<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-05-31.

Permanent Link: http://ufdc.ufl.edu/UFE0024592/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-05-31.
Physical Description: Book
Language: english
Creator: Babb, Michelle
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Communication Sciences and Disorders -- Dissertations, Academic -- UF
Genre: Communication Sciences and Disorders thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Michelle Babb.
Thesis: Thesis (M.A.)--University of Florida, 2009.
Local: Adviser: Edmonds, Lisa Anna.
Electronic Access: INACCESSIBLE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024592:00001

Permanent Link: http://ufdc.ufl.edu/UFE0024592/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2011-05-31.
Physical Description: Book
Language: english
Creator: Babb, Michelle
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2009

Subjects

Subjects / Keywords: Communication Sciences and Disorders -- Dissertations, Academic -- UF
Genre: Communication Sciences and Disorders thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Michelle Babb.
Thesis: Thesis (M.A.)--University of Florida, 2009.
Local: Adviser: Edmonds, Lisa Anna.
Electronic Access: INACCESSIBLE UNTIL 2011-05-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2009
System ID: UFE0024592:00001


This item has the following downloads:


Full Text

PAGE 1

1 GENERALIZATION AND ERROR TRENDS IN TWO PARTICIPANTS WITH MODERATE SEVERE APHASIA FOLLOWING VERB NETWORK STRENGTHENING TREATMENT (VNeST) By MICHELLE E. BABB A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2009

PAGE 2

2 2009 Michelle E. Babb

PAGE 3

3 To my beloved grandfather who passed just -months prior to fulfi llment of this dream whose strength character, and resolve will never be forgotten To my parents for their unceasing support, encouragement, and prayers throughout this journey, and for teaching me dreams are meant to be pursued Finally, to the rest of my family, whose love and support guided me through it all

PAGE 4

4 ACKNOWLEDGMENTS I would like to thank my mentor, Dr. Lisa Edmonds for her instruction, guidance, and encouragement throughout the graduate program and the research process and for instilling in me a passion for research. I would also like to thank Dr. Jamie Reilly, for his comments and collaboration during the writing process Finally, I would like to thank the individuals who participated in the study for their enthusiasm and motivation t hroughout the treatment process.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................................... 4 LIST OF TABLES ................................................................................................................................ 7 LIST OF FIGURES .............................................................................................................................. 8 ABSTRACT .......................................................................................................................................... 9 CHAPTER 1 INTRODUCTION ....................................................................................................................... 11 Genera l Introduction to Aphasia ................................................................................................ 11 Lexical Retrieval Treatments: Single Word Naming ................................................................ 11 Verb Network Strengthening Treatment (VNeST): R ecent Approach .................................... 14 Lexical Retrieval Models ............................................................................................................ 16 Ellis and Young Lexical Retrieval Model (from Kay, Lesser, and Coltheart, 1992) ...... 16 Two -step Lexical Access and Spreading Activation Models ........................................... 18 Errors in Aphasia ......................................................................................................................... 20 Research Questions ..................................................................................................................... 24 2 METHODS .................................................................................................................................. 27 Participants .................................................................................................................................. 27 Pre -testing Measures ................................................................................................................... 28 Stimuli Development .................................................................................................................. 29 Sentence Elicitation Pictures ............................................................................................... 29 Control Task ......................................................................................................................... 30 Experimental Design ................................................................................................................... 30 Procedures .................................................................................................................................... 31 Baseline, Treatment, and Maintenance Probe Measures ................................................... 31 Treatment ............................................................................................................................. 32 Scoring ......................................................................................................................................... 33 Error Analysis .............................................................................................................................. 34 3 RESULTS .................................................................................................................................... 40 Probes ........................................................................................................................................... 40 Participant 1 (P1) ................................................................................................................. 40 Participant 2 (P2) ................................................................................................................. 41 Pre and Post treatment Lexical Retrieval Measures: Accuracy .............................................. 42 Participant 1 (P1) Accuracy ................................................................................................ 42 Participant 2 (P2) Accuracy ................................................................................................ 43 Pre and Post treatment Lexical Retrieval Measures: Error Analysis ..................................... 43

PAGE 6

6 Participant 1 (P1) Error Analysis ........................................................................................ 43 Nouns ............................................................................................................................ 43 Verbs ............................................................................................................................. 44 Participant 2 (P2) Error Analysis ........................................................................................ 45 Nouns ............................................................................................................................ 45 Verbs ............................................................................................................................. 46 4 DISCUSSION .............................................................................................................................. 58 Consideration of Participant 1 (P1) Results .............................................................................. 58 Probes ................................................................................................................................... 58 Pre to Post treatment Language Measures ........................................................................ 60 Consideration of Participant 2 (P2) Results .............................................................................. 64 Probes (Spoken) ................................................................................................................... 64 Pre to Post treatment Language Measures (Spoken) ....................................................... 67 Probes (Written) ................................................................................................................... 69 Pre to Post treatment Language Measures (Written) ....................................................... 71 General Considerations ............................................................................................................... 72 LIST OF REFERENCES ................................................................................................................... 76 BIOGRAPHICAL SKETCH ............................................................................................................. 79

PAGE 7

7 LIST OF TABLES Table page 2 1 Participant demographic information .................................................................................... 36 2 2 Pre and post treatment scores for all administered tests .................................................... 37 2 3 Error analysis categories ........................................................................................................ 38 2 4 Example probe scoring .......................................................................................................... 39

PAGE 8

8 LIST OF FIGURES Figure page 1 1 Ellis and Young Model. ......................................................................................................... 25 1 2 Dell Aphasia Model. .............................................................................................................. 26 3 1 P1s Agent, verb, and patient production within sentences. ................................................ 48 3 2 P1s Adjective control task. ................................................................................................... 48 3 3 P2s Spoken agent, verb, and patient production within sentences. ................................... 49 3 4 P2s Written agent, verb, and patient production within sentences. ................................... 49 3 5 P2s Adjective control task. ................................................................................................... 50 3 6 Visual representation o f P1s accuracy across measures. .................................................... 51 3 7 Visual representation of P2s accuracy across measures. .................................................... 52 3 8 P1s Error analysis graph s.. ................................................................................................... 53 3 9 P2s Spoken error analysis graphs. ....................................................................................... 55 3 10 P2s Written error analysis graphs.. ...................................................................................... 57

PAGE 9

9 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partia l Fulfillment of the Requirements for the Degree of Master of Arts GENERALIZATION AND ERROR TRENDS IN TWO PARTICIPANTS WITH MODERATE SEVERE APHASIA FOLLOWING VERB NETWORK STRENGTHENING TREATMENT (VNeST) By Michelle E. Babb May 2009 Chair: Lisa Edmond s Major: Communication Sciences and Disorders Lexical retrieval deficits characterize a ll aphasia types, directly impacting communication and ultimately, quality of life. Rehabilitation of lexical retrieval deficits most often involves targeting single word naming of items (e.g., nouns); however, both generalization to untrained items and a pplicability to daily communication is limited. Recently, there has been a focus on improving communication beyond single word naming via exploitation of verbs and associated thematic roles (e.g., noun concepts), and engaging the semantic system for sprea d of activation to semantically related untrained items. One such treatment that employs these principles is Verb Network Strengthening Treatment (VNeST), which has revealed promising results in four participants with moderate fluent aphasia. In the curr ent study, two participants with moderate severe nonfluent aphasia received VNeST. Three specific aims were established: 1) examine the effect of VNeST on individuals with more severe aphasia and lexical retrieval impairments 2) examine generalization pat terns in detail, and 3) analyze the evolution of error patterns on lexical retrieval from pre to post -treatment. Following treatment, both participants demonstrated significant improvements in lexical retrieval for trained and untrained verb networks, which include nouns and verbs. Generalization patterns differed between participants, as reflected by

PAGE 10

10 trends in accuracy. Evolution of errors was used to localize levels of impairment within models of lexical retrieval for each participant. Attention to m ultiple modalities, implementation of sensitive scoring measures, and the importance of error analysis in aphasia treatment research is discussed.

PAGE 11

11 CHAPTER 1 INTRODUCTION General Introduction to Aphasia Aphasia is an acquired neurogenic language disorder that results from a stroke, head injury, or infection (e.g., encephalitis). The most common cause of aphasia is stroke, also referred to as a cerebrovascular accident (CVA), with nearly 750,000 new cases reported each year. Men are at a greater risk for stroke; however, women have a much higher mort ality rate. This risk is exponentially increased when women pair the use of oral contraceptives with smoking. Risk factors for stroke include age, hypertension, hypercholesterolemia heart disease, smoking, and obesity (Murray & Clark, 2006). Aphasia do es not affect ones general intellect; instead, it is an impairment in accessing, understanding, and using language following neurologic injury (Chapey & Hallowell, 2001; Murray & Clark, 2006). All modalities of language speaking, writing, comprehending, and reading can be affected following a stroke, although some modalities can be affected to greater extent than others. Depending on the location of the neurologic injury, the disruption of language can be differentially impaired, such that deficits of var ying extent can arise in tasks of comprehension, naming, and repetition (Murray & Clark, 2006). Word retrieval difficulties (anomia) are characteristic of all types of aphasia and persist into the chronic phase (Hillis, 2007; Raymer & Gonzalez Rothi, 2002) Lexical Retrieval Treatments : Single Word Naming A substantial amount of literature discussing treatments for word retrieval deficits in aphasia exists. Most naming treatments tend to address semantic or phonologic deficits with treatments that primar ily consist of picture naming at the single word level. Treatment usually addresses the underlying impairment, such that anomia due to an impaired semantic system

PAGE 12

12 would focus on meanings (e.g., matching a word to the correct picture in the presence of dist ractors, sorting pictures according to semantic category), whereas anomia due to the impairment of retrieving word sounds would require a phonologic treatment (e.g., repetition, rhyming judgments, picture naming using phonemic cueing strategies) (Kiran, 2008; Nickels, 2002). However, Hillis (1998) and Nickels argue that since the semantic system is central to lexical tasks, treatment focusing on semantics should ultimately improve performance across tasks. Furthermore, most clinical strategies for remedia tion of lexical retrieval difficulties simultaneously employ both semantic and phonologic mechanisms (e.g., word -picture matching, or phonemic cueing during picture naming); therefore, connections between semantic and phonologic systems are strengthened as a result of simultaneous activation (Nickels). Treatments for lexical retrieval deficits in aphasia have primarily focused on nouns; however, both nouns and verbs tend to be affected, with historically greater verb retrieval deficits in persons with nonfl uent aphasia and greater noun retrieval deficits in persons with fluent aphasia (Druks 2002; Raymer & Ellsworth, 2002). In recent years, remediation of verb retrieval deficits via verb -centered treatments has received recognition, largely as a result of the assumption that verbs are central to sentence formulation (Raymer & Ellsworth; Raymer & Kohen, 2006). Similar to studies involving lexical retrieval of nouns, treatment for impaired lexical retrieval of verbs may focus on semantic or phonological prin ciples. Raymer and Ellsworth conducted a study comparing the effect of semantic, phonological, and rehearsal (e.g., repetition) treatments for verb retrieval deficits within a single participant. Following the three treatment protocols, the participant i mproved naming of trained verbs and sentences; however, there was no generalization to untrained verbs. This item specific improvement following each

PAGE 13

13 treatment protocol was likely the consequence of semantic influence by way of picture presentation during each protocol (Raymer & Ellsworth). According to Loverso, Prescott, and Selinger (1988), verbs are the core of simple sentences and the catalyst for retrieval of nouns associated with a given verb. Cueing Verbs Treatment (CVT), as described by Loverso et al., involves generating, copying, writing, and repeating noun concepts in response to wh questions for a target verb. Following CVT, participants improved on pre to post -treatment scores for the Porch Index of Communication Ability (PICA ; Porch, 1973), as well as in retrieval of trained and untrained verbs and related nouns (Loverso, et al. ; Prescott, Selinger, & Loverso, 1982). Recently, treatments employing principles of semantic feature analysis (SFA) have gained recognition on the premise that ac tivation and retrieval of features of a particular target strengthen the semantic network surrounding the target, thus promoting retrieval of the trained items and untrained semantically related items that share features with the trained items. (Raymer & G onzalez Rothi, 2002; Wambaugh & Ferguson, 2007, Wilshire & Coslett, 2000). In Boyle (2004) 2 participants (one with mild anomic, one with moderate Wernickes aphasia) received SFA treatment on nouns and improved in confrontation naming on trained and unt rained nouns. Post hoc analysis revealed untrained nouns were from shared as well as unrelated semantic categories. Lexical retrieval in discourse, however, was limited. Kiran and Thompson (2003) applied the complexity effect to semantic treatment of category features with the expectation of increasing generalization. The complexity effect proposes that training more complex (atypical) items (e.g., penguinlays eggs, has wings, but can not fly ) promotes greater generalization than training less complex (t ypical) items (e.g., robin lays eggs, has wings, flies ). This supposition is based on the theoretical principle that

PAGE 14

14 atypical items within a category represent not only the shared features common to all items within the category, but also the diverse vari ations of item features belonging to the category (Kiran & Thompson). Four participants with moderate to moderate -severe fluent aphasia participated in Kiran and Thompsons semantic features analysis treatment. The results suggest that training atypical items generalizes to untrained intermediate and typical items, but not vice versa. Additionally, the participants in the improved on pre to post treatment standardized language measures and exhibited an evolution of errors that reflected increased activation of items semantically related to the target. Although semantic feature treatment promotes activation of features associated with a trained item, encouraging generalization to untrained items sharing similar features, treatment is still directed at single lexical items and has primarily been investigated in nouns. Therefore, functionality is limited, and widespread generalization to untrained items and tasks is also limited (Nickels, 2002). Verb Ne twork Strengthening Treatment (VNeST) : Recent Approach Verb Network Strengthening treatment (VNeST) is a semantic treatment that aims at improving word retrieval in single word, sentence and discourse contexts. The treatment is focused on activating semant ic networks surrounding a trained verb rather than the semantic aspects of single lexical items. The activation of large semantic networks during treatment promotes generalization from trained to untrained items and tasks (Edmonds, Nadeau, & Kiran, 2009; K iran & Bassetto, 2008). Studies conducted by Loverso and colleagues (1988) employed a similar treatment to that of VNeST, with their treatment referred to as verb as core and later, Cueing Verbs Treatment (CVT). The verb is considered to be the anchor (e.g., the core) of sentences and the determiner of concepts associated with it. Raymer and Kohen (2006) concurred that verbs are a significant component of sentences in that they predict argument structure and noun selection. The idea that

PAGE 15

15 verbs influenc e noun retrieval leads directly into the notion of thematic roles, which are the agent (e.g., the doer), patient (e.g., the receiver) and instrument (e.g., the object) of the action represented relationally by the chosen verb. McRae, Ferretti, and colleagu es ( Ferretti McRae, & Hatherell, 2001; McRae, Hare, & Ferretti, 2005) have found in young adults that verbs and their related agents, patients and instruments show bidirectional activation in priming studies (e.g., people respond faster to related noun-ve rb pairs than unrelated noun verb pairs across a variety of tasks) and locations prime verbs but not vice versa Edmonds and Mizrahi (in preparation) have replicated the findings of agent and patient bidirectional priming in young adults and have shown sim ilar priming patterns in older adults. Thus, it appears that verbs and their thematic roles are interrelated, with activation of one supporting activation of the other (Druks, 2002; Edmonds & Mizrahi, in preparation; Ferretti et al; McRae et al.). Most ve rbs can be the feature of numerous agents and patients (e.g., noun concepts), thus connecting these concepts to each other. For example, for a given verb such as measure an agent could be a chef and a patient could be sugar ; however, measure can also act ivate noun concepts for carpenter/lumber and surveyor/land (e.g., agents that can measure something and patients that can be measured) (Edmonds et al., 2009). Activating multiple noun concepts for any one verb is not the only benefit of this type of treat ment. Edmonds et al. (2009) hypothesized that features of one verb can engage features of a semantically related verb. Engaging the semantic system for one verb inherently activates semantically related verbs that share similar features; furthermore, thi s activation of a semantically related verb also activates the noun concepts associated with it, thus representing a large semantic network (Edmonds et al.). For example, the verb measure is activated, along with

PAGE 16

16 its related noun concepts. By way of sprea ding activation, a semantically related verb such as weigh can also be activated along with its noun concepts (e.g., butcher/meat nurse/patient ). In Edmonds et al. (2009), 4 participants with moderate aphasia (two fluent and two nonfluent) received VNeS T for 4 to 6 weeks. All participants generalized to sentence production for picture description of sentences containing trained verbs and semantically related untrained verbs. Further improvement in sentence production was captured on the Northwestern Verb Production Battery (NVPB) (Thompson, 2002), with all participants improving greater than 15% over pre treatment levels. Single word naming of nouns and verbs, as assessed by the Boston Naming Test (BNT ; Goodglass, Kaplan, & Weintraub, 1983) and the NVPB, improved greater than 10% for 3 participants. Connected speech samples in response to picture descriptions and a narrative showed significant improvement of complete utterances (utterances containing subjects and verbs and relevant to the topic) for 3 pa rticipants. The control measure, an adjective retrieval task, showed no significant improvements (Edmonds et al ). Thus, the preliminary findings of this phase I treatment study showed encouraging generalization to various lexical items and tasks, more so than previous semantic treatments for lexical retrieval (Edmonds et al; Raymer & Ellsworth, 2002). Lexical Retrieval Models Ellis and Young Lexical Retrieval Model (from Kay, Lesser, and Coltheart, 1992) There are lexical retrieval models that depict mod ality specific input and output lexicons that contain representations of spoken and written words, as well as a recognition system for visually presented objects. A central semantic system assigns meanings (Raymer & Gonzalez Rothi, 2000). The Ellis and Yo ung model is perhaps one of the most recognized psycholinguistic transcoding models of language processing. The work of Ellis and Young has been adapted and

PAGE 17

17 described by Kay, Lesser, and Coltheart (1992) as a theoretical foundation for the Psycholinguisti c Assessments of Language Processing in Aphasia (PALPA), an assessment tool for evaluating language processing abilities. This model of lexical retrieval is divided into components with pathways that connect them. Each route supports a specific language processing function. Inputs and outputs are organized around a central semantic system, with three inputs (e.g ., speech, viewed pictures and objects, print) and two outputs (e.g., speech, writing). This model functions top-down in that information do es not feed back to previous components. However, it does allow for crossover of information from one modality of input to an alternate method of output, such as for writing to dictation (e.g. spoken input to written output). Kay et al. (1992) broadly de scribe the semantic system as a storehouse of meanings of all the words one knows; however, given that the semantic system is central to all lexical tasks, this description is under -specified. Although not universally accepted the semantic system is gene rally considered amodal (e.g., a unitary mechanism), central to all lexical tasks, and accessible by all input and output modalities. It is thus responsible for assigning meanings to visually presented objects, gestures and spoken and written words. Thus, impairment of the semantic system would result in errors regardless of the modality of input or output (Funnell, 2002; Raymer & Gonzalez Rothi, 2000). During a naming task, some argue that the semantic level and the lexical level interact (Chialant, Cos ta, & Caramazza, 2002). For example in accessing the name for a picture of a cat the semantic system locates features of the target (e.g., domestic walks on four legs ) and activates the lexical node ( e.g., word) that represents the target. However, othe r lexical nodes that are semantically related to the target (e.g., dog, rat ) are also activated, but typically not to

PAGE 18

18 the same extent as the target node. Thus, if the target node receives the highest activation level, it should be selected over the semant ically related lexical items (Funnell, 2002; Morsella & Miozzo, 2002). The current study examines oral and written naming abilities and has some repetition in the treatment, when needed. Thus, only the relevant components and routes of the Ellis and Young model will be discussed. The ability to name pictures and objects is reliant upon the visual object recognition system as well as the semantic system (Kay et al., 1992; Raymer & Gonzalez -Rothi, 2002). Given that there is no direct link from the visual ob ject recognition system to the phonological or orthographic output lexicons, the semantic system is required for naming. Specifically, spoken naming involves the following route: visual object recognition system semantic system pho nological output lexicon phonological output buffer. For written naming, a similar route is followed until the output lexicon: visual object recognition system semantic system orthographic output lexicon orthographic output buffer. A disruption at any of these levels can lead to difficulty in naming; however, careful analysis of a patients abilities/inabilities can enable a clinician or researcher to better localize the level of disruption. Repetition of regular and irregular words have a differe nt input lexicon but like oral naming ends with the phonological output lexicon: auditory phonological analysis phonological input buffer phonological input lexicon phonological output lexicon phonological output buffer. Bypassing the semantic sys tem allows repetition to occur without comprehension. On the other hand, non-words are repeated by a process of auditory phonological analysis acoustic -to -phonological conversion phonological output buffer. Two -s tep Lexical Access and Spreading Activ ation Models It is generally accepted that there are two levels of representation during word retrieval: the conversion of the semantic concept to the lexical node containing the appropriate syntactic

PAGE 19

19 properties (e.g., grammatical class) of the target wor d, and assignment of appropriate phonological properties of the selected node (e.g., the phonemes, syllable length, word structure) (Levelt, Schriefers, Vorberg, Meyer, Pechmann & Havinga 1991). Some models have proposed the existence of a lemma the repr esentation that specifies semantic and grammatical, but not phonological details of the target which mediates the conversion from a conceptual representation to a phonological representation (Dell, Schwartz, Martin, Saffran, Gagnon, 1997). Theories of lex ical retrieval based on spreading activation are divided into two categories: discrete serial activation and cascaded activation assumption. Discrete serial activation assumes information passes from one level of activation to another in one direction wit hout feedback to the previous level. Although multiple conceptual nodes are activated in the semantic system, only the selection of the target node will activate the phonological layer. Conversely, the cascaded activation assumption states that activatio n of multiple nodes in the semantic system will cause activation of corresponding phonological representations before the target lexical node is selected (Chialant et al., 2002). Furthermore, some cascade models support an interactive mechanism, which involves feedback, such as is described by Dell and colleagues (1997). According to Dell et al. (1997), lexical access involves three stages: semantic, lexical (e.g., lemma), and phoneme. When presented with a picture -naming task, the appropriate conceptual units in the semantic system connect with the corresponding word node, with the word node connecting with the related phoneme nodes. Following the notion of a cascaded interactive model, information flows both top-down, and bottom up, resulting in activat ion flowing not only from semantic word node phoneme, but also back up from phoneme word node semantic levels. It is this type of interacting activation that Dell et al. propose accounts for mixed errors (e.g., semantic + phonologic).

PAGE 20

20 According to Dell et al. (1997), the presentation of a picture of a cat activates conceptual units in the semantic system corresponding to features of a cat (e.g., domestic, walks on four legs ); however, units that share semantic features with cat are also activated such as dog (e.g., domestic, walks on four legs ) and rat (e.g., walks on four legs ). Mediation from conceptual to phonological representation occurs at the lemma level, where not only the lexical node corresponding to the activated semantic node is activated (e.g., cat ), but also other lexical nodes that are semantically related to cat are activated (e.g., dog, rat ). Phonological nodes corresponding to the lexical nodes are activated. This in turn sends feedback to the lemma/lexical level to activate no des that are phonologically similar to the current activated lexical nodes, such as mat to cat and fog to dog (Chialant et al. 2002). It is important to note that these phonologically activated lexical nodes do not have additional activation from the semantic level; however, phoneme nodes reactivate lexical nodes that have a semantic connection. Thus, the semantically activated nodes cat, dog, and rat also activate their representative phoneme nodes. These nodes in turn feed backward from the phoneme lev el to the lexical level and activate phonologically similar nodes (e.g., mat, fog ); however, the semantically related nodes (e.g., cat, dog, rat ) receive additional activation. The target cat should receive the highest activation, resulting in its selection for processing at the phonologic level; however, it is possible for the semantically related node rat, which is also phonologically similar to the target cat, to receive higher activation than the target, resulting in the mixed error rat Again, the most highly activated phoneme nodes are selected, and ordering follows the same process of selecting the highest activated onset, vowel, and coda (Dell et al., 1997). Errors in Aphasia Anomia, or the difficulty in retrieving words, is the most persistent ch aracteristic of aphasia. This difficulty can present itself in a variety of manners, including paraphasias and/or

PAGE 21

21 neologisms/jargon. Paraphasias may be semantic or phonemic in nature, with semantic paraphasia describing a word replacement that is semanti cally related to the target word and phonemic paraphasia representing a word that is minimally altered by a substitution, addition, omission, or rearrangement of the phonemes in the target word. Neologisms are nonwords. When content words within a sentenc e are replaced with neologisms, the utterance is defined as jargon (Murray & Clark, 2006; Dell et al., 1997). Since the Kay et al. (1992) model will primarily be used for interpretation of the results from this study, a brief introduction to levels of dam age and consequential impairments is necessary. Error patterns during a lexical task, such as picture naming, can provide clues as to the underlying mechanism responsible for the difficulty in naming (Dell et al., 1997; Raymer & GonzalezRothi, 2000, 2002). Viewing the semantic system as a complex mechanism, neurological injury can result in selective or generalized impairment of the system. According to Raymer and Gonzalez Rothi (2002), impairment at the semantic level will result in semantic errors bot h in naming and comprehension, with further classification of the errors to superordinate ( e.g., vegetable for carrot ), coordinate ( e.g., celery for carrot ), and associated ( e.g. rabbit for carrot ) categories of semantic error. Errors resulting from a tr ue semantic impairment should present themselves each time the system is accessed, regardless of the modality of input or output (Raymer & GonzalezRothi, 2000). Nonetheless, these authors note that semantic errors do not always indicate disruption at the level of the semantic system. For example, semantic errors occurring in response to oral naming and reading but absent in comprehension, written naming, and writing to dictation tasks supports a modality -specific disruption at the phonological output le xicon. Conversely, semantic errors in response to written tasks but absent in spoken tasks supports a disturbance at the level

PAGE 22

22 of the orthographic output lexicon (Aminoff, Boller, & Swaab, 2008). Therefore, careful analysis of error patterns is important, given that semantic errors alone do not support an impairment of the semantic system (Raymer and Gonzalez Rothi, 2000, 2002). In their discussion on models of naming, Chialant et al. (2002) also report that increased errors at one output modality as compa red to the other suggests damage at that level; therefore, if a greater number of errors occur during spoken naming as compared to written naming, disruption at the level of the phonological output lexicon can be assumed. With respect to the Kay et al. (1 992) model, a relatively intact semantic system with damage at either the level of the phonological output lexicon or to the connection between the semantic system and the phonological output lexicon, yet a preserved orthographic output lexicon and connect ion between the semantic system and the orthographic output lexicon may result in better written naming as compared to oral naming. The converse is also true, in that an intact semantic system with damage to either the orthographic output lexicon or the c onnections between the two with both a spared phonological output lexicon and connection between it and the semantic system will likely result in better spoken naming than written naming. According to Dell et al. (1997), lexical errors (e.g., semantic pa raphasias) are due to difficulty in retrieving the correct word node among semantic competitors (e.g., at the level of lemma access), whereas sublexical errors (e.g., phonemic paraphasias and neologisms) reveal difficulty in retrieving the correct phonemic nodes at the level of phonologic representation. Additionally, Dell et al. claim that mixed errors provide evidence for the simultaneous interaction of semantic and phonological level of processing. Dell and colleagues (1997) further categorized errors in lexical retrieval into five categories: semantically related words, phonologically related words, mixed, unrelated words,

PAGE 23

23 and nonwords/neologisms. Semantic word errors are a result of activated word nodes that are semantically related to the target, wi th the semantically related node receiving the highest level of activation needed for selection. Phonological word errors may occur due to an error at the lemma level or at the phonological level. Phonological errors at the lemma level are a result of ba ckward feedback from the phonologic level to the lexical level, with activation of phonologically similar word nodes to the target word. Phonological errors at the phoneme level occur after the selection of the correct word node for the target word, but due to activation of competing phonemes for other word nodes, a phoneme is incorrectly chosen for the target word. Mixed errors occur from a combination of semantic and phonologic errors. Forward and backward feedback between levels reinforces a semantica lly related word node that is phonologically similar to the target; in other words, the mixed error consists of shared semantics and shared phonemes that both receive activation (Chialant et al., 2002). Unrelated word errors occur at the lemma level as a result of activation of word nodes that are distantly related to the target word; however, difficulty at the phonologic level is plausible if the correct word node was selected but was subsequently coded with the incorrect phoneme. Finally, neologisms are strictly considered a disruption at the phonologic level as a consequence of other activated word nodes, resulting in the replacement of one or multiple phonemes of the target word (Dell et al., 1997). The prevalence of error types associated with aphasi a reinforces the importance of conducting error analyses in treatment research in order to capture evolution of errors or the change in predominant error types produced pre to post treatment. According to the literature, there is a predominant trend in evolution of errors following semantic treatment. Pre treatment, the dominant error categories were often neologisms, no responses or I dont know (NR/IDK),

PAGE 24

24 and general semantic. Post treatment, errors often evolved into primarily semantic and phonolog ical errors, revealing increased activation and access to semantic and phonological representations related to the target (Edmonds et al., 2009; Edmonds & Kiran, 2006; Kiran & Thompson, 2003; Kiran, 2008). Research Questions In Edmonds et al. (2009), VNeS T was provided to individuals with moderate aphasia for a relatively short period of time with fairly extensive generalization that was generally described. The purpose of the current study was to 1) examine the effect of VNeST on individuals with more sev ere aphasia and lexical retrieval impairments, 2) examine generalization patterns in detail, and 3) analyze the evolution of error patterns on lexical retrieval tasks from preto post treatment. It was predicted that participants would show generalization to picture description with trained and untrained verb networks, but to a lesser degree and in more time than the moderate participants. Generalization was hypothesized to improve on single word naming of nouns and verbs, with perhaps more improvement to nouns (Edmonds et al.). Improvement to sentence production containing verbs unrelated to treatment was uncertain given the severity of the participants. Errors were anticipated to evolve from those of omission to those of a semantic and/or phonologic natur e, revealing increased activation of representations closer to the target.

PAGE 25

25 Figure 1 1. Ellis and Young M odel (1988; as cited in Kay, Lesser, & Coltheart, 1992 ). [Reprinted with permission from Psychology Press: Taylor and Francis Group. Kay, J., Lesser, R., & Coltheart, M. (199 2 ). Psycholinguistic Assessments of Language Processing in Aphasia (PALPA): An introduction. (Page 15, Figure 9)].

PAGE 26

26 Figure 1 2. Dell Aphasia Model (as cited in Dell, G., Schwartz, M., Martin, N., Saffran, E., & Gagnon, D., 1997). [Reprinted with permission from American Psychological Association. Dell, G., Schwartz, M., Martin, N., Saffran, E., & Gagnon, D. (1997). Lexical Access in Aphasic and Nonaphasic Speakers. Psychological Review, 104 (4), 801838].

PAGE 27

27 CHAPTER 2 ME THODS Participants Two participants, P1 and P2, were recruited from the University of Florida Speech and Hearing Clinic for this study. Participants met several inclusion criteria, including: 1) diagnosis of aphasia, 2) monolingual English speaking, 3) r ight -handedness prior to stroke, 4) considerable lexical retrieval deficits, 5) negative history of diagnosed learning disorder or drug/alcohol addiction. P1 was a 42 -year old, right -handed female with 16 years of education. Forty-nine months before begin ning the study, P1 had a massive left hemisphere hemorrhagic stroke (MCA, ACA territory) involving the entire left frontal lobe that required a decompressive craniotomy. While she was hospitalized for the first stroke, she endured a second stroke in the ri ght hemisphere (minor, ischemic). Prior to her strokes, she was a web designer living independently. At the time of this study, she was living independently in a house a few blocks from her parents house. She received considerable support from her parent s. P1 had received physical, occupational and speech therapy previously, but she had discontinued all therapy for up to two years before starting this study. She decided to pursue speech therapy after two years because she decided she wanted to communicate better. P2 was a 49 -year old, right -handed female with 12 years of education. Nine months prior to beginning the study, P2 had a left hemisphere ischemic stroke (MCA) According to the neurology report, the stroke involved considerable cortical gr ay, gray -white interface, and deep gray matter that resulted in a significant right hemiparesis requiring the use of a wheelchair. Prior to her stroke, she was a choreographer/ma nager living independently. At the time of this study, she was living with her parents for assistance with activities of daily living (ADL).

PAGE 28

28 Although a depression screen was not administered, parental report and clinician observation suggested that she w as having considerable difficulty coping with the effects of the stroke. P2 had received previous speech therapy, but she had discontinued this therapy a few months prior to enrollment in this study. P2 decided to pursue speech therapy in order to improve her communication. Pre -testing Measures Both participants underwent several assessments to characterize aphasia type and severity and lexical retrieval impairments. P1 had an Aphasia Quotient (AQ) of 45.2 as classified by the Western Aphasia Battery (WAB Kertesz, 1982). P2 was also diagnosed with Brocas aphasia, with an AQ of 36.4 (s ee Table 2 2 for details). All assessment sessions were audio and/or videorecorded with patient consent to ensure consistent testing procedures as well as the accuracy of participant responses and examiner scoring. Single word lexical retrieval was evaluated with the Boston Naming Test (BNT ; Goodglass, et al.,1983) and An Object and Action Naming Battery ( An O&A Naming Battery ; Druks & Masterson, 2000). The BNT an assessment of single word lexical retrieval, consists of 60 pictures of objects decreasing in frequency (e.g. bed to abacus ). An O&A Naming Battery is a commercially available assessment that evaluates confrontational naming for bot h nouns ( N = 162) and verbs ( N = 100). The noun and verb stimuli are divided into two forms that are balanced for psycholinguistic (e.g., frequency, age of acquisition) and non-linguistic (e.g., imageability, visual complexity) stimuli related variables. T he stimuli also represent common semantic categories (e.g., animals, body parts). Lexical retrieval in sentences was also evaluated with the Northwestern Assessment of Verbs and Sentences (NAVS Thompson, 2002). The NAVS consists of 50 pictures of common situations with different people and animals depicting some action using one, two and three -

PAGE 29

29 place verbs (e.g. The dog is bark ing The boy is climbing a tree., The woman is putting the box on the shelf ). Contrary to the NAVS protocol which involves showing and reading the verb to the participant the verb was not shown to or heard by the participant s to better determine the lexical retrieval abilities Furthermore, the test shows the two and three place optional verbs in optional and obligatory contexts (e.g. the man is sweeping vs. the man is sweeping the dirt ) during testing but in our adaptation, only the more complex sentence was shown (e.g., each verb/sentence was only shown once). Thus, a total of 36 of 50 pictures were administered ( o ne place N = 8; t wo -place N = 17; t hree -place N = 11). Since lexical retrieva l was of primary interest, participants received credit if all required lexical items of the sentence were present regardless of word order or inflection of the verb. Sentence a ccuracy for P1 and P2 were 25% and 0%, respe ctively. Language measures r eliability. Inter -rater reliability was conducted by a trained Communication Sciences and Disorders undergraduate student on 33%50% of all pre to post treatment language measures. A point to point evaluation was conducted, and agreement in responses and scoring was 97%. Stimuli Development The e xplanation of stimuli development was taken from Edmonds et al. (2009) with permission from the first author. Sentence Elicitation Pictures Twenty -four pictures were developed for baseline and treatment probes. All sentences elicited an agent, verb, and patient. The agents and patients portrayed had specific titles in most cases (e.g., nurse carpenter ) to promote specific language use instead of use of generic words (e.g., lady m an). Sentences were divided into two sets (verb set 1 and verb set 2) Sets were

PAGE 30

30 created so that each verb in one set was semantically related to a verb in the other set (e.g., measure/weigh ). Control T ask A single word adjective retrieval task was developed as a non-verb control task that would rule out the possibility that improvements occurring during treatment reflected a nonspecific effect on semantic knowledge underlying concept representations. Participants were expected to complete a sentence by prov iding a synonym to the adjective provided in the sentence. For example, Someone who is sick is also said to be _____ (target is ill ). The adjectives were matched to the 24 verbs included in the probe stimuli on frequency See Edmonds et al. (2009) for details Experimental Design A multiple baseline approach across subjects was used to evaluate the effects of VNeST Three phases were institute d: 1) baseline, 2) treatment of trained items with administration of generalization and control probes in order to monitor treatment effects, 3) maintenance Prior to treatment and during the acquisition phase stable baselines for generalization and control probes were established To evaluate the effect of VNeST on sentence production, weekly probes of pictures depicting train ed verbs and thematic roles and semantically related verbs and thematic roles were administered (e.g., trained (carpenter -measure -stairs); untrained (nurse -weigh baby) (Edmonds et al., 2009). For P1, maintenance probes were administered 1 month and 5 months post treatment. For P2, three post treatment probes were administered immediately after treatment, and a maintenance probe was administered 1 month after treatment. P2 then enrolled in other treatment studies (for speech and gait) and so was not eligib le for additional maintenance testing.

PAGE 31

31 Procedures Baseline, Treatment, and Maintenance Probe Measures During baseline, treatment, and maintenance probe sessions, 20 pictures and the adjective control task were administered at the beginning of each session. All sessions were audio and/or video recorded with patient consent, and responses were transcribed phonetically. During administration, pictures were presented pseudo randomly with semantically related verbs (e.g. bake/fry ) in non -sequential order. For each picture, participants were instructed to Make a sentence and include him/her, the action, and this (while pointing to the agent ( carpenter ), verb (measure ), and patient ( stairs )). Prompts were generally not provide d unless the participant produced a general word for the target (e.g., cut instead of slice or man instead of carpenter ), in which she was then prompted to use a more specific word. Clarification was provided if the participant produced a n appropriate response that was not the intended target (e.g., The landscaper is mowing the grass instead of The landscaper is pushing the lawnmower ), with the clin ician pointing to the action or the intended target for elucidation These same prompts were given to the normal older group when the pictures were normed as well (Edmonds et al 2009). Given P2s severe spoken output deficits (e.g. jargon and unintelligible speech) with relatively preserved writing, weekly probes were administered in spoken and written modalit ies (on different days) in order to assess potential improvement in both modalities. She received the same cues for the written probes as she did for the spoken probes. For the control task, participants were asked to complete sentences using a synonym for the provided adjective (e.g., Someone who is sick is also said to be _____ (target : ill)) In the event of multiple attempts, the most appropriate adjective produced was scored. Responses consisting of one phonological error per lexical item w ere considered correct

PAGE 32

32 Treatment The t reatment protocol description was taken from Edmonds et al. (2009), with permission from the first author. Stimuli consisted of: 1) 10 cards containing the names of the 10 trained verbs (verb set 1) (e.g., measure ), 2) 6 8 cards for each verb containing 3 4 agents and 3 4 patients that form 3 4 pairs related to each verb (e.g., chef/sugar, carpenter/lumber, surveyor/land, designer/room for the verb measure ) chosen to represent a range of possibilities to maximally expand the variety of scenarios related to each verb, 3) 5 cards containing the following words ( who, what, where, when, why ), and 4) 12 sentences used for semantic judgment (heard but not seen by participants). The 12 sentences contained the target verb bro ken into four categories: a) correct ( The designer measures the room. ), b) inappropriate agent ( The infant measures the lumber. ), c) inappropriate patient ( The chef measures the television. ), d) thematic reversal ( The room measures the designer ). VNeST wa s administered two times per week for 2 -hour sessions. The first hour of the second session each week was dedicated to probes for P1. For P2, probes were presented during the beginning of each session (written or spoken). Participants performed five treatm ent steps that reinforced the semantic meaning of the target verb and encouraged multiple associations between the provided verb and appropriate agents and patients During treatment, participants were asked to produce orally 3 4 thematic role pairs (e.g., carpenter and lumber) for a provided verb (e.g., measure ). When they were unable to produce a word, written o ptions on cards were provided (some appropriate and some foils). Participants were encouraged to provide at least one personal pair, and the responses could change from week to week. After generating agents and patients appropriate to t he target verb, participants read each pair aloud and chose one for answer ing wh questions about it (e.g., when, where, why ).

PAGE 33

33 M odifications were made to accommodate P2s sparse and highly unintelligible/jargon spoken output. Following protocol, she was asked to orally produce thematic role pairs for a provided verb; however, when she was unable to provide an appropriate verbal response, she was allowed to write her response. Oral reading (or repetition with clinician modeling) of her correct written response always followed. This adaptation allowed P2 to produce her own words rather than being provided words by the clinician following an unintelligible response. Treatment was terminated wh en participants produced a minimum of 24 agent -patient pairs (80% accuracy) during treatment step one (e.g., for measure acceptable pairs would include chef/sugar, surveyor/land, designer/room ). Since there were 10 treatment items and 3 opportunities for agent/patient pairs, a total of 30 pairs in one week was possible. Thus, 24 correct pairs (80%) in one week met treatment termination criterion. P1 achieved criterion for ending treatment on week 9 P2 did not achieve criterion for treatment terminati on solely in the spoken modality; however, the provision of spoken and written responses produced in step one during treatment enabled P2 to meet the criterion on week 12 (Edmonds et al, 2009) Treatment r eliability In order to ensure consistency in ex ecution of the treatment protocol within and across participants, a trained Communication Sciences and Disorders undergraduate student observed 25% of the sessions live. The treatment protocol was followed wit h a reliability of 99% (determined by comparing the observed treatment steps to the written protocol). Scoring For probe responses, a point system was established and refined to reflect changes in lexical retrieval over time. This system was developed in order to be sensitive to ch anges in participants with severe lexical retrieval deficits. Even though the probe stimuli elicited sentences, the scoring system evaluated each content word (since entire sentences may not

PAGE 34

34 improve as in Edmonds et al. (2009) where the participants were m oderately impaired). Each content word was assigned a point value to reflect the accuracy and specificity of the response, with point values ranging from correct (1), half (), quarter (), to incorrect (0). Grammatical and/or morphological errors were not considered since these were not targeted in treatment. In the event of multiple attempts, the best response for the target was accepted (s ee Table 2 4 for example scoring protocol ). Scoring reliability. In order to ensure consistency in scoring within and across participants using the refined protocol, a trained Communication Sciences and Disorders undergraduate student completed inter rater reliability for 33% weekly probes. A point to point evaluation was conducted, and agreement in responses and scoring was 95%. Error Analysis An error analysis was conducted on pre to post treatmen t measures to determine trends in the errors made by each participant over the course of treatment. Each lexical item for the BNT and An O&A Naming Battery was assigned to only 1 of 13 possible categories according to the type of error made: Correct ( one phoneme/grapheme error allowed), Description, Semantic Error in the Same Grammatical Class, Semantic Error in a Different Grammatical Class (e.g., a semantically related verb for a target noun), Phoneme/Grapheme Error with Greater than or Eq ual to 50% of Phonemes/Graphemes of the Lexical Item Correct, Phoneme/Grapheme Error with Less than 50% of Phonemes/Graphemes of the Lexical Item Correct, Mixed Error (represents a combined semantic and phonologic error), Unrelated Error in the Sa me Grammatical Class, Unrelated Error in a Different Grammatical Class, Perseveration (e.g., repetition of a previous response, within the last five consecutive responses), Jargon (e.g., a nonsense word), I Dont Know or No Response (NR/IDK), or an Unint elligible response from the participant. In the results and discussion, and for Figures 3 8 through 3 1 0 semantic errors

PAGE 35

35 within the same grammatical class and description errors were combined and collectiv ely referred to as semantic errors within the same grammatical class (s ee Table 2 3 for examples of each error type). The error analysis was modified from the above -mentioned template to accommodate the sentence context of the NAVS and to determine tren ds in the errors as sentences increased in complexity from one place to three -place verbs. Errors in each sentence were assigned to categories reflecting the grammatical function: Subject, Verb, Direct Object, or Indirect Object (s ee Figures 3 8 through 3 10 for pre to post -treatment error analysis data ). Error analysis r eliability Inter rater reliability was conducted by a trained Speech Language Pathologist familiar with aphasic speech on 33% 50% of all error analyses for pre to post treatment language measures. A point -to -point evaluation was conducted, and agreement in responses and scoring was 96%.

PAGE 36

36 Table 2 1. Participant d emographic i nformation Pt Sex Education (years) Age Occupation Site of l esion Aphasia t ype MPO WAB AQ 1 F 16 42 Web Design Large Left Small Right CVAs Brocas 49 45.2 2 F 12 49 Choreographer Large L eft MCA CVA Brocas 9 36.4 Note: Pt: Participant, MCA: Middle Cerebral Artery; MPO: Months Post Onset; WAB AQ: Western Aphasia Battery Aphasia Quotient

PAGE 37

37 Ta ble 2 2. Pre and p ost -t reatment s cores for all a dministered t ests Participant 1 Participant 2 Spoken Spoken Written pre tx post tx pre tx post tx pre tx post tx APHASIA SEVERITY Western Aphasia Battery Aphasia Quotient (AQ) 45.2 55.5 36.4 48.1 N/A N/A Information 7 8 4 8 N/A N/A Fluency 2 5 2 2 N/A N/A Comprehension 7.0 7.55 8.4 8.75 N/A N/A Repetition 3.2 2.5 2.3 1.8 N/A N/A Naming 3.4 4.7 1.5 3.5 N/A N/A COGNITIVE SCREEN Cognitive Linguistic Qui ck Test (CLQT) Composite Severity 2.8 (Mild) 2 (Mod) 2.6 (Mild) 2.2 (Mod) N/A N/A Attention 189 (WNL) 114 (Mod) 188 (WNL) 178 (Mild) N/A N/A Memory 81 (Severe) 83 (Severe) 70 (Severe) 101 (Severe) N/A N/A Executive Functions 27 (WNL) 23 (Mild) 23 (Mild) 22 (Mild) N/A N/A Language 10 (Severe) 12 (Severe) 6.5 (Severe) 12 (Severe) N/A N/A Visuospatial Skills 98 (WNL) 75 (Mild) 92 (WNL) 80 (Mild) N/A N/A SEMANTIC PROCESSING Pyramids and Palm Trees Test ( N = 52) 45 (86.5%) 49 (94.2%) 49 (94.2%) 47 (90.4%) N/A N/A Kissing and Dancing Test ( N = 52) 48 (92.3%) 51 (98.1%) 47 (90.4%) 50 (96.2%) N/A N/A LEXICAL RETRIEVAL Boston Naming Test ( N = 60) 21 (35%) 22 (36.6%) 5 (8.3%) 16 (26.6%) N/A N/A An Object and Action Naming Battery Nouns ( N = 81) 37 (45.7%) 53 (65.4%) 13 (16%) 39 (48.1%) 63 (77.7%) 69 (85.2%) Verbs ( N = 50) 28 (56%) 25 (50%) 6 (12%) 21 (42%) 33 (66%) 43 (86%) Northwestern Assessment of Verbs and Sentences Sentence production w ithout verb provided ( N = 36) 9 (25%) 9 (25%) 0 (0%) 8 (22.2%) N/A N/A

PAGE 38

38 Table 2 3. Error a nalysis c ategories Error t ype Target Example D escription Melting Hot Semantic Same Gra mmatical Class Pencil Pen Semantic Different Grammatical Class Juggling Circus Phonologic: > 50% Phonemes Correct Banana /Anina/ Phonologic: < 50% Phonemes Correct Sheep /Cheet/ Mixed Phoneme or Grapheme Error Sewing Deedle Unrelated Same Grammatical Class Cow Tree Unrelated Different Grammatical Class Pushing Boy Perseveration Clown Circus* Jargon Helicopter /Hoponinee/ IDK/ NR Scissors I D ont Know Unintelligible/ Indiscernible Radio R..? Note: *A word previously answered duri ng the assessment.

PAGE 39

39 Table 2 4. Example p robe s coring Target Agent response (points awarded) Verb response (points awarded) Patient response (points awarded) Summary of p oints Type of e rror The nurse is weighing the baby. Nurse (1) Weigh (1) Baby (1) 1+1+1=3 No errors The carpenter is measuring the stairs. Carfenter (1) Measure (1) Stairs (1) 1+1+1=3 1 phonemic error The chef is baking the cookies. /Chet/ (.5) Bake (1) Cookies (1) .5+1+1=2.5 2 phonemic errors The chauffer is driving the limousine. Pilot (.5) Drive (1) Car (.5) .5+1+.5=2 Semantic or general category error The chauffer is driving the limousine. Man (.25) Drive (1) Limo (1) .25+1+1=2.25 General noun The tailor is sewing the suit. He (0) Sew (1) /chuperdon/ (0) 0+1+0 =1 Pronouns, neologisms

PAGE 40

40 CHAPTER 3 RESULTS Probes For both participants stable, non -rising baseline s w ere established. P1 had 15 treatment baselines and met criteria for ending treatment on week 9. P2 had 15 treatment baselines, but did not meet crit eria for ending treatment solely in the spoken modality; however when paired with spoken and written responses in treatment, P2 met criteria for ending treatment on week 12. P1 did not complete post treatment probes immediately after treatment. Rather, there are maintenance probes of 1 and 5 months post -treatment. P2 had three post treatment probes as well as a 1 month maintenance probe. As mentioned previously, P2 enrolled in other treatment studies (e.g., gait and speech) following conclusion of the current treatment study, and was therefore unavailable for further maintenance testing. To evaluate whether there was significant improvement from the baseline to the acquisition phase of weekly probes, a time series C -statistic was used (Tryon, 198 2). Effect sizes were calculated to determine the magnitude of changes from baseline to post treatment by 1) determining the average of baseline points and the average of post treatment points 2) calculating the difference between the averaged post treatment and baseline points and 3) deriving the standard deviation of baseline points To determine the effect size (e.g., Cohens d ), the difference from post treatment to baseline was divided by the standard deviation of the baseline (Dollaghan 2007). Participant 1 (P1) For P1, impro vements on picture naming for trained verb networks from baseline to acquisition were significant ( C = .867, p = .001). Improvements on untrained networks from baseline to acquisition were also significant ( C = .852, p = .001). The effect size of pre to post -

PAGE 41

41 treatment trained networks was 6.54. The effect size for preto post -treatment untrained networks was 5.06. No significant change baseline to treatment phase was found in the control task ( C = .236, p = .133). The control task effect size was 1.55. See Figures 31 through 3 2. Since individual words were scored on the probes (rather than entire sentences), an analysis of items that improved was conducted. P1 accurately retrieved 43% trained and 27% untrained agents, 28% trained and 24% untrai ned verbs, and 75% trained and 48% untrained patients. An average of maintenance points revealed improvement in P1s ability to retrieve agents (trained: 84%, untrained: 60%), verbs (trained: 58%, untrained: 45%), and patients (trained: 100%, untrained: 80%) P1s accuracy in producing complete sentences on probes (e.g., containing an agent, verb, and patient) was also determined. Pre treatment, 3% of the sentences contained all three content words whereas post treatment, 25% were comp lete. Participant 2 (P2) For P2, statistical analyses were conducted on responses for both the oral and written modalities. Trained networks in the oral modality improved significantly from baseline to acquisition ( C = .756, p = .001). Untrained network s in the oral modality improved significantly from baseline to acquisition ( C = .653, p = .001). The effect size of pre to post treatment trained networks was 11.30. The effect size for preto post treatment untrained networks was 5.42. Improvement s on trained networks in the written modality from baseline to acquisition were significant ( C = .658, p = .001). Improvements on untrained networks in the written modality from baseline to acquisition were also significant ( C = .615, p = .00 1 ). The eff ect size of pre to post treatment trained networks in the written modality was 15.81. The effect size for pre to post treatment untrained networks in the written modality was 3.46. No significant change was found in the control task ( C = .242, p = .126). The effect size was .22. See Figures 3 3 through 3 5.

PAGE 42

42 Since individual words were scored on the probes (rather than entire sentences), an analysi s of items that improved was conducted. In the spoken modality, P2 accurately retrieved 14% trained and 22% untrained agents, 2% trained and 12% untrained verbs, and 39% trained and 14% untrained patients. An average of post -treatment points revealed impr ovement in P2s ability to retrieve agents (trained: 53%, untrained: 55%), verbs (trained: 55%, untrained: 43%), and patients (trained: 75%, untrained: 33%) P2s accuracy in producing complete sentences on probes (e.g., containing an a gent, verb, and patient) was also determined. Pre treatment, 0% of the sentences contained all three content words whereas post treatment, 12% were complete. Analysis of items in the written modality revealed P2 accurately retrieved 64% trained and 75% un trained agents, 40% trained and 51% untrained verbs, and 78% trained and untrained patients. An average of post -treatment point revealed improvement in her ability to retrieve agents (trained: 100%, untrained: 85%), verbs (trained: 67%, untrained: 63%), a nd patients (trained: 97%, untrained: 85%). P2s accuracy in producing complete sentences on written probes (e.g., containing an agent, verb, and patient) was also determined. Pre treatment, 25% of the sentences contained all three content words whereas post treatment, 57% were complete. Pre and Post -treatme nt Lexical Retrieval Measures: Accuracy The accuracy results for each participant will be discussed sequentially. See Table 2 2 and Figure s 3 6 through 3 7 for pre and post treatment language measures results Participant 1 (P1) Accuracy P1 improved in her ability to retrieve nouns, as reflected by percentage accuracy on An O&A Naming Battery. Performance on the BNT pre to post -treatment remained relatively constant (35% to 36.6%) while performance on An O&A Naming Battery improved from 45. 7% to 65.4%. However, no improvement in single verb naming was observed on An O&A Naming Battery, with pre to post -treatment results of 56% and 50% correct, respect ively. Noun and verb

PAGE 43

43 retrieval within a sentence context was assessed using the NAVS. Overall, P1s ability to produce sentences of varying complexity on the NAVS remained unchanged pre to post treatment, with an accuracy of 25 %. Even though there was no change in accuracy, the ability to retrieve nouns and verbs within sentences was assessed for additional information (see P1 error analysis for detailed account of performance on the NAVS ). See Figure 3 6. Participant 2 (P2) Accuracy In the spoken modality, P2 improved in her ability to retrieve nouns with percent accuracy on the BNT increasing from 8.3 % to 26.6 %, and on An O&A Naming Battery from 16% to 48.1%. Verb retrieval improved from 12% to 42% correct. P2s overall ability to produce sentences of varying complexity on the NAVS im proved pre to post -treatment from 0% to 22.2% correct. In the written modality, noun retrieval improved on the An O&A Naming Battery from 77.7% to 85. 2 % accuracy, with verb retrieval also improving on the An O&A Naming Battery fr om 66% to 86% accuracy. See Figure 37. Pre and Post -treatme nt Lexical Retrieval Measures: Error Analysis Results of the error analysis conducted on pre to post -treatment lexical retrieval measures are represented in Figures 3 8 through 3 10. Partici pant 1 (P1) Error Analysis Nouns BNT. Pre treatment, the majority of errors were semantic (48.7 %), with most errors occurring in the same grammatical class (e.g., nouns). Following semantic errors, the next most prominent type of error was NR/IDK (20.5%). Unrelated errors of the same grammatical class occurred 1 2.8 % of the time, while jargon and phonologic errors both occurred 7. 7 % of the time. Within the class of phonologic errors, the majority of errors were confined to the category of > 50% of phonemes produced correctly. Mixed errors occurred the least (2. 6 %). Post -treatment,

PAGE 44

44 semantic errors within the same grammatical class remained the predominant type of error produced (50%), followed by unrelated errors in the same grammatical class (18.4%). Phonologic errors occurred 18.4% of the time, with most errors falling within the category of > 50% of phonemes correct. These percentages reflect an increase in these three error categories pre to post treatment. However, there was a reduct ion in NR/IDK (10.5 %). Mixed errors remained relatively unchanged at 2.6 %. An O&A Naming Battery. Pre treatment, the majority of P1s errors were characterized as semantic (54.5 %), falling within the same grammatical class (e.g., nouns). This was f ollowed by NR/IDK (20. 5%) and unrelated errors of the same grammatical class (11. 4 %). Jargon errors occurred 6.8 %, while phonologic errors of > than 50% of phonemes correct were produced 4.5 % of the time. Occurring the least amount were errors of a mixed nature (2. 3 %). Following treatment, semantic errors of the same grammatical class (75%) remained the highest error category. There was an increase in phonological errors (14. 3 %), with the majority of errors continuing in the category of > 50% o f phonemes produced correctly. There was a reduction in unrelated errors of the same grammatical class (7.1 %), with a slight increase in errors of a mixed nature (3. 6 %). Verbs An O&A Naming Battery. Prior to treatment, the majority of P1s errors fe ll into the category of NR/IDK (40.9%). This was followed by the category of unrelated errors (27. 3 %), with errors occurring equally between grammatical classes (e.g., both nouns and verbs). The next highest error type was semantic (18. 2 %), with err ors again occurring equally between grammatical classes. The final categories of errors produced were jargon (9. 1 %) and mixed (4.5 %). Following treatment, there was a reduction of NR/IDK errors to 0%, replaced by errors of a semantic (56%) nature with errors of the same grammatical class (e.g., verbs) produced

PAGE 45

45 slightly more than errors of a different grammatical class. The next most predominant error type was unrelated (28%), with most errors occurring in a different grammatical class (e.g., nouns). Mixed errors proportionally increased, occurring 12% of the time. Although there was a reduction of jargon errors to 0%, phonological errors with > 50% of phonemes produced correctly increased to 4%. NAVS: Sentences. Nouns decreased from 100% to 94. 4 % correct, while the ability to retrieve verbs improved from 34% to 36.1 % correct. Additional assessment of the ability to retrieve direct and indirect objects of the sentence revealed improvement in the ability to retrieve direct objects (60.7% to 64. 3 %) while the ability to retrieve the indirect object remained unchanged (72.7% correct). See Figure 3 8 for visual representation of P1s error analysis for all measures. Participant 2 (P2) Error Analysis Nouns BNT (Spoken). Pre treatment, the majorit y of P2s errors were jargon (41.8 %), followed by unintelligible responses (25.5% ). The next most frequently produced error was unrelated (16. 4 %), with the majority of these errors falling into same grammatical class (e.g., nouns). Phonologic errors with > 50% of phonemes correct were produced 10.9% of the time. The final categories of errors produced were NR/IDK (3.6 %) and semantic same grammatical class (1.8 %). Post treatment, errors in the jargon category increased to 72.7%. Errors in the phonologic category also increased (13.6%), with errors occurring equally between both categories (e.g., > 50% of phonemes correct, < 50% of phonemes correct). There was a reduction in errors for unrelated same grammatical class (4.5 %) and NR/IDK (2. 3 %), with no unrelated errors in a different grammatical class post -treatment. Semantic errors in the same and different grammatical classes along with mixed errors each occurred 2. 3 % of the time.

PAGE 46

46 An O&A Naming Battery (Spoken). Prior to treatment, t he predominant error type was jargon (33.8%) followed by unintelligible responses (26. 5 %). NR/IDK errors were produced 16. 2 % of the time, while phonologic errors were produced 13.2 %, with slightly more errors in the < 50% of phonemes produced correc tly category. Unrelated errors (7. 4 %) in the same grammatical class were produced twice as often as errors in a different grammatical class (e.g., verbs). Mixed errors were produced 2.9 % of the time. Following treatment, jargon errors increased to 61 .9 %, while phonologic errors increased to 33.3%, with most errors occurring in the > 50% of phonemes produced correctly category. Semantic and unrelated errors within the same grammatical category, along with mixed errors were each produced 2. 4 % of th e time. An O&A Naming Battery (Written). Before treatment began, P2 produced primarily semantic same grammatical class errors (33.3 %), followed by graphemic errors (27. 8 %), with slightly more errors characterized as > 50% graphemes produced correctly. This was followed by and NR/IDK errors (22.2 %). Unrelated same grammatical class errors occurred 11.1 % of the time, followed by mixed errors (5. 6 %). After treatment, errors were predominantly > 50% of graphemes correct (41. 7 %) and semantic (41. 7 %) with more errors in the same grammatical category (e.g., nouns), as compared to the alternate grammatical category. Unrelated errors in the same grammatical category and NR/IDK were each produced 8.3 % of the time. Verbs An O&A Naming Battery (Spoken). Pre -treatment, P2s predominant error type was NR/IDK (54. 6 %) followed by unintelligible responses (20.5%). Jargon and phonologic errors were each produced 6.8 % of the time, with twice as many errors occurring in the > 50% of phonemes produced correctly category as compared to <50% of phonemes produced correctly. Unrelated errors in a different grammatical class (e.g., nouns) and semantic errors occurring equally between grammatical categories were each produced 4.5 % of the time. Errors of a

PAGE 47

47 mixed nature occurred the least (2. 3 %). Post treatment, NR/IDK errors were reduced to 0%, while jargon errors increased to the most predominant error type (51.7 %). Phonologic errors increased to 20. 7 %, with errors in the > 50% of phonemes produced co rrectly category continuing to occur twice as often as those in the <50% of phonemes produced correctly category. Unrelated errors (13. 8 %) occurred mostly in a different grammatical class (e.g., nouns), followed by mixed errors (6 .9%). Semantic e rrors in a same and different grammatical class each occurred 3.4 % of the time. An O&A Naming Battery (Written). Prior to treatment, most errors were semantic (52.9%) in nature, with slightly more errors occurring in a different grammatical class (e.g., nouns). This was followed by unrelated errors (35. 3 %), with almost all of the errors occurring in the different grammatical class, and NR/IDK errors (11. 8 %). Following treatment, the majority of errors were semantic (57.1 %), primarily occurring in a different grammatical class. NR/IDK, jargon, and unrelated different grammatical class errors were each produced 14. 3 % of the time. NAVS: Sentences. Further analysis revealed that the ability to retrieve nouns improved from 55.6% to 88. 9%, whil e the ability to retrieve verbs improved from 0% to 38.9 % correct. P2s ability to retrieve the direct object of a sentence improved from 21.4% to 60.7 % correct; however, her ability to retrieve the indirect object remained unchanged, with 9.1% acc uracy. See Figures 3 9 through 3 10 for visual representation of P2s error analysis on all measures.

PAGE 48

48 0 5 10 15 20 25 30 B1 B2 B3 B4 B5 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 M1 M2 P1 Probes # of Words Trained Untrained TR: Baseline + Treatment = C= .867, p =.001 d = 6.54 UT: Baseline + Treatment = C= .852, p =.001 d = 5.06 Figure 3 1. P1s A gent, v erb, and p atient p roduction within s entences 0 2 4 6 8 10 B1 B2 B3 B4 B5 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 M1 M2 P1 Adjective Control Task # of Words Baseline + Treatment = C= .236, p =.133 d = 1.55 Figure 3 2. P1 s A djective c ontrol t ask

PAGE 49

49 0 5 10 15 20 25 30 B1 B2 B3 B4 B5 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 Post1 Post2 Post3 M 1 P2 Spoken Probes # of Words Trained Untrained TR: Baseline + Treatment = C= .756, p =.001 UT: Baseline + Treatment = C= .653, p =.001 d = 11.30 d = 5.42 Figure 3 3. P2s S poken a gent, v erb, and p atient p roduction within s entence s 0 5 10 15 20 25 30 B1 B2 B3 B4 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 Post 1 Post 2 Post3 M1 P2 Written Probes # of Words Trained Untrained TR: Baseline + Treatment = C= .658, p =.001 d = 15.81 UT: Baseline + Treatment = C= .615, p =.001 d = 3.46 Figure 3 4. P2s W ritten a gent, v erb, and p atient p roduction withi n s entences

PAGE 50

50 0 2 4 6 8 10 B1 B2 B3 B4 B5 P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 P11 P12 P13 P14 P15 Post1 Post2 Post3 M1 P2 Adjective Control Task # of Words Baseline + Treatment = C= .242, p =.126 d = .22 Figure 3 5. P2s A djective c ontrol t ask.

PAGE 51

51 0 20 40 60 80 100 P1 Pre-Post Aphasia Severity and Semantic Processing (Accuracy) AQ and Percent Correct Pre Post Pre 45.2 86.5 92.3 Post 55.5 94.2 98.1 WAB AQ P&P K&D 0 20 40 60 80 100 P1 Pre-Post Changes in Lexical Retrieval (Accuracy) Percent Correct Pre Post Pre 35 45.7 56 25 Post 36.6 65.4 50 25 BNT O&A Nouns O&A Verbs NAVS Figure 3 6. Visual r epresentation of P1s a ccuracy across m easures

PAGE 52

52 0 20 40 60 80 100 AQ and Percent Correct Pre Post Pre 36.4 94.2 90.4 Post 48.1 90.4 96.2 WAB AQ P&P K&D 0 20 40 6080 100 Percent Correct Pre Post Pre 8.3 16 12 77.7 66 0 Post 26.6 48.1 42 85.2 86 22.2 BNT O&A Nouns O&A Verbs O&A Nouns O&A Verbs NAVS Figure 3 7. Visual r epresentation of P2s a ccuracy across m easures.

PAGE 53

53 0 20 40 60 80 100 Semantic Same + Semantic Different Phonologi cal, >50% Phonologi cal, <50% Mixed Unrelated Same Jargon NR/IDK P1 BNT (Nouns) Percent of Errors Pre Post A 0 20 40 60 80 100 Semantic Same + Description Phonological, >50% Correct Phonological, <50% Correct Mixed Unrelated Same Jargon NR/IDK P1 O&A (Nouns) Percent of Errors Pre Post B Figure 3 8 P1s E rror a nalysis g raphs. A) Error a nalysis of P1s r esponse s on the BNT. B) Error a nalysis of P1s r esponses on An O&A Naming Battery ( n ouns). C) Error a nalysis of P1s responses on An O&A Naming Battery ( v erbs). D) Error a nalysis of P1s responses on the NAVS.

PAGE 54

54 0 20 40 60 80 100 Semantic Same + Description Semantic Different Phonological >50% Correct Mixed Unrelated Same Unrelated Different Jargon NR/IDK P1 O&A (Verbs) Percent of Errors Pre Post C 0 20 40 60 80 100 Subjects Verbs Direct Objects Indirect Objects P1 NAVS (Sentences) Percent of Errors Pre Post D Figure 3 8 Continued

PAGE 55

55 0 20 40 60 80 100 Semantic Same + Semantic Different Phonological, >50% Phonological, <50% Mixed Unrelated Same Unrelated Different Jargon Unintelligible NR/IDK P2 BNT (Nouns) Percent of Errors Pre Post A 0 20 40 60 80 100 Semantic Same + Phonologica l, >50% Phonologica l, <50% Mixed Unrelated Same Unrelated Different Jargon Unintelligibl e NR/IDK P2 O&A (Nouns) Percent of Errors Pre Post B Figure 3 9 P2s Spoken e rror a nalysis g raphs. A) Error a nalysis of P2s responses on the BNT. B) Error a nalysis of P2s r esponses on An O&A Naming Battery ( n ouns). C) Error a nalysis of P2s r esponses on An O&A Naming Battery ( v erbs). D) Error a nalysis of P2s responses on the NAVS.

PAGE 56

56 0 20 40 60 80 100 Semantic Same + Semantic Different Phonological, >50% Phonological, <50% Mixed Unrelated Same Unrelated Different Jargon Unintelligible NR/IDK P2 O&A (Verbs) Percent of Errors Pre Post C 0 20 40 60 80 100 Nouns Verbs Direct Objects Indirect Objects P2 NAVS (Sentences) Percent of Errors Pre Post D Figure 3 9 Continued

PAGE 57

57 0 20 40 60 80 100 Semantic Same + Description Semantic Different Orthographic, >50% Correct Orthographic, <50% Correct Mixed Unrelated Same NR/IDK P2 O&A (Nouns) Percent of Errors Pre Post A 0 20 40 60 80 100 Semantic Same + Description Semantic Different Unrelated Same Unrelated Different Jargon NR/IDK P2 O&A Verbs Percent of Errors Pre Post B Figure 3 10. P2s Written e rror a nalysis g raphs. A) Error a nalysis of P2s r esponses on An O&A Naming Battery ( n ouns). B) Error a nalysis of P2s r esponses on An O&A Naming Battery ( v erbs).

PAGE 58

58 CHAPTER 4 DISCUSSION The purpose of this study was to examine the effect of VNeST on individuals with more severe aphasia and lexical retrieval impairments than those who participated in the first study. E xamination of generalization patterns and analysis of errors for determination of error pattern evolution on lexical retrieval tasks from pre to post -treatment were also conducted Evolution of errors provide s additional evidence for le vel of impairment for interpretation within lexical retrieval models, primarily the Ellis and Young model as adapted by Kay, Lesser and Coltheart (1992) Level of impairment for each participant will be briefly discussed fo llowing the discussion of each participants results. Consideration of Participant 1 (P1) Results Probes For P1, both trained and untrained verb networks improved significantly from the baseline to the treatment phase. Further, effect sizes for pre treatm ent to maintenance scores for sentence production for trained and untrained verb network s were very large ( Beeson & Robey, 2006). Thus, treatment resulted in generalization to retrieval of content words within sentences for both trained and untrained verb networks in untrained tasks (e.g., picture description), and that improvement was maintained five months post treatment. This general pattern is consistent with Edmonds et al. (2009). However, as predict ed, the overall improvement in percent of sentences produced completely (e.g., containing an agent, verb, and patient) was less than the first study, as there was a 22 percentage point improvement on sentence production for P1 as com pared to margins of 40 to 80 percentage point improvements in the 4 less impaired participants in Edmonds et al.

PAGE 59

59 E xamination of individual agent, verb, and patient retrieval revealed that P1 was more accurate in noun retri eval than verb retrieval during baseline probes (e.g., pre treatment) From pre to post treatment, all components of the trained and untrained sentences improved from 21 to 41 percentage points over baseline, with untrained verbs improving the least and t rained agents improving the most. Thus, there was not a large disparity in improvement of trained and untrained items, which was also observed in Edmonds et al. (2009). However, verb retrieval remained relatively weaker than noun retrieval. Additionally, n oun retrieval abilities were different across agents and patients pre and post -treatment. Specifically, P1 retrieved more patients as compared to agents. This disparity in accuracy may be attributed to specificity. Patients were typically gen eral objects that were more familiar (e.g., baby, package, airplane ), whereas agents were more specific and specialized professions (e.g., pilot, chauffeur ). Closer inspection of improvement trends revealed patients improved first, followed by agents and then verbs. Agents and patients were relatively consistent in their improvement trend, while verbs improved to a lesser extent This trend may be due to the treatment paradigm, which encouraged more retrieval and repetition of nouns than verbs. However, it is also possible that noun retrieval of patients and agents assisted in activation of the verb over the course of treatment. While this may be contrary to traditional thought where verbs support noun retrieval, priming paradigms (e.g., Edmonds & Mizrahi (in preparation); Ferretti et al. (2001), and McRae et al. (2005)) have s hown bi -directional priming, such that verbs prime related nouns and nouns also prime verbs. Although the treatment is called verb strengthening treatment, it focuses its direction from nouns inward to the verb rather than working from the verb out towa rd the nouns. Therefore, it is plausible that treatment reinforced P1s pre treatment strength (e.g., nouns) and resulted in greater post -treatment accuracy of nouns. Thus, due to a low number of correct

PAGE 60

60 sentences and continued difficulty in verb retri eval for P1 more attention to verb retrieval in the treatment paradigm may be warranted. Pre -treatment, P1s content words for probes were of decreased semantic weight (e.g., man/woman, boy/girl ). Most of her errors were semantic, primarily within the sa me grammatical class, as well as some phonological errors that were close to the target ( > 50% phonemes correct). Post treatment, her responses were much more specific (e.g., instead of man she produced bartender ). Her most prominent type of error remain ed semantic, and within the same grammatical class as the target. Most responses that differed in grammatical class from the target were the result of noun retrieval (more specifically, an instrument) for a verb (e.g., sew needle ). For an example of P1s pre to post -treatment change in semantic specificity, the target pilot fly plane was produced as man drivin plane at pre treatment and as pilot race chet (jet) post treatment. Her phonological errors after treatment were few, with most containing one or two phoneme s different than the target. Pre to Post -treatment Language Measures P1 exhibited relatively no change in accuracy on nouns when assessed via the BNT ; however, she did exhibit an improvement of 20 percentage points on nouns t ested through An O&A Naming Battery A preliminary explanation for this discrepancy in noun retrieval may be attributed to the organization of the BNT from items of high frequency to low frequency. Thus, P1s relatively static performance on the BN T may reflect difficulty in lexical retrieval of objects (e.g., nouns) that occur less frequently in the language. Had P1 only been examined with the BNT it would have been concluded that she did not improve on nouns, when in fact she exhibited a large improvement in noun retrieval as reflected by An O&A Naming Battery. Error trends on An O&A Naming Battery reveal ed that P1 made more attempts at naming with a significant reduction and/or elimination of jargon and IDK/NR responses. The majority of

PAGE 61

61 post treatment errors were semantic errors followed by she was not able to retrieve correctly were much closer to the target as compared to pre treatment, which indicates improved lexical retrieval. In contrast to noun accuracy, P1s accuracy in naming verbs on An O&A Naming Battery declined slightly from pre to post -treatment. However, there was an evolution of her errors that revealed increased processing. Prior to treatment, the majority of errors were NR/IDK, follo wed by unrelated and semantic errors (with equal errors between grammatical classes), as well as jargon. Following treatment, there was a large increase in the proportion of semantic errors (more errors within the same grammatical class), with an elimination of jargon and NR/IDK responses, which suggests not only increased attempts at naming, but also increased processing despite not gaining in accuracy. Increased proportion of unrelated errors in a different gra mmatical class (e.g., nouns) may have been a consequence of retrieval and repetition of primarily nouns in treatment. Phonological errors also increased, which provides further support that errors were clo ser to the target following treatment. Thus noun and verb errors in single word naming evolved pre to post treatment with comparable trends of 1) reduction and/or elimination of NR/IDK and jargon errors that predominated pre treatment response s, and 2) primarily semantic and phonologic errors (within a few phonemes of the target) post treatment, revealing a higher level of processing with responses closer to the target. The pattern of noun retrieval improving more than verb retrieval was also observed in Edmonds et al. (2009) in some participants. However, no improvement in verb naming accuracy was observed in P1 in the current study despite generalization to verb retrieval in the probes. There are a number of reasons why this might have occurred. First, the verbs in the probes were semantically related to the treatment verbs, and thus would have received activation during

PAGE 62

62 treatment and thus improved their likelihood of retrieval. Second, this participant exhibits more impairment in verb r etrieval than noun retrieval, so perhaps more verb retrieval should be required in treatment to be in balance with nouns, which are retrieved more often and are given more feedback. However, it is important to remember that there was an evolution of lexica l retrieval ability in verbs as evidenced by the error analysis. Thus, the importance of error analysis in aphasia research is evident given that judge d solely on accuracy P1 did not improve on verbs. On the NAVS sentences assessment, P1s a ccuracy remained constant (at 25% correct) from pre to post treatment. Closer analysis of retrieval accuracy for each element of the sentence (e.g., subject, verb, direct/indirect objects) showed that pre treatment, P1 retrieved m ore subjects than objects, which contradicts probe findings pre -treatment (where patients obtained higher accuracy than agents). However, the subjects on the NAVS are general and of low semantic weight (e.g., man/boy ) as compared to probe a gents, which were very specific and specialized professions (e.g., chauffeur ). The finding from this closer analysis supports the presumption from probes that while agents and patients are both nouns, the patients were retrieved with higher accuracy becaus e they were more general and less specialized than the agents. Likewise, although subjects were retrieved with higher accuracy as compared to objects on the NAVS these subjects lacked semantic specificity. Similar to An O&A Naming Battery, there was rela tively no improvement in verb retrieval on the NAVS, as reflected by pre to post accuracy scores of 34% and 36.1% ( N = 1). Similarly P1s overall ability to produce correct sentences on the NAVS remained static preto post treatment at 25% of s entences correct. Closer analysis revealed that three sentences correct pre treatment became incorrect post treatment as a result of a semantically related verb (e.g., pinch

PAGE 63

63 bite ). Nonetheless, t hree sentences improved post -treatment as a result of retrieving a correct verb instead of a semantically related counterpart in pre -treatment (e.g., pull push). Analysis of errors on the NAVS revealed that pre to post treatment, jargon was eliminated and there was a reduction in NR/IDK respo nses. The reduction of NR/IDK reveals increased attempts at naming, even though accuracy did not capture this trend. Semantic and unrelated errors were proportionate pre to post treatment, with their presence post -treatment consistent with P1s other er ror trends. The lack of improvement (in accuracy) in both verbs and sentences on the NAVS further implicates that improvement in verbs is necessary for improvement in sentences, although retrieval of the verb does not necessarily ensure a correct sentence since P1 named approximately 50% of probe verbs but only 25% of probe sentences were completely correct. This overall trend is observed by comparing pre to post treatment performance within the probes and also to the NAVS Overall, nouns improved more than verbs on probes; however, trained nouns and verbs improved more than untrained nouns and verbs. Trained verbs improved 30 percentage points pre to post treatment, untrained verbs improved 21 percentage points, while verbs on the NAVS improved 3 per centage points. Therefore, generalization to a different task such as the NAVS was not achieved, presumably because there was not enough activation of verbs. Treatment for P1 focused on activation, retrieval, and repetition of nouns associated with the provided verb and not the actual verb. For example, in response to who can fly something and what can be flown, P1s correct retrieval of pilot and plane resulted in repetition of pilot and plane only, not fly Thus, the finding is consistent: P1s verbs may have required more activation in treatment before generalization could be achieved in pre to post treatment verb related assessments.

PAGE 64

64 More noun retrieval is an intentional component of VNeST, as the idea is to activate multipl e event schemas/semantic networks surrounding the target verb. The theoretical principle, then, is that the verb is being activated systematically and multi-dimensionally with the retrieval of multiple related noun pairs (Edmonds et al. (2009). However, w ith the relative verb retrieval difficulties noted in P1, an attempt to have P2 produce the target verb along with every noun pair was instituted. Thus, it was a small adaptation to protocol, but one that would theoretically aid generalization to verb retr ieval. Consideration of P articipant 2 (P2) Results Probes (Spoken) For P2, both trained and untrained verb networks improved significantly from the baseline to treatment phase, with a large magnitude of pre to post treatment change in both item sets as d etermined by effect sizes ( Beeson & Robey, 2006). Thus, treatment resulted in generalization to retrieval of content words within sentences for both trained and untrained verb networks in untrained tasks (e.g., picture description), thou gh untrained items were predictably lower. Maintenance scores were only conducted 1 month post treatment, and untrained items appeared to be maintained somewhat better than trained items, although the trained items were still considerably above baseline. Following predictions and consistent with P1 s performance, the overall improvement in percent of sentences produced completely (e.g., containing an agent, verb, and patient) was less than the first study, as there was a 12 percentage point improvement on sentence production for P2. Analysis of individual agent, verb, and patient retrieval revealed that P2 was more accurate in noun retrieval than verb retrieval during baseline probes. From pre to post -treatment, all elements of the trained and untrain ed sentences improved from 19 to 53 percentage points over baseline levels with untrained patients improving the least and trained verbs improving the most.

PAGE 65

65 Although trained verbs improved the most, they were the least accurate of all the elements pre tr eatment and therefore had a greater opportunity for improvement. Post -treatment, P2 retrieved approximately half of all agents, verbs, and patients, with slightly greater accuracy for agents and patients. Patients were retrieved with higher accuracy pretreatment whereas both agents and patients were retrieved equally post treatment T his trend differs from P1 who not only exhibited a greater discrepancy between nouns and verbs, but also retrieved patients more accurately than agents. Post -treatment accuracy of complete sentences on probes (e.g., containing an agent, verb, and patient) improved 12 percentage points over baseline levels, yet P2 retrieved nearly half of the verbs in post -treatment probes. While P2 improved in both single verb retrieva l and in complete sentences, the discrepancy in accuracy implies that retrieval of the verb does not guarantee a correct sentence. This trend of retrieving approximately half of the verbs in probes but only formulating a few complete sentences was also observed in P1 Closer inspection of improvement trends in the spoken modality for trained networks revealed that acquisition of patients and agents was variable, with patients improving earlier and to a greater extent than ag ents. This increased accuracy for retrieval of patients further supports the theory that these items are less specific and therefore easier to retrieve than the agents. Improvement trends for untrained networks revealed that agen ts improved initially, while acquisition of patients and verbs was variable. One explanation for agents improving prior to patients would be the theory of atypicality, where training a less typical item (e.g., penguin) results in generalization to typical items (e.g., robin) (Kiran & Thompson, 2003) Thus, if trained agents were more atypical, generalization to untrained agents that were more typical would not be unexpected. There also appeared to be a trend that when agents improved patients

PAGE 66

66 declined. This trend, coupled with P2s large improvement in verb retrieval but low number of correct sentences suggests possible processing difficulties. Pre -treatment, P2s recognizable content words for probes were of decreased semanti c weight (e.g., man, boy, her ). The vast majority of her errors were IDK/NR and jargon, with a few phonologic errors close to the target (e.g., clown wown ; brush bwut ). Post treatment, her most prominent type of error remained jargon, but there was a shift in the proportion of her errors to those phonologically closer to the target (e.g., > 50% of the phonemes produced correctly). This significant reduction in NR/IDK errors reveals P2s increased attempts at naming. Following treatment there was an increased proportion of verb nominalizations evidenced by retrieving instruments in place of verbs, which often resulted in mixed errors (e.g., slice knife nait ; write pencil pensho; measure tape take ). According to McRae et al. (2005) and Ferretti et al. (2001) verbs prime not only agents and patients, but als o instruments. Thus, it appears that this network was strengthened, as evidenced by P2s naming of instruments related to the target action. Furthermore, retrieval of these instruments reaffirms P2s relative strength in nouns. Careful observation ove r time suggests that P2 abandoned responses of lower semantic weight (such as man ) for more specific items, which often resulted in a neologism. For example, instead of man she attempted the target admiral which was produced as aperway or adnerbit In t his study, general responses such as man received partial credit. However, when P2 abandoned this response for one of greater specificity, her subsequent neologism resulted in a reduction in accuracy, even though her response reflected an attempt at a more difficult word Although probe graphs are expected to visually reflect participant performance, the underlying mechanism

PAGE 67

67 prompting variability in data points is not always apparent. This again draws attention to the importan ce of error analysis in aphasia treatment research. Pre to Post -treatment Language Measures (Spoken) P2 improved in her ability to name nouns as assessed by the BNT and An O&A Naming Battery. Specifically, she improved 19 percentage points on the BNT and 32 percentage points on An O&A Naming Battery. As suggested with P1, lower frequency words on the BNT may have fostered a more reserved improvement in comparison to items on An O&A Naming Battery. However, given her severity, P2s improvement on both assessments is impressive. Error patterns for noun retrieval were similar for both the BNT and An O&A Naming Battery. The general trend pre treatment on both assessments was the predominance of jargon an d unintelligible responses, followed by NR/IDK and phonological errors that were variable in their relation to the target. Post treatment, the majority of errors continued as jargon and phonological, with both error categories exhibiting a proportion shif t in the number of errors from pre treatment levels. Pre to post -treatment, semantic, unrelated, and mixed errors were rare. Elimination of unintelligible responses with proportional increases in jargon and phonological errors post treatment confirm tha t P2 s responses were closer to the target Similar to nouns, verbs also improved in accuracy as assessed via An O&A Naming Battery with an improvement of 30 percentage points from pre to post -treatment. There was an evolution of erro rs, with pre treatment errors predominantly NR/IDK responses, followed by unintelligible and jargon errors. P2s pre treatment trend of predominantly NR/IDK and unintelligible responses implies a greater deficit in retrieval of verbs as compared to nouns which is consistent with pre treatment probe findings. Post treatment, both unintelligible and NR/IDK were eliminated, with proportional increases in jargon and phonological errors within a few phonemes of the targe t. Thus evolution of errors pre to post -treatment for nouns and verbs

PAGE 68

68 in tasks of lexical retrieval followed similar patterns of 1) reduction and/or elimination of NR/IDK and unintelligible responses, and 2) proportional increases in jargon and phonological errors post treatment, suggesting increased attempts at naming with response s shifting closer to the target. Unlike P1, P2 improved in both noun and verb retrieval that generalized beyond the treatment items, as evidenced by accuracy on th e BNT and An O&A Naming Battery. Although noun retrieval was a pre -treatment strength for both participants, P2 exhibited greater improvement in verb retrieval on both the probes and An O&A Naming Battery. Thus, P2 may have presented with a lesser defici t in verbs than P1. Another plausible explanation for P2s improvement in verbs across tasks may be the modification to the treatment protocol that encouraged P2 to produce the verb along with every noun pair, thus providing opportunities for retrieval, repetition, and feedback. Sentences on the NAVS improved 22 percentage points, from a pre treatment level of 0% complete sentences (e.g., containing all required elements: noun, verb, direct/indirect objects). P2s post -treatment improvement in complete sentences on the NAVS reflects not just her ability to construct a sentence, but also an increase in overall lexical retrieval. Closer analysis of retrieval accuracy for each element of the sentence revealed that P2 retrieved more subjects than objects pre -treatment which contradicts pre treatment probe findings where patients attained higher accuracy than agents. However, consistent with P1, P2s responses fo r subjects on the NAVS were general and of low semantic weight (e.g., man/boy ), and as mentioned previously, probe agents were very specific and specialized professions. Therefore, accuracy inflation of subjects compared to objects on the NAVS for both participants was a direct result of decreased semantic specificity.

PAGE 69

69 Further evidence that P2s pre treatment strength was nouns, she did not retrieve a single verb accurately pre -treatment on the NAVS. Post -treatment, all elements impro ved (e.g., subjects, verbs, direct objects). Specifically, verb retrieval improved almost 39 percentage points with a 22-percentage point improvement in complete sentences. Error analysis of pre to post treat ment responses revealed that NR/IDK responses were significantly reduced, with a proportional increase in jargon errors. This trend supports the assumption that P2s error trends in single noun and verb retrieval reflect increased attempts at naming (e.g., elimination of NR/IDK) and errors approaching the phonological target. Overall, nouns improved greater than verbs on probes; however, trained nouns and verbs improved more than untrained nouns and verbs. Trained verbs improved 53 percentage points pre to post treatment, untrained verbs improved 31 percentage points, while verbs on the NAVS improved 39 percentage points. Therefore, generalization to different tasks was achieved, and this generalization was likely a direct result of increased verb activation in treatment. Treatment for P2 focused on activation, retrieval, and repetition of nouns associated with the provided verb, as well as repetition of the actual verb. Thus, P2s generalization to all lexical r etrieval tasks is likely attributed to activation and retrieval of both nouns and verbs in treatment. Probes (Written) Pre -treatment, P2s written modality was more preserved than her spoken; however, significant improvement on the probes with large effec t sizes was observed in this modality as well. Thus, treatment resulted in generalization to retrieval of content words within sentences for both trained and untrained verb networks in untrained tasks (e.g., picture description), though untrained items to a slightly less extent Maintenance scores were only conducted one month post treatment, and trained items appeared to be maintained better than untrained items.

PAGE 70

70 Post -treatment accuracy of complete sentences on probes (e.g., containing a n agent, verb, and patient) improved 32 percentage points over baseline levels. At the same time, P2 retrieved more than half of the verbs in post -treatment probes. Given that probe elicitation pictures were identical for both spoken and written modaliti es, the discrepancy between oral and written modalities in the number of correct sentences post treatment allows for additional interpretation of level of impairment. Analysis of individual agent, verb, and patient retrieval revealed that P2 was more a ccurate in noun retrieval than verb retrieval during baseline probes. From pre to post -treatment, all elements of trained and untrained sentences improved from 7 to 36 percentage points over baseline, with untrained patients improving the least and tr ained agents improving the most. Similar to spoken probes and to P1, P2 retrieved more patients as compared to agents pre treatment. P2 was consistent with her performance on spoken probes retrieving agents and patients almost equ ally post -treatment; however, there was greater disparity between noun and verb accuracy in the written modality. P2 retrieved almost all agents and patients and more than half of the verbs. Thus, further support is offered that nouns were a relative s trength for P2. Pre -treatment, P2s errors in the written modality differed greatly from the spoken modality Errors in the written modality were predominantly graphemic errors within a few graphemes of the target, which suggests a highe r level of processing as compared to the spoken modality (e.g., NR/IDK, unintelligible, jargon) Graphemic errors were followed by semantic errors occurring in both grammatical classes; however, the majority of responses in a different grammatical class f rom the target were usually the result of noun retrieval (more specifically, an instrument) for a verb (e.g., slice knife ). In this modality she presented with very few instances of NR/IDK or jargon, with most responses extremely close to the target. T his supports

PAGE 71

71 the theory that the written modality was more preserved than spoken prior to treatment, and that the type of errors in the written modality reflect a higher level of processing than what was captured in the spoken modality. Post -treatment, ja rgon was eliminated and the predominant error type remained graphemic (> 50% graphemes produced correctly) followed by semantic errors in both grammatical classes, with errors in the different grammatical class typically the result of ver b nominalization (e.g., instruments retrieved for verbs ). While this pattern was captured only post treatment in the spoken modality, P2 retrieved an instrument for an associated action in the written modality during pre treatment prob es. This is not surprising given her relative pre treatment strength in nouns. Overall, P2 achieved almost perfect accuracy for thematic roles in the written modality. To further substantiate the earlier claim that P2 was abandoning spoken responses of decreased semantic weight (e.g., man ) for those of greater specificity, P2 attempted written naming of colonel (an untrained agent for probes), which was written as colonal Pre to Post -treatment Language Measures (Written) Written naming of nouns improved on An O&A Naming Battery increasing 8 percentage points over pre treatment levels. P2s primary errors pre treatment were semantic same grammatical class, followed closely by graphemic and unrelated errors Post -treatment, there w as a proportional increase in graphemic errors within a couple graphemes of the target as well as NR/IDK, with a reduction in semantic errors Thus, P2 increased her naming attempts with responses that were extremely close to target following treatment. Written naming of verbs also improved on An O&A Naming Battery, increasing 20 percentage points over pre treatment levels. The primary error type pre treatment was semantic (within both grammatical classes) and unrelated of a different gramma tical class. Post -treatment, semantic errors predominated and were confined primarily to a different grammatical class (e.g.,

PAGE 72

72 nouns), followed by a mixture of unrelated and jargon errors. This trend suggests that P2 relied on her relative strength of nouns, which is consistent with her errors post treatment on written probes. Given that pre treatment testing for nouns and verbs revealed increased accuracy in the written modality as c ompared to spoken, this more reserved improvement evidenced in post testing is not unexpected. Pre treatment testing in the written modality for P2 not only established this as her more preserved modality but also revealed a high er level of processing F urthermore, testing this modality also localized a disruption between the semantic system and the output lexicon as evidenced by the magnit ude of semantic errors in the written modality. Thus, attention to multiple modalities as well as error patterns within each modality is essential for localizing levels of impairment and determining the most accurate representation of the participants level of processing General Considerations This study set out to determine whether VNeST was an appropriate treatment for persons with more severe aphasia. Both participants in this study improved significantly on trained and untrained verb networks, alt hough only one of the participants (P2) showed generaliz ation to all lexical retrieval measures. As predicted in research question one trained and untrained verb networks for P1 and P2 generalized to a lesser extent and required more treat ment sessions than the moderate participants in the first study. Although both participants entered the study with different manifestations of communication impairment, detailed examination of their errors allowed for comparisons to be drawn. Pre treatment, both participants retrieved nouns more accurately than verbs. Further investigation revealed that although both participants improved pre to post treatment in their lexical retrieval abilities, nouns remained a relative strength. Thus, in t he current study, noun retrieval supported the activation of the verb, as proposed by McRae et

PAGE 73

73 al. (2005) and Ferretti et al. (2001). P1 and P2s more preserved noun retrieval confirms the prediction proposed in research question two, and is consist ent with the literature reporting lexical retrieval abilities for persons with nonfluent aphasia (Druks, 2002, Raymer & Ellsworth, 2002). Close examination of generalization patterns and completion of an error analysis to determine evolution of errors pre to post treatment determined that P1 exhibited a more severe verb retrieval deficit than P2. Another hypothesis for the discrepancy in generalization between P1 and P2 may be attributed to the modification made in treatment for P2 of verb inclusion with each noun pair. Retrieval, repetition, and feedback on both the nouns and the verb in therapy may have increased verb activation to levels needed for generalization to occur; however, further investigation would be needed to fully support this claim. N umerous factors should be taken into consideration in aphasia treatment research. Depending on severity, modifications to treatment protocol such as the current study made for provision of written responses during treatment for P2 may be necessary. Furth ermore implementation of a hierarchical scoring system that better capture s improvement in lexical retrieval (e.g., responses within a few phonemes of the target) and level of processing ( e.g., general vs. specific response ) may be warrante d. Error analysis in aphasia treatment research, particularly with more severe participants, is essential. In this study, without taking into account performance in different modalities, as well as detailed analysis of participant e rrors across a variety of tasks, subtle trends may have gone unnoticed (e.g., decreased specificity of patients as compared to specialized agents on probes). Furthermore, studying errors for trends may guide the researcher or clinician in localizing a par ticipants level of impairment. Given P1s relatively preserved semantic system, as

PAGE 74

74 determined by performance on the Py ramids and Palm Trees and Kissing and Dancing test s her level of impairment can be localized between the semantic system and the phonological output lexicon which explains the predominance of semantic errors H owever, th e presence of phonologic errors close to the target suggest s additional disruption at the level of the phonologic output lexicon (Kay et al., 1992). According to Dell et al. (1997) the abundance of semantic errors implies disruption at the lemma level (e.g. mediation between semantic and phonological levels) where activation of numerous semantic competitors to the target results in the selection of a higher activated semantically related node. For these rea son s careful analysis of errors is imperative, as semantic errors alone do not indicate impairment at the level of the semantic system (Raymer & GonzalezRothi, 2000, 2002) According to Kay et al. (1992) predominance of jargon and phonologica l errors post treatment for P2 implicates primary disruption at the level of the phonological output lexicon. Similarly, Dell et al.s (1997) model localizes P2s impairment at the phonological level, with activation of numerous phoneme nodes and subseque nt selections resulting in jargon and phonological errors. Kay et al. s (1992) model accounts for disturbances in multiple modalities and as such, is useful for interpreting P2s written e rror trends P2s errors in the written modality suggest disruption at the level of the orthographic output lexicon, but to a much less er extent than the phonological output lexicon F urthermore, errors within the written modality exposed interferen ce between the semantic system and the orthographic output lexicon thus explaining the presence of semantic errors, which would have gone unnoticed solely in the spoken modality. Aminoff et al. (2008) further inter pret semantic errors present in the written but absent in the spoken modality as a modality -specific disturbance of the orthographic output

PAGE 75

75 lexicon. P2s more preserved written as compared to spoken modality lends further support that the phonological output lexicon is primarily disrupt ed (Chialant et al., 2002) Finally, the findings for research question three were as predicted. E volution of errors for both participants reveal ed a higher level of processing pre to post treatment that accuracy did not capture. Although both participants presented with different manifestations in communication impairments pre -treatment, improved lexical retrieval and levels of processing post treatment suggests treatment was benef icial, despite differences in levels of impairment. Therefore treatment s focusing on shared semantic features, engagement of the semantic system for spreading of activation, and interaction between verbs and thematic roles for increasing lexical retrieva l abilities and the extent of generalization are promising ( Boyle, 2004; Edmonds et al., 2009; Kiran & Thompson, 2003) Factors impacting generalization and maintenance performance s (e.g., cognition, depression, communication opportunities grammatical cl ass retrieval deficits ), as well as t he applicability of VNeST to a wider range of aphasia severity is warranted and currently under consideration. In summary results of this study demonstrate the importance of attention to multiple modalities, err or analysis implementation, and familiarity with models of lexical retrieval for capturing the subtle nuances of language impairments and improvements in aphasia treatment for individuals with severe impairments

PAGE 76

76 L IST OF REFERENCES Beeson P. M., & Robey, R. R. (2006) Evaluating single -subject treatment research: Lessons learned from the aphasia literature. Neuropsychological Review 16(4), 161-169. Boyle M. (2004) Semantic feature analysis treatment for anom ia in two fluent aphasia syndromes. American Journal of Speech-Language Pathology 13 236249. Chapey, R. & Hallowell, B. (2001). Introduction to language intervention strategies in adult aphasia. In R. Chapey (Ed.) Language interventi on strategies in aphasia and related neurogenic communication disorders. Philadelphia : Lippincott, Williams & Wilkins Chialant, D., Costa, A., & Caramazza, A. (2002). Models of n aming. In A. E. Hillis (Ed.) The h andbook of a dult l anguage d isorders. New York: Psychology Press Dell, G. S. Schwartz, M. F. Martin, N. M. Saffran, E. M. & Gagnon, D. A. (1997). Lexical access in aphasic and nonaphasic speakers. Psychological Review, 104 (4), 801 838. Dollaghan, C A. (2007). Importance of evidence: An overview. In Handbook for evidence based practice in communication disorders Baltimore : Paul H. Brookes Publishing Company, Inc. Druks, J. (2002). Verbs and nouns: A review of the literature. Journal of Neurolinguistics, 15, 289315. Druks, J., & Masterson, J. (2000). An O bject & A ction Naming B attery. Hove: Psychology Press. Edmonds L., Nadeau, S., & Kiran, S. (2009) Effect of Verb Network Strengthening Treatment (VNeST) on lexical retrieval of content words in sentences in persons with aphasia. Aphasiology, 23(3), 402424. Edmonds L., & Kiran, S. (2006). Effect of semantic based treatment on cross linguistic generalization in bilingual aphasia. Journal of Speech, Language, and Hearing Research, 49, 729748. Edmonds L. & Mizrahi S. (in preparation) Priming of verbs and their them atic roles in younger and older adults. Ferretti T. R., McRae, K., & Hatherell, A. (2001). Integrating verbs, situation schemas, and thematic role concepts. Journal of Memory and Language, 44, 516 547. Funnell, E. (2002.) Semantic Memory. In A. E. Hi llis (Ed.), The handbook of adult language disorders. New York: Psychology Press. Goodglass, H., Kaplan, E., & Weintraub, S. (1983). Boston Naming T est. Philadelphia : Lea & Febiger.

PAGE 77

77 Hillis A. E. (1998) Treatment of naming disorders: New issues regarding old therapies. Journal of the International Neuropsychological Society, 4, 648660. Hillis, A. E. (2007). Aphasia: Progress in the last quarter of a century. Neurology, 69, 200213. Howard, D., & Patterson, K. (1992). P yramids and p alm t rees London: Harcourt Assessment. Kay, J., Lesser, R., & Coltheart, M. (1992). Psycholinguistic Assessment s of Language Processing in Aphasia (PALPA) New York : Psychology Press Kertesz, A. (1982). Western A phasia B attery Austin, TX: Pro -ed. Kiran, S. (2008) Typicality of inanimate category exemplars in aphasia treatment: Further evidence for semantic complexity. Journal of Speech, Language, and Hearing Research, 51, 15501568. Kiran, S., & Bassetto, G. (2008). Evalua ting the e ffectiveness of s emantic -b ased t reatment for n aming d eficits in a phasia: What w orks? Seminars in Speech and Language, 29(1), 71 82. Kiran S., & Thompson C.K. (2003) The role of semantic complexity in treatment of naming deficits: Training semantic categories in fluent aphasia by controlling exemplar typicality. Journal of Speech, Language, and Hearing Research, 46, 773 787. Levelt, W. J., Schriefers, H., Vorberg, D., Meyer, A. S., Pechmann, T., & Havinga J. (1991). The time course of lexical access in speech production: A study of picture naming. Psychological Review, 98, 122142. Loverso F. L., Prescott, T. E., & Selinger, M. (1988) Cueing verbs: A treatment strategy for aphasic adults (CVT). Journal of Rehabilitation Research, 25 (2), 47 60. McRae K., Hare, E., & Ferretti, T. R. (2005). A basis for generating expectancies for verbs from nouns. Memory and Cognition, 33(7), 117484. Murray, L L., & Clark, H. M. (2006). Neurogenic d isorders of language : Theory d ri ven c linical p ractice New York: Thomson Delmar Learning. Nickels L. (2002) Therapy for naming disorders: Revisiting, revising, and reviewing. Aphasiology 16, 935 979. Porch B. E. (1973) The Porch Index of Communicative Abilities: Administration, scoring and interpretation. Palo Alto, CA: Consulting Psychologists Press.

PAGE 78

78 Prescott T. E., Selinger, M., & Loverso, F. L. (1982) An analysis of learning, generalization, and maintenance of verbs by an aphasic patient. In Clinical Aphasiology Confe rence BRK Publishers, 178182. Raymer A. M., & Ellsworth T. A. (2002) Response to contrasting verb retrieval treatments: A case study. Aphasiology, 16(10), 10311045. Raymer, A .M., & Gonzalez-Rothi L .J. (2000). The semantic system. In S. E. Nadeau, L. J. GonzalezRothi, & B. Crosson (Eds. ), Aphasia and language: Theory to practice New York: The Guilford Press Raymer, A. M., & Gonzalez -Rothi, L. J. (2002) Cl inical diagnosis and treatment of naming disorders. In A. E. Hillis (Ed.), The handbook of adult language disorders New York: Psychology Press. Raymer, A. M. & Kohen, F. (2006). Wordretrieval treatment in aphasia: Effects of sentence context. Journal of Rehabilitation Research and Development, 43(3), 367378. Thompson, C. K. (2002). The Northwestern V erb P roduction B attery /The Northwestern Assessment of Verbs and Sentences Unpublished. Tryon W.W. (1982) A simplified time -series analysis for e valuating treatment interventions. Journal of Applied Behavior Analysis, 15, 423429. Wambaugh J. L., & Ferguson, M. (2007) Application of semantic feature analysis to retrieval of action names in aphasia. Journal of Rehabilitation Research and Develop ment, 44(3), 381 394. Wilshire, C. E., & Coslett, H. B. (2000). Disorders of word retrieval in aphasia: Theories and potential applications. In S. E. Nadeau L. J. Gonzalez Rothi, & B. Crosson (Eds.) Aphasia and language: Theory to practice New York: Th e Guilford Press.

PAGE 79

79 BIOGRAPHICAL SKETCH Michelle E. (Lucas) Babb was born in Ft. Lauderdale, Florida and was raised in Miami and Jacksonville, Florida. In 2004 she moved to Gainesville, and received a Bachelor of Arts degree in c ommunication s ciences an d d isorders from the University of Florida in May 2007. In August 2007 she began her graduate program at the University of Florida and received a Master of Arts degree in c ommunication s ciences and d isorders in May 2009 Following graduation, she wil l begin her career in s peech l anguage p atholog y with the eventual attainment of a PhD in communication sciences and d isorders Her primary research interests include aphasia rehabilitation, lexical retrieval in persons with aphasia a nd factors affecting generalizatio n.