language Processing in Bilinguals: Psycholinguistic and Ne uropsychological Perspectives ed. Jyotsna Vaid, 320. Hillsdale: Lawrence Erlbaum Associates, Inc. Sekerina, Irina A., and Brooks, Patricia J. 2006. Pervasiveness of shallow processing. Applied Psycholinguistics 27:84-88. Sinclair, J McH. 1991. Corpus, Concordance, Collocation Oxford: Oxford University Press. Skehan, Peter. 1989. Individual Differences in Second-language Learning London: Edward Arnold. Skehan, Peter. 1998. A Cognitive Approach to Language Learning Oxford: Oxford University Press. Staub, Adrian. 2007. The parser doesnt ignore intransiti vity, after all. Journal of Experimental Psychology: Learning, Memory, and Cognition 33:550-560. Steinhauer, Karsten. 2006. How dynamic is second language acquisition? Applied Psycholinguistics 27:92-95. Steinhauer, Karsten, White, Erin, King, Erin, Cornel l, Sonia, Genesee, Fred, and White, Lydia. 2006. The neural dynamics of second language acquisition: evidence from event-related potentials. Poster presented at Cognitive Neuroscience Society San Francisco. Stowe, Laurie. 1986. Parsing WH-construc tions: evidence for on-line gap location. Language and Cognitive Processes :227-245. Swinney, David, Ford, M., and Bresnan, J. 198 9. Coreference assignment during sentence processing. In Language Structure and Processing eds. B. Grosz, R. Kaplan, M. Macken and I. Sag. Stanford: CSLI. Tanenhaus, Michael K., Boland, Julie, Garnsey, Susan M., and Carlson, Greg N. 1989. Lexical structure in parsing long-distance dependencies. Journal of Psycholinguistic Research 18:37-50. Tokowicz, N., and MacWhinney, B. 2005. Implicit and explicit measures of sensitivity to violations in second language grammar. Studies in Second Language Acquisition 2:173204. Townsend, D.J., and Bever, T.G. 2001. Sentence comprehension: th e integration of habits and rules Cambridge: MIT Press.

PAGE 228

228 Trueswell, John C., Tanenhaus, Michael K., and Kello, Christopher. 1993. Verb-specific constraints in sentence proces sing: separating effects of lexi cal preference from gardenpaths. Journal of Experimental Psychology: Learning, Memory, and Cognition 19:528553. Trueswell, John C., Tannenhaus, Michael K., an d Garnsey, Susan M. 1994. Semantic influence on parsing: use of thematic role inform ation in syntactic ambiguity resolution. Journal of Memory and Language 33:285-318. Trueswell, John C. 1996. The role of lexical frequency in syntactic ambiguity resolution. Journal of Memory and Language 35:566-585. Tyler, L. K. 1989. The role of lexical repr esentations in language comprehension. In Lexical Representation and Process ed. William Marslen-Wilson. Cambridge: MIT Press. Ullman, Michael. 2001. The Neural Basis of Lexicon and Grammar in First and Second Language: The Declarative/Procedural Model. Bilingualism: Language and Cognition 4:105-122. Weber-Fox, Christine M., and Neville, Helen J. 1996. Maturational cons traints on functional specializations for language processing: ERP and behavior al evidence in bilingual speakers. Journal of Cognitive Neuroscience 8:231-256. Weber-Fox, Christine M., and Neville, Helen J. 1999. Functional neur al subsystems are differentially affected by delays in second language immersion: ERP and behavioral evidence. In Second Language Acquisition and the Critical Period Hypothesis ed. David Birdsong, 23-37. Mahwah: Lawrence Erlbaum Associates. White, Lydia, and Genesee, Fred. 1996. How native is near-native? Th e issue of ultimate attainment in adult second language acquisition. Second Language Research 12:233-265. White, Lydia, and Juffs, Alan. 1998. Constraints on wh-movement in two different contexts of nonnative language acquisition: competence and processing. In The Generative Study of Second Langauge Acquisition eds. Suzanne Flynn, Gita Martohardjono and Wayne O'Neil, 111-129. Mahwah, NJ: Lawrence Erlbaum. Williams, John N., Mobius, Peter, and Kim, Choonkyong. 2001. Native and non-native processing of English wh-questions: parsing strategies and plausibility constraints. Applied Psycholinguistics 22:509-540. Williams, John N. 2006. Incremental interpretati on in second language sentence processing. Bilingualism: Language and Cognition 9:71-88.

PAGE 229

229 BIOGRAPHICAL SKETCH Andrea Dallas has been working toward her Ph.D in linguistics at the University of Florida (UF) since 2004. She completed her M. A. in linguistics at UF in 2004 with a specialization in psycholinguistics and second language acquisition. During her time at UF, she worked as a teaching assistant for UF's Academ ic Spoken English Program and the Program in Linguistics, and worked as a res earch assistant in the Kaan Br ain & Language Lab. Before doing graduate work in linguistics, she lived in Sos nowiec, Poland for 2 years, teaching English and completing her M.Ed. via Framingham State College. She received her B.A. in English with a certificate in linguistics from Flor ida International University in 1999. Her research interests are psycholinguistics and second language acqui sition; more specifically, second language sentence processing and the role of individual differences. Her doctoral study was supported by a grant from Language Learning: A Journal of Research in Language Studies.