<%BANNER%>

When Things Do (Not) Refer

University of Florida Institutional Repository
Permanent Link: http://ufdc.ufl.edu/UFE0022274/00001

Material Information

Title: When Things Do (Not) Refer An fMRI Study on Anaphora in Discourse
Physical Description: 1 online resource (69 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: anaphora, brain, discourse, fmri, processing, quantifier, reference
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: How do we analyze referents in multi-sentence discourses when they do not match? Discourse processing is generally considered a right hemisphere function; however this stance is somewhat controversial, as experiments until now have not elicited consistent results. To determine what areas of the brain speakers activate and thereby gain a better understanding of discourse and quantifier processing, we designed and conducted the following fMRI experiment. Fourteen neurologically healthy, monolingual speakers of English viewed 128 sets of two-sentence trials (constructed based on the results of 3 offline pretests). Experimental sentences either had matching quantifiers or mismatched ones: Four tires were leaking some air. All four had already been patched last week (plausible); Four tires were leaking some air. All three had already been patched last week (implausible). The first sentence of each set was presented two words at a time, for 400ms each (200ms blank in between). The first two words of the second sentence (containing the second quantifier) were presented in the same manner; the rest of the second sentence was shown word-by-word for 300ms each (separated by 200 ms blank). Subjects were then cued to judge (by pressing buttons) whether the two sentences were plausible or implausible together. An analysis of the data seen in the current study does not support the idea that discourse processing is lateralized to the right hemisphere in the brain. No areas of activation were statistically significant in a comparison of the implausible to the plausible trials. However, comparing the plausible to the implausible trials yielded nine significantly activated areas. Activation in the left hemisphere's prefrontal cortex, which has been implicated in inferencing and integration processes (Mason & Just 2004, Kuperberg et al. 2006), in addition to activation seen in the right superior temporal gyrus, support the idea that these areas are involved in the successful integration of semantic relationships (propositions) between the first and second sentences in the plausible but not the implausible trials, as semantic integration in implausible trials is impossible. Further analysis looking at the functional connectivity among these regions will contribute to the understanding of the network that may exist, possibly relating the activations which appear in the present study.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis: Thesis (M.A.)--University of Florida, 2008.
Local: Adviser: Kaan, Edith.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022274:00001

Permanent Link: http://ufdc.ufl.edu/UFE0022274/00001

Material Information

Title: When Things Do (Not) Refer An fMRI Study on Anaphora in Discourse
Physical Description: 1 online resource (69 p.)
Language: english
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2008

Subjects

Subjects / Keywords: anaphora, brain, discourse, fmri, processing, quantifier, reference
Linguistics -- Dissertations, Academic -- UF
Genre: Linguistics thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: How do we analyze referents in multi-sentence discourses when they do not match? Discourse processing is generally considered a right hemisphere function; however this stance is somewhat controversial, as experiments until now have not elicited consistent results. To determine what areas of the brain speakers activate and thereby gain a better understanding of discourse and quantifier processing, we designed and conducted the following fMRI experiment. Fourteen neurologically healthy, monolingual speakers of English viewed 128 sets of two-sentence trials (constructed based on the results of 3 offline pretests). Experimental sentences either had matching quantifiers or mismatched ones: Four tires were leaking some air. All four had already been patched last week (plausible); Four tires were leaking some air. All three had already been patched last week (implausible). The first sentence of each set was presented two words at a time, for 400ms each (200ms blank in between). The first two words of the second sentence (containing the second quantifier) were presented in the same manner; the rest of the second sentence was shown word-by-word for 300ms each (separated by 200 ms blank). Subjects were then cued to judge (by pressing buttons) whether the two sentences were plausible or implausible together. An analysis of the data seen in the current study does not support the idea that discourse processing is lateralized to the right hemisphere in the brain. No areas of activation were statistically significant in a comparison of the implausible to the plausible trials. However, comparing the plausible to the implausible trials yielded nine significantly activated areas. Activation in the left hemisphere's prefrontal cortex, which has been implicated in inferencing and integration processes (Mason & Just 2004, Kuperberg et al. 2006), in addition to activation seen in the right superior temporal gyrus, support the idea that these areas are involved in the successful integration of semantic relationships (propositions) between the first and second sentences in the plausible but not the implausible trials, as semantic integration in implausible trials is impossible. Further analysis looking at the functional connectivity among these regions will contribute to the understanding of the network that may exist, possibly relating the activations which appear in the present study.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis: Thesis (M.A.)--University of Florida, 2008.
Local: Adviser: Kaan, Edith.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2008
System ID: UFE0022274:00001


This item has the following downloads:


Full Text

PAGE 1

WHEN THINGS DO (NOT) REFER: AN FM RI STUDY ON ANAPHORA IN DISCOURSE By NICOLE A. CHEVALIER A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2008 1

PAGE 2

2008 Nicole A. Chevalier 2

PAGE 3

To Brian, Dorito, and Gwen, who have now dealt with the writing of two theses. 3

PAGE 4

ACKNOWLEDGMENTS Many people have contributed to the ultimate out put of this thesis; though I will attempt to thank them all here, I doubt this will be as comp rehensive a list as perhaps is necessary. I would like to thank my family for their love and continued support of my academic pursuits to my parents especially, for having gone above and beyond the call of dut y. If they had a nickel for every time I vented about this project or schoolwo rk in general, they c ould feed a small country. My fianc, Brian, deserves thanks not only for dealing with my long hours in the lab, lengthy statistics arguments with myself, and gene ral preoccupation with all things brain-related, but also for allowing himself to be persuaded into participating in experiments. His mom Beverly also merits gratitude, for giving me a pep talk at a much needed time. Dorito and Gwen deserve thanks as well, for reminding me that it is okay to stop and breathe ( once in awhile) and for providing a welcome distraction (perha ps more often than necessary). The faculty and staff of the Linguistics Progr am have also been wonderfully supportive over the past five and a half years. I have benef ited particularly from the assistance and guidance of Wind Cowles and Gary Miller, who taught me both inside and out side of the classroom. I am very grateful to all of you. I wish to also thank all of the participants in this study, most of whom are (were?) my friends, and my friends who offered to particip ate but did not meet th e subject requirements. Their willingness to have their brains scanne d for the sake of science is admirable. I must also acknowledge Paul Wright, Guojun Alex He, Zhenyu Zhou, and Nelson Klahr of the McKnight Brain Institute as integral to this experiment. These fellows trained me in and provided me with the tools necessary to acquire and analyze fMRI data correctly, as well as providing countless hours of tech support. Thanks are due also to Yijun Liu, for agreeing to be on my committee and helping me navigate the hitherto unknown world of connectivity analyses. 4

PAGE 5

Last but most assuredly not least, I would like to express my apprec iation to the chair of my committee, Edith Kaan, for giving me the opport unity to participate on this project and for her continuous support throughout the many ups and downs of the past few years. I cannot thank her enough. 5

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES.........................................................................................................................9 LIST OF ABBREVIATIONS........................................................................................................11 ABSTRACT...................................................................................................................................13 CHAPTER 1 INTRODUCTION................................................................................................................. .15 Discourse Processing........................................................................................................... ...15 Levels of Representation........................................................................................................16 Quantifiers.................................................................................................................... ..........18 2 BACKGROUND................................................................................................................... .20 Discourse Processing and the Right Hemisphere...................................................................20 Patient Studies.................................................................................................................20 FMRI and PET................................................................................................................21 Inferencing and Integration.................................................................................................... .24 Coreference.............................................................................................................................26 Present Studys Contributions................................................................................................27 Predictions..............................................................................................................................28 3 METHODS...................................................................................................................... .......29 Participants.............................................................................................................................29 Materials.................................................................................................................................29 Procedure................................................................................................................................31 MRI Data Acquisition........................................................................................................... ..32 Data Analysis..........................................................................................................................33 4 RESULTS...................................................................................................................... .........36 Behavioral Data................................................................................................................ ......36 Imaging Data................................................................................................................... .......36 5 DISCUSSION................................................................................................................... ......51 Behavioral Data................................................................................................................ ......51 Imaging Data................................................................................................................... .......51 6

PAGE 7

Bilateral Activations........................................................................................................51 Right Superior Temporal Gyrus......................................................................................52 Prefrontal Cortex.............................................................................................................5 3 Limitations and Future Experiments......................................................................................55 APPENDIX A LIST OF MATERIALS USED..............................................................................................56 LIST OF REFERENCES...............................................................................................................66 BIOGRAPHICAL SKETCH.........................................................................................................69 7

PAGE 8

LIST OF TABLES Table page 3-1. Example of stimuli used. | indicates th e division of frames, as presented to the participant. Critical words in e xperimental items are underlined......................................35 4-1. Analysis of behavioral data, including RT Accuracy, and standard deviations for each condition............................................................................................................................38 4-2. ROI FFX GLM Analyses, showing degrees of freedom, mean and standard deviations of activation, t value, and significance level......................................................................38 4-3. Activated clusters at q(FDR) = 0.05; cluster threshold size = 128. The second through fourth columns refer to the coordinates (in Talairach space) of the center of activation for each cluster..................................................................................................39 A-1 List 1, Run 1 Material s, sorted by condition..........................................................................56 A-2. List 1, Run 2 Material s, sorted by condition.........................................................................57 A-3. List 1, Run 3 Material s, sorted by condition.........................................................................58 A-4. List 1, Run 4 Material s, sorted by condition.........................................................................59 A-5. List 1, Run 5 Material s, sorted by condition.........................................................................60 A-6. List 2, Run 1 Material s, sorted by condition.........................................................................61 A-7. List 2, Run 2 Material s, sorted by condition.........................................................................62 A-8. List 2, Run 3 Material s, sorted by condition.........................................................................63 A-9. List 2, Run 4 Material s sorted by condition..........................................................................64 A-10. List 2, Run 5 Materi als sorted by condition........................................................................65 8

PAGE 9

LIST OF FIGURES Figure page 3-1. The GLM model used. The conditions are co lor-coded according to the legend on the right. The white spikes are the HRF response modeled for that condition (implausible (c), in this case); this was done for each condition, so that each had an HRF response modeled. The numbers at the botto m represent the volume number, and the numbers on the left axis re present the % of BOLD signal change (such that 1.0 = 100%).................................................................................................................................35 4-1. The significant areas of activation (in orange and blue) in a glass brain; view of the left hemisphere..................................................................................................................... ....40 4-2. The significant areas of activation (in oran ge and blue) in a glass brain; view of the right hemsiphere............................................................................................................... ..41 4-3. The significant activation in the superior temporal gyrus of the right hemisphere (RSTG), along with the event-related averag ing plot of the same region. The green line signifies the time course of activati on for the plausible co ndition, and the red the implausible.................................................................................................................... .....42 4-4. The significant activation in the precentral gyrus of the right hemisphere (RPreCG), along with the event-related averaging plot of the same region (green signifying the time course of activation for the plau sible condition, red the implausible).......................43 4-5. The significant activation in the right middle frontal gyrus (RMFG), along with the event-related averaging plot of the same region (green signifying the time course of activation for the plausible condi tion, red the implausible)...............................................44 4-6. The significant activation in the right claustrum, along with the event-related averaging plot of the same region (green signifyi ng the time course of activation for the plausible condition, red the implausible)...........................................................................45 4-7. The significant activation in the left s uperior frontal gyrus (LSFG), along with the event-related averaging plot of the same region (green signifying the time course of activation for the plausible condi tion, red the implausible)...............................................46 4-8. The significant activation in the left me dial frontal gyrus (LMedFG), along with the event-related averaging plot of the same region (green signifying the time course of activation for the plausible condi tion, red the implausible)...............................................47 4-9. The significant activation in the left cl austrum (Lclaus), along w ith the event-related averaging plot of the same region (green si gnifying the time cour se of activation for the plausible condition, red the implausible).....................................................................48 9

PAGE 10

4-10. The significant activation in the left middle frontal gyrus (LMFG), along with the event-related averaging plot of the same region (green signifying the time course of activation for the plausible condi tion, red the implausible)...............................................49 4-11. The significant activation in the left pre central gyrus (LPreCG), along with the eventrelated averaging plot of the same region (green signifying the time course of activation for the plausible condi tion, red the implausible)...............................................50 10

PAGE 11

LIST OF ABBREVIATIONS ANOVA Analysis of variance aTL Anterior temporal lobe BOLD Blood oxygen level dependent BVQX BrainVoyager QX CI Construction-Integration dlPFC Dorso-lateral prefrontal cortex dmPFC Dorso-medial prefrontal cortex ELAN Early left anterior negativity EPI Echo-planar imaging ERP Event-related potential FFX Fixed-effects analysis fMRI Functional magnetic resonance imaging FOV Field of view GLM General linear model HRF Hemodynamic response function IFG Inferior frontal gyrus IPC Inferior parietal cortex ISI Inter-trial stimulus LH Left hemisphere LIFG Left inferior frontal gyrus LMedFG Left medial frontal gyrus LMFG Left middle frontal gyrus LPreCG Left precentral gyrus LSFG Left superior frontal gyrus 11

PAGE 12

LSTS Left superior temporal sulcus MRI Magnetic resonance imaging ms Milliseconds PCC Posterior cingulate cortex PET Positron emission tomography pIPFC Posterior inferior prefrontal cortex pTL Posterior temporal lobe RaTL Right anterior temporal lobe RF Radio frequency RFX Random effects analysis RH Right hemisphere RIFG Right inferior frontal gyrus RIFS Right inferior frontal sulcus RMFG Right middle frontal gyrus ROI Region of interest RPreCG Right precentral gyrus RSTG Right superior temporal gyrus RT Reaction time s Seconds SENSE Sensitivity encoding; a specific type of imaging technique utilized to reduce scan time TD Talairach Daemon ToM Theory of mind TR Repetition time UTHSCSA University of Texas Health Science Center San Antonio WM Working memory 12

PAGE 13

Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Arts WHEN THINGS DO NOT REFER: AN FMRI STUDY ON ANAPHORA IN DISCOURSE By Nicole A. Chevalier May 2008 Chair: Edith Kaan Major: Linguistics How do we analyze referents in multi-sentence discourses when they do not match? Discourse processing is generally considered a right hemisphere f unction; however this stance is somewhat controversial, as experiments until now have not elicited co nsistent results. To determine what areas of the brain speakers activ ate and thereby gain a better understanding of discourse and quantifier proces sing, we designed and conducted the following fMRI experiment. Fourteen neurologically healt hy, monolingual speakers of Eng lish viewed 128 sets of twosentence trials (constructed base d on the results of 3 offline pretests). Experimental sentences either had matching quantifiers or mismatched ones: Four tires were leaking some air. All four had already been patched last week (plausible); Four tires were leaking some air. All three had already been patched last week (implausible). The first sentence of each set was presented tw o words at a time, for 400ms each (200ms blank in between). The first two words of the se cond sentence (containing the second quantifier) were presented in the same manner; the rest of the second sentence was shown word-by-word for 300ms each (separated by 200 ms blank). Subjects were then cued to judge (by pressing buttons) whether the two sentences were plau sible or implausible together. 13

PAGE 14

14 An analysis of the data seen in the current study does not support the idea that discourse processing is lateralized to th e right hemisphere in the brai n. No areas of activation were statistically significant in a comparison of the implausible to the plau sible trials. However, comparing the plausible to the implausible tria ls yielded nine signifi cantly activated areas. Activation in the left hemispheres prefrontal co rtex, which has been implicated in inferencing and integration processes (Mason & Just 2004, Kupe rberg et al. 2006), in addition to activation seen in the right superior temporal gyrus, support the idea that these areas are involved in the successful integration of semantic relationships (propositions) between the first and second sentences in the plausible but not the implausible trials, as semantic integration in implausible trials is impossible. Further analysis looking at the functiona l connectivity among these regions will contribute to the understanding of the network that may exist, possibly relating the activations which appear in the present study.

PAGE 15

CHAPTER 1 INTRODUCTION Although many functional magnetic resonanc e imaging (fMRI) studies have been conducted looking at discourse pr ocessing (e.g. Chow et al. 2007; Robertson et al. 2000) and reference processing (e.g. Almor et al. 2007, McM illan et al. 2005), none ha s specifically studied the interpretation (or misinterpr etation) of referents in disc ourse. Our experiment sought to remedy that deficit. We looked at how participan ts interpreted quantifiers (number words) in a two sentence discourse, both when the two numbered groups matched and when the groups numbers did not match. Specifically, we were intere sted in the effects of implausible quantifier mismatches on patterns of activation in the brain, and whether th ere were hemispheric asymmetries in resolving this type of refe rence problem. Because previous research has implicated areas in the right hemisphere (RH) for the processing of discourse, we anticipated activation in the RH if previous hypotheses were sound. Discourse Processing When processing sentences in a discourse, the reader or listene r creates a mental representation of the people, objects, and re lationships mentioned in the discourse. The representation is continua lly updated and maintained as the r eader progresses further along in the discourse, incorporating new part icipants and events or modi fying the representation as it pertains to previously mentione d referents. For example, when readers interpret the sentence The kitten and the puppy were adopted from the animal shelter, they create a discourse model in which the two noun phrases, the kitten and the puppy, exist as entities. If a following sentence were to begin with a noun phrase The kitten, readers could either assume that the kitten in the previous sentence is being referred to or that a new kitten is being referred to, and 15

PAGE 16

add another kitten referent into their discourse model. (If a new sentence were to begin The chameleon, readers would be forced to set up a new discourse referent.) Studies have shown that reader s prefer to interpret a sec ond noun phrase as referring to a previously mentioned entity, rather than set up a new discourse refe rent. McKoon and Ratliff (1992) in particular have demonstrated through behavioral methods that processing and creating new referents is harder than relating new inform ation to entities previously mentioned in the discourse. In other words, it is easier to ascrib e new information to something currently being talked about than create an entirely new referent for that new information. Neurolinguistic results have also supported this idea of new referents be ing hard to process; Kaan, Dallas, and Barkley (2007) elicited a Late Positive Complex when a ne w referent needed to be set up in their eventrelated potential (ERP) st udy of the processing of bare quant ifiers in discourse, which they concluded was due to difficulty in updating the participants mental models. Psycholinguistic results also support this conclu sion. Wijnen and Kaans (2006) behavioral and Frazier and colleagues (2005) eyetracking resu lts also attest to the difference in processing between creating new referents and applying informa tion directly to a previously mentioned entity; participants preferred a reading of the discour se so that the second quantified entity was a subset of the previously introduced one. Levels of Representation Most processing theories so fa r postulate that readers construct (at least) two levels of representation when interpreting discourse: a propositional representation and a situational model. These two levels of pro cessing reflect the ways in whic h the discourse can be described. The propositional level reflects what words and ideas are contained within the sentences given. At this level of processing, representations of semantic structure (called propositions) are formed. Anaphora are incorporated into the discou rse model at the propositional level, as well, 16

PAGE 17

according to Long, Baynes, and Prat (2005). The s econd level, the situational level, reflects a more global, abstract view encompassing what th e text is actually about and incorporating general world knowledge. The situati onal level, then, includes the readers own interpretation of the events. Walter Kintsch and Teun A. van Dijk (1978) we re some of the firs t to hypothesize these two levels of processing. Many re searchers also assume that it is necessary to also build and maintain coherence at each leve l (only local coherence for the propositional, and local and global coherence for the situational level). The form ation of local coheren ce involves the reader mapping the new information onto the recent, previously stated context; global coherence involves mapping the new information onto relevan t information presented earlier in the text but that is no longer available in working me mory (OBrien, Rizzella, Albrecht, and Halleran 1998). Global coherence also involves the in corporation of general world-knowledge. This distinction between (at least) two levels of discourse processi ng has gained support from experimental results. Long and Baynes (2002) used an item-priming-in-recognition paradigm and found that propositionalization wa s largely a left hemisphere (LH) function, whereas situational level eff ects occurred bilaterally. Long, Ba ynes, and Prat (2005) further expanded on this result by demonstrating their LH result for propositionalization is due to neither the close proximity of the noun phrases nor the standard noun-verb-noun syntactic structure of the sentences; the result stands even with the presence of an embedded clause interrupting this normal word order. These results lend support to the idea that there are qualitative and quantitative differences in activation between the propositional and si tuational levels of discourse. 17

PAGE 18

Quantifiers The current experiment made use of quantif iers to investigate reference processing. Quantifiers are words such as most, nine, s eventy, and some. For a quantifier to make sense, it needs to specify the quantity of something namely, a noun phrase. However, depending on context, a noun phrase need not be present in the act ual sentence for the meaning to be understood. (If someone were talking about cats and said M ost are grotesquely overfed, a listener would interpret the quantifier most as referring to the discourse topic cats.) When used without an accompanying nound phrase, quantifiers are a type of anaphora in that they refer to a group or entity, usually previously mentioned or in some way related to another discourse referent. Van Benthem (1986) demonstrat ed a difference between two main types of quantifiers: first-order quantifiers and higher-order quantifie rs. By modeling each type with automatons; he showed that while first-order quantifiers (e.g. six, four) can be modeled with a simple computing device (a finite state automaton), higher-order quantifie rs (e.g. most, few) needed a more complex automaton that has a WM-like device. McMillan et al. (2005) provides a wellwritten distinction between the two types; while first-order quantifiers activate one numerical quantity higher-order quantifie rs must activate a numerical quantity and then perform judgments upon that quantity utilizing resources such as working memory (WM) to do so. First-order quantifiers do not have this additional resource usage. The present study uses quantifiers to further investigate th e functional areas involved in discourse and reference proces sing. The manipulation of whethe r the quantifier in the second sentence matched or did not match the quantifie r of the first, being the basis for our two conditions, should elicit a qualita tively and possibly quantitatively different result based on the construction of propositional and s ituational models. In other wo rds, based on the ability to 18

PAGE 19

19 construct an integrated semantic representati on of the sentences, we should see differential activation (either a stronger or weaker signa l in the same place or a different location).

PAGE 20

CHAPTER 2 BACKGROUND Discourse Processing and the Right Hemisphere Most researchers ascribe to a model in which the right hemisphere (R H) functions as the processor for discourse comprehension and pragmatic qualities (e.g. Bookheimer 2002, Robertson et al. 2000, Hoen et al. 2006, Xu et al 2005). Some, such as Rapp et al. 2004, ascribe some sentence-level processes to the RH as well. While the RH no doubt plays at least a small or supporting role in discourse proce ssing, the data are still inconsis tent and have mixed results. Additionally, researchers have tended to overgen eralize their findings; when data was acquired only from a specific region of the brain, conc luding that discourse comprehension was accompanied by more neural activity in the right than the left hemisphere (Robertson et al. 2000) does not seem to follow. Patient Studies One area of investigation that has lead resear chers down this RH path is patient studies. Patients with RH lesions tend to show probl ems with inferencing, revising incorrect interpretations (Ferstl 2007), and maintaining coherence (M ason and Just 2004). However, patient data (while a valuable resource to gain a starting point for investigation) cannot tell us much at this level of language comprehension, for two reasons. First, the extent of most patients lesions is not confined to one sp ecific area nor are the lesions readily comparable to each other. Some lesions extend into frontal regions as well. Brain injury pa tients with frontal lobe damage also exhibit many of the same kind of deficits as RH lesion patients. While the RH and frontal lobes text comprehension proce sses are likely not the same, there does not seem to exist an account outside of psycholinguistic models whic h has reconciled this with neuropsychological proposals (Ferstl 2007). 20

PAGE 21

FMRI and PET While many neuroimaging studies examin ing discourse proce ssing find significant activation in the RH, th ere are many which do not. Task diffe rences and confounded factors take some of the blame for this lack of consistency (F erstl 2007), but another factor is also a player in the differences seen for discourse processing st udies: the neuroimaging method used. Unless care is taken when scanning with fMRI in orbitofron tal and anterior temporal regions (due to the sinus cavities), signal loss can occur. Positron emission tomogra phy (PET) is better at imaging these regions, but it is often more difficult to tim e-lock stimuli to the acquisition of data. While there are some fMRI experiments that show anterior temporal l obe (aTL) activation (e.g. Ferstl, Rinck, and Von Cramon 2005), others lack this result (e.g. Roberts on et al. 2000). Blocked designs, which are seen in some studies in bot h methodologies (e.g. Robertson et al. 2000), also make it difficult to determine which activation is due to what mani pulation. Comparison of activation across studies is further impeded by the regions of interest chosen prior to scanning; in some studies, the aTL was not included as an acq uisition area at all (Mason and Just 2004). All of these factors together make it difficult to determine to verify to what extent the RH is involved in discourse and reference processing. Xu et al. (2005) is of the opi nion that this lateralization of activation into one hemisphere or the other is taskor contextdriven. They found more bilateral activation at the sentence level and more RH activation at the discourse level, in contrast to Long and Bayness (2002) results. Xu and colleagues also report fi nding that the hemispheric response was modulated over the discourse being read, so that th ere was more LH activation at th e beginning of the discourse (as propositional structures were be ing set up) and more RH activa tion near the end and at the conclusion of the discourse ( possibly representing the form ation of a globally coherent structure). Xu and colleagues interpret this incr eased RH activation as a representation of a 21

PAGE 22

globally coherent structure; howev er, another analysis is possible: the RH activation increases as more and more propositions enter the discourse, taxing executive resources. This result could then be interpreted as representing increased working memory or computational load. Bilateral activation of the ante rior temporal lobes (aTL), as well as left-dominant activation of the superior and inferior front al regions and inferior parietal areas has also been observed (Mazoyer et al. 1993). Mazoyer and colleagues co nducted a PET study comparing a rest baseline with conditions in which participan ts were presented with (French) word lists, stories in a foreign language, stories in French, stor ies containing French pseudoword s, stories with semantically anomalous content words, and stories in Frenc h. In their conclusions, they implicate their anterior temporal lobe (aTL) ac tivation to an involvement in s yntactic processing, as they saw this activation even in pseudoword stories. Fers tl and Von Cramon (2001) also observed bilateral aTL activation, and additional areas of activation in the posterior te mporal sulcus (pTL) and leftdominant, bilateral involvement of the IFG and post inferior prefr ontal cortex (pIPFC) in their comparison of two short texts versus a non-word baseline. When compared with a non-la nguage baseline, activation ac ross neuroimaging studies for discourse processing seems to focus on the left s uperior temporal sulcus (LSTS) and the pars opercularis and/or triangul aris of the left IFG. Ferstl (2007) adds a caveat in her review of these discourse processing studies, in th at none of the activations menti oned is specific to contextual language comprehension, being present even in th e comparison of a resting state to an attention task (Binder, Frost, Hammeke et al. 1999). Ergo, we cannot conclude that activation in the LIFG and LSTS areas result specifically from processi ng discourse, though they ma y be functionally or anatomically linked. 22

PAGE 23

Certain areas of the brain have been imp licated in each posited level of discourse processing. However, the story remains unclear as to which areas of the brain are related to each level of processing; overlap abounds. Ferstl and von Cramons (2001) study examining local coherence (propositional) processe s also found activation they belie ve is related to situational model updating. The previously mentioned study by Mazoyer et al. f ound activation to be significant only in the narrative condition for the dorso-medial prefrontal cortex (dmPFC), whereas activation in the aTLs was present for both the sentence and story condition. They ultimately conclude that the aTLs play a role in semantic encoding thereby linking these areas to the propositional level of a model of di scourse comprehension. Long and Baynes (2002), however, utilized a hemi-field priming task to show evidence for LH effects of propositional structure. They also found bilateral hemispheric effects for processing at the situational level. Other areas have been implicated in the situ ational level of discourse processing. Ferstl (2007) also discussed results im plicating the posterior cingulate cortex (PCC) in situational modeling. Based on results from Ferstl and Von Cr amon (2002) and Maguire et al. (1999), which found activation in the PCC, she and the other authors concluded the PCC was involved in the integration of novel information with pr ior context or general world-knowledge. Another important area of activation is the pr efrontal cortex. The dorso-medial prefrontal cortex (dmPFC BA 8, 9, and 10) activation fou nd in several discourse processing studies (e.g. Mazoyer et al. 1993, Ferstl and Von Cramon 2001) does not have a clear justification as of yet. Ferstl (2007) discusses two possibl e explanations: theo ry of mind (ToM) processes or general inferencing processes. Theory of mind refers to the ability to recogni ze that other peoples actions are related to their beliefs and mental stat es. Based on the results of another study (Ferstl and von Cramon 2002), and the demonstrated result of the dmPFC being more active for 23

PAGE 24

coherent as opposed to incoherent trials, Ferstl (2007) concludes act ivation in this region is due to general inferencing processes. The dorso-later al prefrontal cortex (dlPFC) has also been implicated in inferencing processes; or more specifically, in the generatio n of inferences (Mason and Just 2004). Inferencing and Integration Events in a discourse are often not explicitly connected to one another, necessitating that the hearer or reader generate inferences to de termine the relationship between the two (or more) events for comprehension. For example, in a discourse such as I became angry with my cat. The kitchen counter had paw prints on it readers must infer that the cat jumped on the counter, that this is the reason the speaker is angry with her, and then integrate thos e inferences into their discourse model explaining the relationship between the two sentences. The constructionintegration (CI) model of text comprehensi on (Kintsch 1988) reflects these processes. As explained by the CI model, readers ge nerate many possible inferences, and then through an integration process select the inferences that relate most to the text-base (what has been stated explicitly). For th e example discourse given above, al ternative inferences that might also be made that an animal had jumped on the c ounter, or that the cat fa iled to clean up the paw prints. These are ultimately rejected because th e inference above (that the cat jumped on the counter) is more directly related to the discourse, and ultimately makes more sense. The resulting successful discourse model, then, reflects both th e propositions generated from the discourse and the propositions of the successfully integrated inference. Some researchers have come to the conclusion that the RH is involved in inferencin g processes as a part of its supposed dominance for discourse processing. Robertson and colleagues (2000) looked at inferencing processes, which they call mapping. They ultimately found two areas of activ ation for more coherent sentence blocks: the 24

PAGE 25

right inferior frontal sulcus (RIFS) and the right inferior frontal gyrus (RIFG). However, Robertson et al. did not include most of the temp oral lobe in their anal ysis (supposedly to avoid susceptibility artifact). As noted above, this can be a problem with fMRI data acquisition; however, we cannot draw conclusions based on a l ack of finding for the temporal lobe in this study. Robertson and colleagues conclude that, base d on previous results that implicate the area in allocation of internal atten tion, this sub-process must be in volved in mapping or inferencing processes. Since parts of the temporal lobe we re not included in the comparison, they may be involved in this process as well. As mentioned previously, Fers tl and von Cramon (2001) were in terested in investigating local coherence processes, without global text factors or task effects. Usi ng event-related fMRI, they conducted an experiment with two-sentence tr ials in which they manipulated coherence and cohesiveness (as measured by the insertion of a conjunction or related pronoun). Their participants were told to res pond whether the sentences in a tria l were pragmatically related to each other, necessitating the gene ration of inferences to link th e two sentences together (for example, The lights have been on since last night. The car doesnt start ). Ferstl and von Cramon could not, however, replicate the results of R obertson et al.s study, leading to the continued murky role of the right hemisphe re in inferencing processes. Mason and Just (2004) conduc ted an fMRI experiment examining causal-related inferencing. According to previous work, an in termediately-related sentence pair was better recalled than either a highlyor distantly-related pair; the theory proposed was that the increase in recall was due to readers having generated a causal inference between the two sentences, which in turn might have increased the number of possible retrieval cues for better recall later on in the task. Mason and Just saw differential activation in what th ey define as the RH-language 25

PAGE 26

areas (including the inferior front al, inferior parietal, and tempor al gyri) for the intermediate versus the highlyand distantly-related pairs. The RH activation seen for this intermediate condition is equal to the LH ar eas; in the other two conditions, LH activation was greater. Mason and Just use this data as support for Kintschs co nstruction-integration model. They interpret the RH activation as involved in integration proce sses, after the dlPFC has been involved in the generation of the inferences. They interpret th e LH activation as due to the construction of propositional representations, whic h they say would likely remain constant across all conditions. Their RH temporal finding is intriguing, as it sh ows a dramatic increase in activation for the intermediately-related co ndition, but they conflate the supe rior and middle gyri without giving specific localization information. They ultimately c onclude that the bilateral dlPFC is involved in the generation of an inference, and the RH ar eas then integrate the successfully generated inference into the readers discourse model. Coreference Almor et al. (2007) examined another area of processing relevant to the current study: anaphora processing and coreference. These researchers were primarily interested in the repeated name penalty (in which repeated use of a name instead of a pronoun increases processing time). Because pronouns are often ambiguous, comprehenders preference for them over precise, repeated names may seem paradoxical at first. Ho wever, when one ascribes to the view that a name causes an automatic generation of a new disc ourse referent (which as described above, is more difficult to process than relating new inform ation to a previous referent), the preference for pronouns is clearer if they do not cause th e generation of a new referent but instead immediately relate and integrate information in to preexisting referents. In this way, pronouns would reduce WM load. Hypothesizi ng that a repeated name would activate areas related to WM and integration processes, they expected to see act ivation in the temporal and parietal lobes for a 26

PAGE 27

contrast between their repeated name and prono un conditions. They used three-sentence trials, with either a repeated name or a pronoun appeari ng in the second and third sentences. No areas of the brain were found to be more accura te in the pronoun > repeated name condition. Conversely, the repeated name condition led to mo re activation in the superior parietal lobule and precuneus, left fusiform gyrus and left mi ddle and inferior temporal gyri. Almor and colleagues interpret these results as evidence for generation of new referents for repeated names and for increased integration effects when a newl y created referent is analogous to a previously created discourse referent and must be integrated into the discourse model. Present Studys Contributions As previously mentioned, this experiment was conducted with the aim of looking at brain areas involved in reference problems in disc ourse, and whether th ere were hemispheric asymmetries in the activation see n. The limitations of the literatu re, discussed above, stem from inconsistencies as to task, imaging methods, a nd confounded materials; additionally, this single aspect of discourse processing has not been so lely investigated befo re. Many of the studies presented so far have had lexical confounds in th eir materials, which precluded their use in the current study. For example, in Ferstl and von Cramons (2001) study, one of their stimuli was The lights have been on since last night. The car doesnt start While this does require that the reader generate an inference that the lights refer to the headlights of the car (and investigates the inferencing sub-component process of discourse) the introduction of an other noun phrase (or a repeated noun, as shown in Almor et al.s study ) and the creation of another referent in the discourse would confound our reference investigati on. By utilizing quantifiers, whic h are able to be used without an accompanying noun (see above), we can avoid this effect. Hence, a set of materials was designed with several factors in mind, based on the results of three offline pretests. These are detailed in the next chapter. 27

PAGE 28

28 Predictions In accordance with previous studies on anomalous referent processing and new referent creation, we expected there to be significant activation in the superi or and inferior frontal gyri, as those areas have been implicated in inferencing, semantic integration and conflict (Ferstl & Von Cramon 2001). We expected larger activations in general for the right hemisphere (as compared to the LH), if the right hemisphere actually does play an increased role in constructing discourse models. Additionally, we expected activation in th e median frontal lobe of the left hemisphere because of its involvement in plausibility evalua tions (Stowe et al. 2004). We also expected to find activation in the thalamus and anterior cingul ate if their activation, as McMillan et al. (2005) state, does pertain to selective attention. As stated above, Robertson et al. (2000) found significant activation in the right prefrontal cortex for a more coherent version of discourse (utilizing definite articles vs. i ndefinite articles). If the quan tifiers we used are interpreted similarly to definite articles, we can expect activation there as well. In the current study, participants had to relate the first quantity to the second for the plausibility task which might lead us to expect some activ ation relating to working memory (WM) for both conditions. We would not see a differential activation, however, because this has to be done for all trials. Activation in the RHs inferior parietal cortex (IPC), which has been implicated in magnitude comparison (McMillan et al. 2005, Cohen et al. 2000), therefore would not be differentially activated for one condition over another, as this pr ocess is required of al l conditions for the task the participants had to perform.

PAGE 29

CHAPTER 3 METHODS Participants Eighteen monolingual speakers of English (9 female; aged 18-26, mean age 22) participated in this study. A ll were right-handed and had norma l or corrected-to-normal vision. Based on responses from health and education questionnaires, none had any history of brain trauma, language or learning disability, reading a bnormality, substance abuse, or indicated to be on any psychiatric drugs. All par ticipants were compensated for their time and participated according to University of Florida Institutional Re view Board procedure; most were University of Florida students. None of thes e participants participated in any pretest of the materials. Two of the eighteen participants were omitted from the analysis due to equipment malfunction (a female, 18, and a male, 25). Another two (a female, 18, and a male, 24) were omitted on the basis of their accuracy scores (which fell significantly outside the rest of the data). Hence, the data analysis reported below incl udes data from a total of 14 participants. Materials The materials went through three different pret ests, which were primarily directed toward two goals: confirming the experimenters plausi bility judgments and removing any significant differences in reaction time (RT) and accuracy between the conditions. The materials were presented in the same way as the current study (machine-paced, not self-paced), because we were interested in replicating conditions in the MR I scanner. Using self-paced reading would have varied the rate at which partic ipants could have processed the sentences (possibly resulting in a mismatch between stimuli presentation and imag e acquisition) and elicited increased, prolonged activation of the motor cortex. Additionally, we di d not want to present entire sentences at a 29

PAGE 30

time; doing so would not allow us to know which words the participants are viewing at what time, and could potentially lead to eye movement artifacts. Keeping these ideas in mind, sentence pairs we re created. Each sentence began with a quantifier. The total length of a trial was either 13 or 14 word s (two of the distractor conditions had one fewer word than the other conditions). The first sentence in a pair always had six words, which were divided up into pairs for presentation. The second sentence presented the quantifier phrase (e.g. All six ) at once, and then six more words, one per frame. No noun accompanied this second quantifier. Since we do not introduce another noun phrase (m odified by the quantifier) in the second sentence, lexical confounds by the in troduction of such a noun are avoided. Another discourse referent is not automatically created, as readers prefer as subset interpretation (as discussed above). Participants viewed 128 two-senten ce trials in total. Sixty-four were experimental items in which the number in the second sentence either ma tched or did not match the number of the first. Each experimental trial created had an implausible and a plausi ble variation. The experimental materials were Latin-squared: two lists were ultimately created by counter-balancing the conditions across the lists. Each participant saw only one list and therefore one variation of each trial: 32 items for each of the two conditions. Distractor items composed the remainder of th e sets. There were four kinds of distractor items. Twenty-four of the sixty-four distractor sets were constr ucted to allow for a plausible quantifier mismatch, thereby reducing the probabili ty that participants wo uld base their judgment on the second quantifier alone befo re waiting for possible disambigua tion with a restrictive (that) clause. For example, in the sentence Four roofs were repaired this week. All eight that were fixed previously still leak, the participant must wait until the presentation of that to 30

PAGE 31

understand that this sentence does no t refer to a subset of the pr eviously mentioned roofs, and therefore arrive at the correct conclusion. These twenty-four compri sed the first distractor type. Eight of the distractors (the second type) had the same number in both sentences, but ended implausibly. Thirty-two distractor sets used other quantifiers (a ll, few, etc.), with an equal amount ending plausibly and implausibly (types three and four). Table 3-1 illustrates each condition and distractor type. Experimental ite ms and distractors were pseudo-randomized so that no more than two trials of the same condition followed each other. The total set of 128 items was split into five blocks of 25 or 26 tria ls, each block beginning with a distractor item. See Figure 3-1 for a sample block; the colored vert ical bars represent the stimuli, as encoded on the righthand legend of the figure. Procedure The protocol was first created in EPrime, Version 1.1.4, produced by Psychology Software Tools Inc. Eloquence presentation software, created by the Invivo Corporation, was used to control the presentation of the protoc ol and collection of th e behavioral responses. Participants were first shown a brief demo of the task they would be performing on a laptop computer, and the entire scanning procedure was explained to them. Once participants were placed inside the scanner, they were shown a practice block of four trials during the anatomical scan to allow them to become familiar with th e presentation rate and response procedure. They were instructed to minimize movement and focus their gaze on a fixation cross when not actively reading or responding. The sentences were presented in white on a black background and were viewable on a mirror positioned above the participants head wh ich reflected a computer monitor behind him or her. At the beginning of each block, a fixati on cross (+) was presented for 8000ms. Seven hundred milliseconds before the first trial began, the color of the cross changed from white to red 31

PAGE 32

to signal to the participant that a sentence was ab out to be presented. The first sentence was then presented, two words per frame, for 400 millisec onds per frame. The first two words of the second sentence (or one word, in th e case of the third and fourth ty pe of distractors) and every word thereafter was presented for 300ms. Ever y presentation of text was followed by a 200ms gap. The critical word here is the fifth frame of the trial; at this point, it is clear that there is no further modifier. Each trial had ten frames, as indicated by |s in Table 1. Participants were cued at the end of each pair by the 2000ms presentation of a string of three question marks (???), signa ling that they needed to judge whether the sequence of two sentences together were plausibl e or implausible. A plausible response was indicated with the right index finger and an implausible response was indicated with the right middle finger, using an MR-compatible button box. This kept the part icipant actively paying a ttention through the end of the sentence, because nearly a fifth of the sentences became plausible after the quantifier phrase; in order to judge correc tly, attention had to be given. There was an inter-stimulus interval (ISI) of between seven and eleven seconds, varying in half-second increments. During the ISI, the fixation cross was shown; just as at the beginning of each block, before each trial began, the cross b ecame red. The ISI was jittered to allow for a varied sampling rate throughout the hemodynamic response to the critical word. Each experimental block lasted approximately seve n and a half minutes. After the five blocks requiring the reading and judgment task, participants were asked to relax and cl ose their eyes while a final scan was performed. MRI Data Acquisition The scanning was conducted on a 3 Tesla Ph ilips Whole-Body System located at the McKnight Brain Institute at the University of Florida. Particip ants wore earplugs and MR-safe headphones to reduce noise. Anatomical T1-weighted TFE SENSE structural images were 32

PAGE 33

obtained (TR=8.04ms, 144 axial slices, F OV=240 240 144mm, 256 s duration), then functional images were acquired. The functional data were obtained using a T2 -weighted echo-planar imaging (EPI) pulse sequence (TR=2000ms, TE=30ms, flip angle= 80, matrix=3.753 mm, 36 axial slices, ascending interleaved acquisition parallel to the AC-PC plane) sensitive to blood oxygen level dependent (BOLD) contrast. Two hundred and thirty-one vo lumes were acquired over each of the five experimental blocks. At the beginning of each block, two disdaq (dummy RF excitation) pulse sequences were conducted to allow for more st ability in the signal acquired. The imaging computer then produced an RF pulse when the acquisition of data began, time-locked with the presentation of the stimuli. Data Analysis BrainVoyager QX software (BVQX), created by Brain Innovation, Inc., was used in the analysis of the imaging data. Structural and func tional images were first coregistered and then converted to standardized Talairach space. Func tional images were next preprocessed to reduce artifacts. The preprocessing incl uded: (i) slice scan time correc tion, (ii) motion correction, (iii) linear trend removal, (iv) spatial smoothing (4 mm full width at half ma ximum Gaussian kernel applied to reduce differences in intersubject lo calization), and temporal smoothing (high-pass, 3*data points). Significant activity was mapped by using a Ge neral Linear Model (GLM) with fixed effects (FFX). In a fixed effects analysis, the degrees of freedom relate to the number of voxels, as opposed to a random effects analysis (RFX) in which the degrees of freedom are related to the number of participants. Hemodynamic response func tions were modeled to start at the beginning of each trial, with one function per trial; this provided the analysis a standard activation to compare the actual activation to. 33

PAGE 34

The critical cluster size threshold was set to 128, meaning that an activation area had to consist of at least 128 significant voxels to be reported. Clusters that were significantly activated in this analysis then had their coordinate info rmation exported. Talairac h Daemon (TD) software, created by Jack Lancaster and Peter Fox at the Re search Imaging Center of the University of Texas Health Science Center San Antonio (UTHSCSA ), was used to identify regions of interest (ROIs), based on the nearest gray matter. These ROIs were confirmed in the Talairach atlas available in the Liu lab. ROI GLMs were then conducted for each clus ter to determine the specific degree of activation for each. 34

PAGE 35

35 Table 3-1. Example of stimuli used. | indicates the division of frames, as presented to the participant. Critical words in e xperimental items are underlined. Condition Stimulus Plausible (32) Five ships | were in | the port. | All five | had | just | left | on | their | voyage. Implausible (32) Five ships | we re in | the port. | All *three | had | just | left | on | their | voyage. Distractor 1 (24) Eight girls | were on | the squad. | All five | that | competed | last | year | had | graduated. Distractor 2 (8) Eight students | ha ve very | good grades. | All eight | are | failing | all | of | their | classes. Distractor 3 (16) Few lectures | have sparked | my interest. | Most | were | so | boring | that | I | slept. Distractor 4 (16) Some tenant s | have just | moved in. | All | have | lived | here | for | many | years. Figure 3-1. The GLM model used. The conditions ar e color-coded accordi ng to the legend on the right. The white spikes are the HRF response mode led for that condition (implausible (c), in this case); this was done for each condition, so that each had an HRF response modeled. The numbers at the bottom represent the volume numb er, and the numbers on the left axis represent the % of BOLD signal cha nge (such that 1.0 = 100%).

PAGE 36

CHAPTER 4 RESULTS Behavioral Data The behavioral data was analyzed using SAS version 9.1 (Statistical Analysis System). The mean of the reaction times (RTs) for all re sponses was calculated first, and outliers more than two and a half standard devi ations from the mean were repla ced with the value of two and a half standard deviations from the mean. A repeated measures ANOVA was conducted with list as a between-subjects factor a nd condition as a within-subjects factor. Only the List factor approached significance [F(1,12) = 5.28; p = 0.02]. The main effect of condition was insignificant [F(1,12) = 0.33; p > 0.5], as was the interaction between cond ition and list [F(1,12) = 0.5; p = 0.49]. Table 4-1 shows the means and standard deviations for both the RT and accuracy data. The average accuracy for all of th e subjects for the all of conditions was 87.6%; the average accuracy for the critical conditi ons (overall) was 92.7%. A repeated measures ANOVA was conducted for accuracy within the tw o experimental conditions, and found there was no significant difference in accuracy leve ls between the two [F(1,13) = 3.18; p > 0.9]. Imaging Data A General Linear Model was set up, and th e contrast between the implausible and plausible condition was first examined; however, no clusters were significantly activated at either a corrected (q(FDR)) or an uncorrected p value of 0.05. When the opposite contrast was scrutinized (plausible vs. implausible), nine clusters showed significant activation. They are listed in Table 4-3, and visual ized in Figures 4-1 and 4-2. When activation is seen in the RH for this experiment, it is usually bilateral in nature. Figures 4-6 and 4-9 (right and le ft claustrum), 4-5 and 4-10 (right and left middle frontal gyrus), and 4-4 and 4-11 (right and left precentral gyrus) are the bila teral activations found for this 36

PAGE 37

experiment. The sole RH activation seen that was not bilateral in nature was the small but significant (384 voxels; p = 0.000000) act ivation of the right superior temporal gyrus (RSTG). Overall for the plausible > implausible contrast activation is observed more in the LH than the RH (5417 voxels in the LH vs. 2324 voxels in the RH). Effectively, 70% of the activation seen is in the LH. The bilatera l middle frontal gyrus activation is an order of magnitude greater in the LH than the RH (4254 vs. 439 voxels). Th e RPreCG and RClaustrum activations are both somewhat larger than their LH homologues, but overall activation is mo re left-lateralized. In the LPreCG, where we might expect activat ion in general (due to motor response), we find stronger activation for the plausible trials than the implausible trials (see Figure 4-11). While the activation levels for both conditions are the same at the beginning of the stimulus presentation, by the end of the presentation and for 12 seconds afterward they remain more negative for the implausible condition versus th e plausible condition. For the plausible condition, activation remains significantly more positive (vs. the implausible) in the 8-12sec (4-6 scan) and in the 18-20sec (9-10 scan) ra nges, the first corresponding to the hemodynamic response we could expect to see at the presentation of the second referent and the second range approximately occurring for the response seen for the judgment task. Prefrontal cortex activation is also seen for th e plausible > implausible contrast in the left superior frontal gyrus (L SFG BA 6), the left medial frontal gyrus (LMedFG BA 6), the right middle frontal gyrus (R MFG BA 9), and the left middle frontal gyrus (LMFG BA 9). 37

PAGE 38

38 Table 4-1. Analysis of behavioral data, includi ng RT, Accuracy, and standard deviations for each condition. Condition Mean RT St. Dev. Mean Accuracy St. Dev. Plausible 1071ms 605ms 95.3% 3.4% Implausible 969ms 637ms 90.2% 10.3% Table 4-2. ROI FFX GLM Analyses, showing degrees of freedom, mean and standard deviations of activation, t value, and significance level. ROI df Mean Standard error t value p value Overall 16169 10.031 2.38 4.216 0.000025 Cluster df Mean Standard error t value p value R STG 16169 10.217 1.956 5.223 0.000000 R PreCG 16169 9.217 1.861 4.953 0.000001 R MFG 16169 9.501 1.967 4.83 0.000001 R Claustrum 16169 10.327 1.861 5.502 0.000000 L SFG 16169 11.934 2.727 4.376 0.000012 L MedFG 16169 9.43 2.173 4.34 0.000014 L Claustrum 16169 10.903 2.312 4.716 0.000002 L MFG 16169 14.648 2.683 5.46 0.000000 L PreCG 16169 10.031 2.38 4.216 0.000025

PAGE 39

Table 4-3. Activated clusters at q(FDR) = 0.05; cluster threshold size = 128. The s econd through fourth co lumns refer to the coordinates (in Talairach space) of the center of activation for each cluster. Cluster X coor Y coor Z coor Hemi sphere Lobe Anatomical location Brodmanns Area Extent (in voxels) 1 59 -15 -3.2 Right Cerebrum Temporal Lobe Superior Temporal Gyrus Brodmann area 21 384 2 57 4 34 Right Cerebrum Frontal Lobe Precentral Gyrus Brodmann area 6 486 3 39 28 31 Right Cerebrum Frontal Lobe Middle Frontal Gyrus Brodmann area 9 439 4 28 14 11 Right Cerebrum Sub-lobar Claustrum 1015 5 0.74 14 54 Left Cerebrum Frontal Lobe Superior Frontal Gyrus Brodmann area 6 516 6 -5.5 0.74 61 Left Cerebrum Frontal Lobe Medial Frontal Gyrus Broadmann area 6 190 7 -28 16 11 Left Cerebrum Sub-lobar Claustrum 216 8 -44 16 31 Left Cerebrum Frontal Lobe Middle Frontal Gyrus Brodmann area 9 4254 9 -44 -5.8 45 Left Cerebrum Frontal Lobe Precentral Gyrus Brodmann area 6 241 7741 total 39

PAGE 40

40Figure 4-1. The significant areas of activ ation (in orange and blue) in a glass brain; view of the left hemisphere.

PAGE 41

41Figure 4-2. The significant areas of activ ation (in orange and blue) in a glass brain; view of the right hemsiphere.

PAGE 42

42Figure 4-3. The significant activation in the superior temporal gyrus of the right hemisphere (RSTG), along with the event-rela ted averaging plot of the same region. The green line signifies th e time course of activation for the plausible condition, and the red the implausible.

PAGE 43

43Figure 4-4. The significant activation in the precentral gyrus of the ri ght hemisphere (RPreCG), al ong with the event-related a veraging plot of the same region (green signifyi ng the time course of activation for the plausible condition, red the implausible).

PAGE 44

44Figure 4-5. The significant activation in th e right middle frontal gyrus (RMFG), along w ith the event-relate d averaging plot of the same region (green signifying the time course of activ ation for the plausible condi tion, red the implausible).

PAGE 45

45Figure 4-6. The significant activation in the right claustrum, along with the event-related averaging plot of the same region ( green signifying the time course of activation for the plausible condition, red the implausible).

PAGE 46

46Figure 4-7. The significant activation in the left superior frontal gyru s (LSFG), along with the even t-related averaging plot o f the same region (green signifying the time course of activa tion for the plausible cond ition, red the implausible).

PAGE 47

47Figure 4-8. The significant activation in the left medial frontal gyrus (LMedFG), along with the even t-related averaging plot o f the same region (green signifying the time course of activ ation for the plausible condi tion, red the implausible).

PAGE 48

48Figure 4-9. The significant activation in the left claustrum (Lclaus), along with the ev ent-related averaging plot of the same region (green signifying the time course of activation for the plausible conditi on, red the implausible).

PAGE 49

49Figure 4-10. The significant activation in th e left middle frontal gyrus (LMFG), along with the even t-related averaging plot of the same region (green signifying the time course of activ ation for the plausible condi tion, red the implausible).

PAGE 50

50 Figure 4-11. The significant activation in th e left precentral gyrus (LPreCG), along with the event-related averaging plot of t he same region (green signifying the time course of activa tion for the plausible cond ition, red the implausible).

PAGE 51

CHAPTER 5 DISCUSSION Behavioral Data Consistent with the pretests, there were no significant differences in RT or accuracy between conditions. Therefore, it could be r easonably concluded that there was no overall increase in difficulty between processing one versus processing the other condition; this conclusion would be based on offline measures of pr ocessing, but provides evid ence that trials of different conditions did not differ in terms of overall difficulty or processing load. Ferstl (2007) interprets inferior frontal gyrus (IFG) activation as a measure of increasing processing demand or difficulty; coupling this interpretation with our be havioral data, and the fact that we do not find IFG activation for either contrast lends creden ce to the interpretation that both conditions are processed with equal or nearly e qual demands on processing resources. Imaging Data One of the more important results of this study has been the limite d RH activation seen. Even for the plausible condition, in which cohe rent discourse models could be built, RHdominant activation is not seen. With the exception of the RSTG, all RH activations are bilateral in nature. Bilateral Activations These bilateral activations are interesting; their interpretation is up for debate. For the RPreCG, we see greater activation in the RH than the LH, though our participants all used their right hand to respond to the task in all conditions (and therefore, should activate their LPreCG, as the contralateral hemisphere controls moto r response). One possible explanation for the differential LPreCG activation is that participan ts are more certain of their decision for one condition over the other. If this is the case, they might initiall y recruit a bilateral area (the 51

PAGE 52

RPreCG) to try and assist with the decision of which finger to use. Feedback from the RPreCG might in turn increase activation in the LPreCG, leading to the contrast we see. Ferstl, Rinck, and Von Cramon (2005) found ac tivation in the R IFG/PreCG junction for inconsistent > consistent trials, but only in contra sting one of their conditio ns. They attribute this activation to more demands on inferencing proc esses when inconsistencies are present. The current study also found activation in the R PreCG, but for plausible > imp lausible trials. Since the behavioral data show no significant difference in eith er accuracy or RT, we cannot automatically conclude that there was a dramatic increase in difficulty fo r the plausible condition resulting in more activation in this area. What might instead be happening is a decision or integration effect in that the RPreCG would assist with the decisionmaking process of which finger to respond with. Right Superior Temporal Gyrus The activation in the RSTG would seem to s upport the RH discourse processing model, as it does not have a LH homologue, but its extent is small. Meyer, Friederici, and Von Cramon (2000) have implicated the RaTL in assisting it s LH counterpart when processing difficulties occur, but in the present study we find activation more toward th e middle of the R STG, and for the plausible condition as opposed to the implausi ble. Their analysis would not seem to apply here, as we see neither bilateral or LH activation, and according to our behavioral data, participants did not have signi ficantly more trouble with one condition over another. Ferstl, Rinck, and Von Cramon (2005) found activation in the RaTL (slightly more anterior than current results) to be associated with local detection of inconsistencies, at the propositional level of processing. As Ferstl (2007) interprets this finding, we might see this activation due to successful proposition building for the plausi ble condition. For the implausible condition, when participants begin processing the second sentence of trial an d encounter an anomaly, they may stop forming 52

PAGE 53

propositions and integrating th e rest of the sentence. This conclusion is supported by ERP evidence, as well. When readers are presented wi th a combination of a syntactic and semantic violation, they do not produce neural responses for the semantic violation (where an N400 component is normally expected). Only synt actic neural responses (an ELAN and a P600 component) for the violation are seen (Friederici 2001). Prefrontal Cortex The prefrontal cortex makes up the remainder of the areas differentially activated for the plausible condition. With the exception of a small (439 voxels) activation of the RMFG, activation is almost exclusively seen in the LH the SFG (BA 6), the MedFG (BA 6), and the LMFG (BA 9). The LH medial fr ontal lobe has already been im plicated in the plausibility judgment task (Stowe et al. 2004), so it does not come as a total surprise th at there is activation in this area. Ferstl and Von Cramon (2001) also saw activation in this area for coherent > incoherent trials. An interpretation as to w hy there is more activation for the plausible > implausible condition might be the more integrat ory process of judging something as plausible. If a sentence or discourse is implausible, the participant has to recogni ze just one implausible feature to make a decision. On the other hand, to judge a sentence as plau sible, the participant has to check and integrate all features of the se ntence, then make a decision. This activation for the plausible condition might refl ect this type of processing strategy for the task at hand. Kuperberg et al. (2006) make a similar ar gument in looking at anomaly processing. In mapping new information onto information in semantic memory and determining normalcy, these researchers saw more activation in the left frontal and temporal areas to pragmatically anomalous sentences > normal sentences. In fact, they state that it requires more effort to judge a sentence implausible on pragmatic grounds rather than plausible. While that may be true for sentence processing in isolation, this interpretation does not seem to mesh well with our results. 53

PAGE 54

Given the way our experimental conditions were formulated and that participants could potentially judge by the fifth frame the (im)plausibility of the tria l as a whole, it would not be logical to assume participants continued incorpor ating the rest of the se ntence once judgment had passed. We also cued participants to respond, which could be masking th e early resolution of some implausible trials. The remainder of activa tion seen in the prefrontal cortex could be ascribed to inferencing generation (e.g. Ma son and Just 2004, Ferstl and Von Cramon 2001). Xu and colleagues (2005) found activation in this area as well, selectiv ely for the narrative condition (to the exclusion of thei r sentence activation). So this differential activation seen for the plausible condition might not, in fact, be due wholly to (automatic) inferencing processes, bu t also to the final integration of propositions present in the discourse. In other words, because a final construction or interpretation is possible with the plausible condition, and su ch an interpretation cannot be constructed for the implausible condition (as it does not make sense, or allow for a final result to be reached), the activation may reflect this incorporation process of arriving at a final, coherent discourse model. The right hemisphere is undeniably involved in language processing. However, to attribute discourse processing as a whole to the hemisphere is not warranted by the data presented in the literature, nor is that view supported by th e data presented in the current study. Two interpretations are possible with the present data. First, the RH is not as involved in discourse processing as previously thought in that it does not specialize in multi-sentence processing. The second interpretation possible is that the RH plays a larger ro le in discourse processing when there are a larger amount of sentences (and propo sitions) in discourse than the two-sentence discourses used in our study. These two interpreta tions are not mutually exclusive; this second hypothesis also gives support for a le ss specialized view of the processes occurring in the RH. As 54

PAGE 55

55 the amount of sentences increases in a discourse, the WM load increases as well. Activation in this case would naturally be found in the RH (as it has been shown that the RH supports LH processing when processing load increases (Ferstl and von Cramon 2001)). Based on the evidence presented so far, both in the literature and in the present study, we cannot conclude that the right hemisphere specializ es in discourse processing. Limitations and Future Experiments There are limitations in every experiment, and this one is no exception to the rule. We were interested in examining (problems with) refe rence processing in disc ourse, and designed our experiment accordingly. Our results indicate d that the RH does not hold a monopoly over discourse processing as a whole. Du e to the fact that participants were cued to respond at the end of a trial, we cannot be sure at what point they recognized the im plausibility. We therefore do not know (if they recognized it early in the second se ntence) whether they pr ocessed the remainder of the sentence or not. Future experiments coul d help ascertain the answer to this question. Presenting the same materials in the same way but giving participants a diffe rent task might yield insight into this limitation. Fo r instance, instead of asking for a plausibility judgment, participants could be asked about a feature that appears (or does not) in the second sentence (e.g. Had all tenants lived there for many years? (see Table 3-1, distractor type 4)). Another further avenue of research would be investigating exactly wh at areas are activated for the creation of new discourse referents. Such an experiment would have to take into account the automatic generation of referents associat ed with noun phrases, and utilize anaphora (or words such as quantifiers that can function as anaphora). No experime nt has looked at this specific process so far, and the results from such an experiment would allow researchers to begin to understand the beginnings of the conglomerati on of processes which allow humans to process language in context.

PAGE 56

APPENDIX A LIST OF MATERIALS USED Table A-1 List 1, Run 1 Materials, sorted by condition. Condition Stimulus Correct Response b Four drinks are being filled up. All f our will be for the thirsty athletes. 2 b Three skirts are on sale today. All three are way too expensive for me. 2 b Eight ducks are in the pond. All eigh t are eating bread thrown by kids. 2 b Eight workers struggled through the day. All eight will be fired later this evening. 2 b Five squirrels are climbing a tree. All five are holding something in their mouths. 2 c Four tires were leaking some air. All three had already been pa tched last week. 3 b Six cups are in the cabinet. All six just came out of the dishwasher. 2 b Six eggs are in the carton. All si x will be used for the recipe. 2 b Five poems made my girlfriend cry. All five were really stupid in my opinion. 2 c Five senators have addressed the issue. All three are worki ng on very important things. 3 c Six chairs are in the room. A ll eight are for guests to sit on. 3 c Three restaurants have opened for business. All five seem like they will make money. 3 c Eight tiles need to be replaced. All six are covered in very tiny cracks. 3 f1-plaus Three men were building a house. All si x that were watching made many comments. 2 f1-plaus Four roofs were repaired this week. All eight that were fixed pr eviously still leak. 2 f2-impl Six channels are dedicated to children. All six have political documentaries and independent films. 3 f2-impl Three guests did not eat seafood. All three enjoyed the lobster and the shrimp. 3 f2-impl Five lockers are full of school books. All five are not being used at all. 3 f2-impl Five movies were shot in color. All five were filmed in black and white. 3 f3-plaus Few shops sell only kid's clothing. So me have large collectio ns of men's suits. 2 f3-plaus Many people died in the crash. Fe w lived to tell about the accident. 2 f3-plaus These flowers are deadly to animals. All contain toxins that may cause rashes. 2 f3-plaus Some companies pay employ ees almost nothing. Some pay a moderate but reasonable salary. 2 f-4imp Some drivers eat in their cars. Mo st need their tires rotated each year. 3 f-4imp These crackers are made from wheat. Most are filled with chunks of ice. 3 56 Correct Response 2 = plausible, 3 = implausible

PAGE 57

Table A-2. List 1, Run 2 Mate rials, sorted by condition. Condition Stimulus Correct Response b Three sparrows are on the table. All thr ee are begging people for some crumbs. 2 b Six lawyers worked on the case. All si x have just joined the law firm. 2 b Eight sailboats were in the race. All ei ght sunk before they crossed the line. 2 b Three teachers had offered to help. All thr ee would much rather do something else. 2 b Three companies are in the building. All three will be moving away very soon. 2 b Five plates broke during the meal. All five fell off of the kitchen table. 2 c Three viruses have attacked our computer. All f our have already caused large scale problems. 3 c Three journals are in my desk. All five contain all my top secret information. 3 c Six gifts have been opened already. All four were from my brothers and sisters. 3 c Eight surfboards are in th e display. All five just arri ved at the store today. 3 c Eight frogs jumped into the ditch. All f our were looking for some tasty flies. 3 c Six monkeys have climbed the tree. All eight are busy eating lots of fruit. 3 f1-plaus Three toddlers were crying very loudly. All six that were quiet before became annoyed. 2 f1-plaus Eight fans were cheering very loud ly. All six who supported the opponent were quiet. 2 f1-plaus Six cats played with the yarn. All f our that were asleep didn't notice anything. 2 f1-plaus Four storms have passed through here. All three that formed this summer hit elsewhere. 2 f1-plaus Four elephants were performing an act. A ll three that performed la st year had died. 2 f1-plaus Six branches are on the tree. All three that touched the roof have been removed. 2 f1-plaus Three musicians just recorded an album. All five who recorded here before became famous. 2 f3-plaus Some couples are having minor problems A few hardly ever talk to each other. 2 f3-plaus Some peaches were only slightly bruised. Mo st had completely rotted and were trashed. 2 f-4imp Most kings lived a long life. All were poisoned by their own servants. 3 f-4imp Most Mexican sauces are extremel y hot. Few are way too spicy to eat. 3 f-4imp Few fish live in sweet water. Mo st live in rivers, lakes or ponds. 3 f-4imp No problem appears to be unsolvabl e. All seem like they have no solution. 3 57 Correct Response 2 = plausible, 3 = implausible

PAGE 58

Table A-3. List 1, Run 3 Mate rials, sorted by condition. Condition Stimulus Correct Response b Six waiters have asked to leave. All six are looking for other summer jobs. 2 b Five flowers are in the vase. All five have already wilted from the heat. 2 b Four nations have signed the treaty. All four will obey it beginning next year. 2 b Four vines have covered the gate. All four are now growing down the path. 2 b Three policemen are making an arrest. All three are holding down a violent suspect. 2 b Four coaches are pacing the sidelines. A ll four are too nervous to do anything. 2 c Three suitcases were on a cart. All six we re just taken from baggage claim. 3 c Four plays moved the entire a udience. All six were directed towards very young children. 3 c Five airplanes have been serviced today. All three needed major repa irs that took hours. 3 c Three authors are writing a book. All five haven't published in a long time. 3 c Five items are in stock now. All eight ha ve just been received this afternoon. 3 c Eight dishes are cooking right now. All four take a long time to prepare. 3 f1-plaus Eight bikes were on the rack. All four that were very expensive were inside. 2 f1-plaus Five motorcycles raced down the track. A ll eight that crashed earli er could not start. 2 f1-plaus Eight girls were on the squad. All five that competed last year had graduated. 2 f1-plaus Six buses are making their rounds. A ll three at the station are not running. 2 f1-plaus Five kids wanted to go outside. Al l six that were reading didn't want to. 2 f2-impl Eight students have very good grades. A ll eight are failing all of their classes. 3 f2-impl Three pencils have a sharp tip. All three are too dull to write with. 3 f3-plaus Some truckers were awake all night. Others slept soundly in a comfortable bed. 2 f3-plaus Some insects are harmless to human s. Others can kill you with one bite. 2 f3-plaus Most chairs are actually very comfor table. A few are always painful to sit in. 2 f3-plaus Some cups have broken during ship ment. Those in the box are still intact. 2 f3-plaus Most plants had bright yellow flower s. Some had red petals with purple tips. 2 f-4imp Some species live in the arctic. A ll prefer to live in tropical climates. 3 58 Correct Response 2 = plausible, 3 = implausible

PAGE 59

Table A-4. List 1, Run 4 Mate rials, sorted by condition. Condition Stimulus Correct Response b Eight rats are gnawing on something. All eigh t are rooting around in the dumpster. 2 b Four dogs have learned to fetch. All four seemed to have difficulty with it. 2 b Six plants have died this winter. All six have been kept inside all year. 2 b Three trains are preparing to depart. All three are still fi lling up with passengers. 2 b Five cakes are baking right now. All five will be decorated with white flowers. 2 b Eight firemen are in the fire. All eigh t are trying to extinguish the blaze. 2 c Four notebooks are in my backpack. A ll three are full of my math notes. 3 c Eight beers are on the table. All five were just taken from the fridge. 3 c Eight shirts need to be washed. All six may have to be slightly mended. 3 c Three songs have made me happy. All five are my best friend's favorite songs. 3 c Five bears jumped in the stream. All th ree were looking for so mething to eat. 3 c Six programs crashed on the computer. All f our were designed to be very stable. 3 c Four steaks are on the counter. All eight just came off of the grill. 3 f1-plaus Three tents collapsed in the storm. All six that had flexib le poles did not. 2 f1-plaus Three women were at the table. A ll five who came earlier had already left. 2 f1-plaus Four cabs arrived at the airport. A ll eight that were ther e earlier had left. 2 f1-plaus Four girls had shopped all day. All six that were broke stayed at home. 2 f1-plaus Three courses opened in the department. A ll five that were taught before are not offered. 2 f2-impl Eight raccoons were vicious and dangerous. All eight were gentle and loved by everyone. 3 f2-impl Four essays were an exciting read. All four were dull and I lost interest. 3 f3-plaus Most printers aren't working at all. Few are running their jobs very smoothly. 2 f3-plaus These tools are not very expensive. All of them are cheaper than was listed on-line. 2 f3-plaus Few lectures have sparked my intere st. Most were so boring that I slept. 2 f-4imp Most girls were very strict vegetari ans. Few wouldn't eat any meat at all. 3 f-4imp Many people like to talk a lot. Some usually bring their lunch from home. 3 f-4imp Most knives are made of metal. Some are used to sift fine flour. 3 59 Correct Response 2 = plausible, 3 = implausible

PAGE 60

Table A-5. List 1, Run 5 Mate rials, sorted by condition. Condition Stimulus Correct Response b Three roses bloomed in our garden. A ll three will need a lot of care. 2 b Eight mice are running around me. All eight ha ve been hiding during the daytime. 2 b Six researchers analyzed all the data. All six worked on making charts and graphs. 2 b Four designs are in the catalog. All four can be seen on the website. 2 b Three actors read their new script. All three couldn't seem to remember their lines. 2 b Five turtles are swimming unde r water. All five are looking for something to eat. 2 c Three folders are on my desk. All four belong to one of my classmates. 3 c Six pigs are in the pen. All thr ee are eating out of a bucket. 3 c Four packages were shipped la st night. All eight have dead lines at ten this morning. 3 c Six brands have sold out quickly. All thr ee are the most expensive ones available. 3 c Five players are on the court. All th ree had been sitting on the bench. 3 c Four helicopters have conducte d the tour. All six are usually used to cover traffic. 3 c Five prisoners are hoping for release. All eight have much more time to serve. 3 f1-plaus Five paintings had sold very quickly. All three that I like are still there. 2 f1-plaus Six ships were in the port. All four that had left earlier had not returned. 2 f1-plaus Three classes have been very helpful. All four that I took before were useless. 2 f1-plaus Eight copies were smudged with ink. All four that I made earlier were useless. 2 f1-plaus Five singers have asked to perform. All three who had been i nvited had cancelled. 2 f3-plaus Most homes were very badly damaged. Few didn't have a scratch on them. 2 f3-plaus Most boxes are full of clothes. Some are only holding a few things. 2 f-4imp All cutting boards were completely ro und. Some were shaped like a perfect square. 3 f-4imp Some sausages are made of por k. Some are made of raw cow's milk. 3 f-4imp Some people are afraid of snakes. Some are scared of mice and cheese. 3 f-4imp Some classes are taught for free. None can be atte nded without paying anything. 3 f-4imp Some stamps are saved by collectors. They use them to mail their bills. 3 f-4imp Some tenants have just moved in. All have lived here for many years. 3 60 Correct Response 2 = plausible, 3 = implausible

PAGE 61

Table A-6. List 2, Run 1 Mate rials, sorted by condition. Condition Stimulus Correct Response b Six chairs are in the room. A ll six are for guests to sit on. 2 b Three poachers hunted during the night. All three were stopped by the park ranger. 2 b Three restaurants have opened for business. All three seem like they will make money. 2 b Eight tiles need to be replaced. All ei ght are covered in very tiny cracks. 2 b Four tires were leaking some air. All f our had already been patched last week. 2 b Five senators have adressed the issue. All five are working on very important things. 2 c Four drinks are being filled up. All eight will be fo r the thirsty athletes. 3 c Three skirts are on sale today. All f our are way too expensive for me. 3 c Eight ducks are in the pond. All six are eating bread thrown by kids. 3 c Eight workers struggled through the day. All four will be fired later this evening. 3 c Five squirrels are climbing a tree. All thr ee are holding something in their mouths. 3 c Six cups are in the cabinet. All eight just came out of the dishwasher. 3 c Six eggs are in the carton. All three will be used for the recipe. 3 c Five poems made my girlfriend cry. All thr ee were really stupid in my opinion. 3 f1-plaus Three men were building a house. All si x that were watching made many comments. 2 f1-plaus Four roofs were repaired this week. All eight that were fixed pr eviously still leak. 2 f2-impl Six channels are dedicated to children. All si x have political documentaries and independent films. 3 f2-impl Three guests did not eat seafood. All three enjoyed the lobster and the shrimp. 3 f2-impl Five lockers are full of school books. All five are not being used at all. 3 f2-impl Five movies were shot in color. Al l five were filmed in black and white. 3 f3-plaus Many people died in the crash. Fe w lived to tell about the accident. 2 f3-plaus These flowers are deadly to animals. All contain toxins that may cause rashes. 2 f3-plaus Some companies pay employ ees almost nothing. Some pay a moderate but reasonable salary. 2 f3-plaus Few shops sell only kid's clothing. Some have large collections of men's suits. 2 f-4imp These crackers are made from wheat. Most are filled with chunks of ice. 3 f-4imp Some drivers eat in their cars. Mo st need their tires rotated each year. 3 61 Correct Response 2 = plausible, 3 = implausible

PAGE 62

Table A-7. List 2, Run 2 Mate rials, sorted by condition. Condition Stimulus Correct Response b Three viruses have attacked our computer. All th ree have already caused large scale problems. 2 b Three journals are in my desk. All three contain all my top secret information. 2 b Six gifts have been opened already. All six were from my brothers and sisters. 2 b Eight surfboards are in th e display. All eight just arri ved at the store today. 2 b Eight frogs jumped into the ditch. All ei ght were looking for some tasty flies. 2 b Six monkeys have climbed the tree. All six are busy eating lots of fruit. 2 c Three teachers had offered to help. All five would much rather do something else. 3 c Three companies are in the building. All five will be moving away very soon. 3 c Five plates broke during the meal. All eight fell off of the kitchen table. 3 c Three sparrows are on the table. All si x are begging people for some crumbs. 3 c Six lawyers worked on the case. All three have just joined the law firm. 3 c Eight sailboats were in the race. All fi ve sunk before they crossed the line. 3 f1-plaus Three musicians just recorded an album. All five who recorded here before became famous. 2 f1-plaus Eight fans were cheering very loudly All six who supported th e opponent were quiet. 2 f1-plaus Six cats played with the yarn. All f our that were asleep didn't notice anything. 2 f1-plaus Four storms have passed through here. All th ree that formed this summer hit elsewhere. 2 f1-plaus Four elephants were performing an act. A ll three that performed la st year had died. 2 f1-plaus Six branches are on the tree. All three that touched the roof have been removed. 2 f1-plaus Three toddlers were crying very loudly. All six that were quiet before became annoyed. 2 f3-plaus Some peaches were only slightly bruised. Mo st had completely rotted and were trashed. 2 f3-plaus Some couples are having minor problems A few hardly ever talk to each other. 2 f-4imp Most kings lived a long life. All were poisoned by their own servants. 3 f-4imp Most Mexican sauces are extremel y hot. Few are way too spicy to eat. 3 f-4imp Few fish live in sweet water. Mo st live in rivers, lakes or ponds. 3 f-4imp No problem appears to be unsolvable. All seem like they have no solution. 3 62 Correct Response 2 = plausible, 3 = implausible

PAGE 63

Table A-8. List 2, Run 3 Mate rials, sorted by condition. Condition Stimulus Correct Response b Three suitcases were on a cart. All three were just taken from baggage claim. 2 b Four plays moved the entire a udience. All four were directed towards very young children. 2 b Five airplanes have been serviced today. All five needed major repair s that took hours. 2 b Three authors are writing a book. All thr ee haven't published in a long time. 2 b Five items are in stock now. All five ha ve just been received this afternoon. 2 b Eight dishes are cooking right now. All eight take a long time to prepare. 2 c Four coaches are pacing the sidelines. All six are too nervous to do anything. 3 c Six waiters have asked to leave. All f our are looking for other summer jobs. 3 c Five flowers are in the vase. All eight have already wilted from the heat. 3 c Four nations have signed the treaty. A ll six will obey it beginning next year. 3 c Four vines have covered the gate. All three are now growing down the path. 3 c Three policemen are making an arrest. All five are holding down a violent suspect. 3 f1-plaus Five motorcycles raced down the track. All eight that crashed earlier could not start. 2 f1-plaus Eight girls were on the squad. All five that competed last year had graduated. 2 f1-plaus Six buses are making their rounds. A ll three at the station are not running. 2 f1-plaus Five kids wanted to go outside. Al l six that were reading didn't want to. 2 f1-plaus Eight bikes were on the rack. All four that were very expensive were inside. 2 f2-impl Eight students have very good grades. All eight are failing all of their classes. 3 f2-impl Three pencils have a sharp tip. Al l three are too dull to write with. 3 f3-plaus Some truckers were awake all night. Others slept soundly in a comfortable bed. 2 f3-plaus Some insects are harmless to human s. Others can kill you with one bite. 2 f3-plaus Most chairs are actually very comfor table. A few are always painful to sit in. 2 f3-plaus Some cups have broken during ship ment. Those in the box are still intact. 2 f3-plaus Most plants had bright yellow flower s. Some had red petals with purple tips. 2 f-4imp Some species live in the arctic. All prefer to live in tropical climates. 3 63 Correct Response 2 = plausible, 3 = implausible

PAGE 64

Table A-9. List 2, Run 4 Materials sorted by condition. Condition Stimulus Correct Response b Four notebooks are in my backpack. A ll four are full of my math notes. 2 b Eight beers are on the tabl e. All eight were just taken from the fridge. 2 b Eight shirts need to be washed. All ei ght may have to be slightly mended. 2 b Three songs have made me happy. All thr ee are my best friend's favorite songs. 2 b Five bears jumped in the stream. All five were looking for something to eat. 2 b Six programs crashed on the computer. All six were designed to be very stable. 2 b Four steaks are on the counter. All f our just came off of the grill. 2 c Five cakes are baking right now. All three wi ll be decorated with white flowers. 3 c Eight firemen are in the fire. All six are trying to extinguish the blaze. 3 c Eight rats are gnawing on something. All five are rooting around in the dumpster. 3 c Four dogs have learned to fetch. All thr ee seemed to have difficulty with it. 3 c Six plants have died this winter. All eight have been kept inside all year. 3 c Three trains are preparing to depart. All f our are still filling up with passengers. 3 f1-plaus Three courses opened in the department. All five that were taught be fore are not offered. 2 f1-plaus Three tents collapsed in the storm. All six that had flexib le poles did not. 2 f1-plaus Three women were at the table. A ll five who came earlier had already left. 2 f1-plaus Four cabs arrived at the airport. A ll eight that were ther e earlier had left. 2 f1-plaus Four girls had shopped all day. All six that were broke stayed at home. 2 f2-impl Eight raccoons were vicious and dangerous. All eight were gentle and loved by everyone. 3 f2-impl Four essays were an exciting read. All four were dull and I lost interest. 3 f3-plaus Most printers aren't working at all. Few are running their jobs very smoothly. 2 f3-plaus Few lectures have sparked my intere st. Most were so boring that I slept. 2 f3-plaus These tools are not very expensive. A ll of them are cheaper than was listed on-line. 2 f-4imp Most girls were very strict vegetari ans. Few wouldn't eat any meat at all. 3 f-4imp Many people like to talk a lot. Some usually bring their lunch from home. 3 f-4imp Most knives are made of metal. Some are used to sift fine flour. 3 64 Correct Response 2 = plausible, 3 = implausible

PAGE 65

65Table A-10. List 2, Run 5 Materials sorted by condition. Condition Stimulus Correct Response b Six pigs are in the pen. All si x are eating out of a bucket. 2 b Four packages were shipped last night. All four have dead lines at ten this morning. 2 b Six brands have sold out quickly. All six are the most expensive ones available. 2 b Five players are on the court. All fi ve had been sitting on the bench. 2 b Four helicopters have conducte d the tour. All four are usuall y used to cover traffic. 2 b Three folders are on my desk. All thr ee belong to one of my classmates. 2 b Five prisoners are hoping for release. All five have much more time to serve. 2 c Three roses bloomed in our garden. All six will need a lot of care. 3 c Eight mice are running around me. All four ha ve been hiding during the daytime. 3 c Six researchers analyzed all the data. All four worked on making charts and graphs. 3 c Four designs are in the catalog. All eight can be seen on the website. 3 c Three actors read their new script. All five couldn't seem to remember their lines. 3 c Five turtles are swimming unde r water. All three are looki ng for something to eat. 3 f1-plaus Three classes have been very helpful. All four that I took before were useless. 2 f1-plaus Eight copies were smudged with ink. All four that I made earlier were useless. 2 f1-plaus Five singers have asked to perform. All three who had been i nvited had cancelled. 2 f1-plaus Five paintings had sold very quickly. All three that I like are still there. 2 f1-plaus Six ships were in the port. All four that had left earlier had not returned. 2 f3-plaus Most homes were very badly damaged. Few didn't have a scratch on them. 2 f3-plaus Most boxes are full of clothes. Some are only holding a few things. 2 f-4imp All cutting boards were completely ro und. Some were shaped like a perfect square. 3 f-4imp Some sausages are made of pork. Some are made of raw cow's milk. 3 f-4imp Some people are afraid of snakes. Some are scared of mice and cheese. 3 f-4imp Some classes are taught for free. None can be atte nded without paying anything. 3 f-4imp Some stamps are saved by collectors. They use them to mail their bills. 3 f-4imp Some tenants have just moved in. All have lived here for many years. 3 Correct Response 2 = plausible, 3 = implausible

PAGE 66

LIST OF REFERENCES Almor A, Smith, D, Bonilha, L, Fridriksson, J, Rorden, C. 2007. What is in a name? Spatial brain circuits are used to track discourse references. NeuroReport, 18(12):1215-1219. Binder JR, Frost JA, Hammeke TA, Bellgow an PSF, Rao SM, Cox RW. 1999. Conceptual processing during the conscious resting state: a functional MRI study. Journal of Cognitive Neuroscience, 11:80-93. Bookheimer S. 2002. Functional MRI of language: New approaches to unders tanding the cortical organization of semantic processing. A nnual Review of Neuroscience, 25:151-188. Chow HM, Kaup B, Raabe M, Greenlee MW. 2008. Ev idence of fronto-temporal interactions for strategic inference processes during language comprehension. NeuroImage, 40(2):940-954. Cohen L, Dehaene S, Chochon F, Lehericy S, Naccache L. 2000. Language and calculation within the parietal lobe: A combined cognitive, anatomical, and fMRI study. Neuropsychologia, 38(10):1426-1440. Ferstl EC. 2007. The functional neuroanatomy of text comprehension: Whats the story so far? In: Schmalhofer F, Perfetti C (Eds.), Highe r Level Language Processes in the Brain: Inference and Comprehension Processes. Lawrence Erlbaum, New Jersey, pp. 53-102. Ferstl EC, Rinck M, von Cramon DY. 2005. Emotional and temporal aspects of situation model processing during text comprehension: An event-related fMRI study. Journal of Cognitive Neuroscience, 17(5):724-739. Ferstl EC, von Cramon DY. 2001. The role of cohe rence and cohesion in text comprehension: An event-related fMRI study. Cognitive Brain Research, 11(3):325-340. Ferstl EC, von Cramon DY. 2002. What does the fr ontomedian cortex contribute to language processing: Coherence or theory of mind? NeuroImage, 17(3):1599-1612. Frazier L, Clifton CJ, Rayner K, Deevy P, Koh S, Bader M. 2005. Interfact problems: Structural constraints on interpretation? Journal of Psycholinguistic Research, 34:201-231. Friederici A.2001. Syntactic, prosodic, and seman tic processes in the brain: Evidence from event-related neuroimaging. Journal of Psycholinguistic Research, 30(3):237-250. Hoen M, Pachot-Clouard M, Segebarth C, Do miney PF. 2006. When broca experiences the janus syndrome: An ER-FMRI study comparing sent ence comprehension and cognitive sequence processing. Cortex, 42(4):605-623. Kaan E, Dallas AC, Barkley CM. 2007. Processing ba re quantifiers in discourse. Brain Research, 1146:199-209. 66

PAGE 67

Kintsch W. 1988. The role of knowledge in discour se comprehension: A construction-integration model. Psychological Review, 95(2):163-182. Kintsch W, van Dijk TA. 1978. Toward a mode l of text comprehension and production. Psychological Review, 85(5):363-394. Kuperberg G, Lakshmanan BM, Caplan D, Holc omb P. Making sense of discourse: An fMRI study of causal inferencing across sentences. NeuroImage, 33:343-361. Long DL, Baynes K. Discourse re presentation in the two cerebral hemispheres. Journal of Cognitive Neuroscience, 14(2):228-242. Long DL, Baynes K, Prat CS. 2005. The propositional structure of discourse in the two cerebral hemispheres. Brain and Language, 95(3):383-394. Maguire EA, Frith CD, Morris RGM. 1999. The func tional neuroanatomy of comprehension and memory: The importance of prio r knowledge. Brain, 122(10):1839-1850. Mason RA, Just MA. 2004. How the brain processe s causal inferences in text: A theoretical account of generation and integration co mponent processes ut ilizing both cerebral hemispheres. Psychological Science, 15(1):1-7. Mazoyer BM, Tzourio N, Frak V, Syrota A, Murayama N, Levrier O, Salamon G, Dehaene S, Cohen L, Mehler J. 1993. The cortical representation of speech. Journal of Cognitive Neuroscience, 5(4):467-479. McKoon G, Ratcliff R. 1992. Inference during reading. Psychological Review, 99(3):440-466. McMillan CT, Clark R, Moore P, Devita C, Grossman M. 2005. Neural basis for generalized quantifier comprehension. Neuropsychologia, 43(12):1729-1737. Meyer M, Friederici AD, von Cramon DY. 2000. Neurocognition of auditory sentence comprehension: Event related fMRI reveals se nsitivity to syntactic violations and task demands. Cognitive Brain Research, 9(1):19-33. O'Brien EJ, Rizzella ML, Albrecht JE, Halle ran JG. 1998. Updating a situation model: A memory-based text processing view. Journal of Experimental Ps ychology: Learning, Memory, and Cognition, 24(5):1200-1210. Rapp AM, Leube DT, Erb M, Grodd W, Kirche r TTJ. 2004. Neural corre lates of metaphor processing. Cognitive Brain Research, 20(3):395-402. Robertson DA, Gernsbacher MA, Guidotti SJ, R obertson R, Irwin W, Mock B, Campana ME. 2000. Functional neuroanatomy of the cogniti ve process of mapping during discourse comprehension. Psychological Science, 11(3):255-260. Stowe LA, Paans AMJ, Wijers AA, Zwarts F. 2004. Activations of motor and other nonlanguage structures during sentence comp rehension. Brain and Language, 89(2):290-299. 67

PAGE 68

68 Wijnen F, Kaan E. 2006. Dynamics of seman tic processing: The interpretation of bare quantifiers. Language & Cognitive Processes, 21(6):1-1. Xu J, Kemeny S, Park G, Frattali C, Braun A. 2005. Language in context: Emergent features of word, sentence, and narrative comp rehension. NeuroImage, 25:1002-1015. Van Benthem J. 1986. Essays in logical semantic s. Dordrecht: D. Reid el Publishing Company.

PAGE 69

BIOGRAPHICAL SKETCH Nicole A. Chevalier was born and raised in South Florida, where she lived until attending college. She received her Bachelor of Arts in linguistics from the University of Florida in 2006, with a minor in medieval and early modern studie s. She received her Master of Arts degree from UF in 2008.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101203_AAAABG INGEST_TIME 2010-12-03T11:05:06Z PACKAGE UFE0022274_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 19196 DFID F20101203_AAAOAW ORIGIN DEPOSITOR PATH chevalier_n_Page_05.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
127389debe9e9f2e7900cd71599c6191
SHA-1
bd99ea79aca7fc53f948c22c38cee69a8e120a15
25271604 F20101203_AAAOKQ chevalier_n_Page_38.tif
30efef40c1d8f37328217b551111190f
09bd707155937a86fa8a4d4db69aae10fa256a26
1031404 F20101203_AAAOFT chevalier_n_Page_51.jp2
400c0119768dbe8d908dbb0e3d4df586
e7fd1761d77fa2e34746f03d0ec765669f1e5a8b
26929 F20101203_AAAOAX chevalier_n_Page_24.QC.jpg
0798d04f3db539a2cddf745feaa8fec9
8e5ca30968d31c90209c0d63cfdfd435a000a62a
25265604 F20101203_AAAOKR chevalier_n_Page_39.tif
c7cb60c52db5a691ff92158e2d732bf6
1e77b14422c334e65360095a0b6cc9057959e63b
30141 F20101203_AAAOFU chevalier_n_Page_10.jpg
bd93d6b1e19fb5865e56bb2a1fa7eaa6
3aa22340b819ea956b0a22762c17180e8626e155
F20101203_AAAOKS chevalier_n_Page_40.tif
64f29746d205bd9b1f0a0b01fde16d00
c7c541f62d0ed156e8c60a5864adffea24e96b50
1051981 F20101203_AAAOFV chevalier_n_Page_20.jp2
53c095743b8026e5c90cb4d07d7fd476
d5d3a7c2108cff01f666564f2ba233d6c3252755
1051957 F20101203_AAAOAY chevalier_n_Page_60.jp2
58bd3279fa65d26636aab9e31cbb410e
7ea4d5c8636af1e5bb87d195eebc8fad4ad29757
F20101203_AAAOKT chevalier_n_Page_41.tif
cdb8bb8c43d3523a8e9e26b4cd933289
499bf28d953fd01e5cd85f18a59170442ce2b6e1
2211 F20101203_AAAOFW chevalier_n_Page_69thm.jpg
293953cee86c9ec18a1d35feacb731f1
9453a62f743f5e52614906f6a29f36c062fcc9f8
F20101203_AAAOAZ chevalier_n_Page_65.tif
6faeaca31c3be2612e1db00d063ea924
72397c0f31ba598c5266a964d3c39933d36860e4
F20101203_AAAOKU chevalier_n_Page_46.tif
af5d1791b8ada1b63bf56cd87bf1fa51
7aca276b1d118c5326a64f635861d0c4897506b9
25110 F20101203_AAAOFX chevalier_n_Page_36.QC.jpg
2b8b1937b353ba15e4104b12c5a52376
278af5d4e521050ea4d4dc970d7b149f486a8224
F20101203_AAAOKV chevalier_n_Page_48.tif
301d2eb9527c78c53ddc5663622e8ccd
c864795d4c0c5c45f8447ecd94492cfe7e0a03b2
1051966 F20101203_AAAOFY chevalier_n_Page_17.jp2
74bd5d712a4871772ed80e79e560c28c
a3f4d0721178580eae4174248c81f0e2b440a4c7
F20101203_AAAOKW chevalier_n_Page_58.tif
7210ed15a00da37aa3398fb2cd8a010b
678f74c46ed9066f2439807932e172460e696202
98634 F20101203_AAAODA chevalier_n_Page_67.jpg
27a64e62df2270751c36ebfd909b47a3
84d130ffbb7c5fd517adf48ceb1801a4a9ff8b95
1023196 F20101203_AAAOFZ chevalier_n_Page_28.jp2
ffa1123ceba8edd8909d0b7b8e092056
305c93856bef2a1b355f1faea3c56850602a4375
F20101203_AAAOKX chevalier_n_Page_59.tif
89df77795470374f84bf7348200e32f7
60cfca44ab996a2d33d60b57f7d172c540487ade
1051905 F20101203_AAAODB chevalier_n_Page_63.jp2
d5029b7ae339690fa806e3eced0618a4
fe7d919d28df5dab38ff6cc68e39f6bcbdbe4217
F20101203_AAAOKY chevalier_n_Page_61.tif
3dabf0d97c48c4d2a1365cc39d7af859
ed4749e45592774538b6b73a23c3572d5143df9b
1051983 F20101203_AAAODC chevalier_n_Page_29.jp2
5c52481fb9ef7a032bf3db5569bccc1d
e9e2d38a29cfd740f743f64ee41ccdbfe03228e9
11170 F20101203_AAAOIA chevalier_n_Page_03.jpg
b7baf94edb8d283be118b62b20326762
dd64a08aed299555a2aaf7c744d405b2662e084a
F20101203_AAAOKZ chevalier_n_Page_67.tif
457028f83dbd520d7da30595b2632993
c9b8db6c26ae0dfc31cbd74ba417a6bcd55983d7
15726 F20101203_AAAODD chevalier_n_Page_61.QC.jpg
fcd74721b71bd1e5e6400d1f5fd33cbe
60ea9ce45a4ceed95028eb811a684a139788ca21
87123 F20101203_AAAOIB chevalier_n_Page_04.jpg
26fec78b2eb14b8e3791a94d574972fd
c998a80b2ed1ce89283c24259eac8174010058c2
F20101203_AAAODE chevalier_n_Page_62.tif
374b7b727830f1bc4694868242ab4b0e
f354a0f4c364a79405fb107a255ba01704a29f55
102760 F20101203_AAAODF chevalier_n_Page_66.jpg
632fce1e8e0603c56df9db60ed59cd47
cc655e2c0d097b715593e7b13f5829fda90fc39b
4271 F20101203_AAAONA chevalier_n_Page_61thm.jpg
3b48ff299dc3bf8e8fd2295077a216af
9aae21a98548d3a797060eacf21d298ba244022b
24874 F20101203_AAAOIC chevalier_n_Page_07.jpg
94aa747e6a8cb312b04ee62d135a722b
159269a6377465d605f3d85044531e6c8c32ad6b
7312 F20101203_AAAODG chevalier_n_Page_18thm.jpg
49b1c6402d1c752683b230365cdfb507
f2c33bf4c40c51de3892655fe8a03e12b146ca86
15173 F20101203_AAAONB chevalier_n_Page_62.QC.jpg
30616d467d198a41f3bb143a4c8322af
2d6d51bb46eb8374bc03791266b877a983440cea
64875 F20101203_AAAOID chevalier_n_Page_08.jpg
3ac6a50150b74ac1b5d27d44e80c4d66
0fe80bbba0f90ae3cd9b26f9e57a10926d415116
F20101203_AAAODH chevalier_n_Page_36.tif
3a2e163450bce820d2d05fdf3e252a3b
0b5a8b1eba070c43173369cfee3d8f1244d2657b
14736 F20101203_AAAONC chevalier_n_Page_63.QC.jpg
083041a74cff32c105d4ec5ddcfa726b
2cd1f48f0820048579b76386c1a792850fee7c86
108689 F20101203_AAAOIE chevalier_n_Page_09.jpg
f5f4e9cef98c0a29962b7f3de8be219f
d51696341bdc78282ead8e8d0bc8e8efce576ef7
7450 F20101203_AAAODI chevalier_n_Page_26thm.jpg
7927327b6228202c60f95e9fe6504ccb
b093906f7ac5e72179833a1cc6612962be366b68
15057 F20101203_AAAOND chevalier_n_Page_65.QC.jpg
7f01fbd4267228e300cf76608304fdb8
215b47a540695dea4ed22e630d2a53a94d5246a0
39372 F20101203_AAAOIF chevalier_n_Page_11.jpg
a7958d79944a360e8e76e28982b379a4
d4cb695dee294695d3c3e5405d501499a1257ec5
1051870 F20101203_AAAODJ chevalier_n_Page_56.jp2
87499ee5a144b2edd727822d2917a0dd
77383fe8f4a23401a7b369486b2d2c12dea4af5a
29282 F20101203_AAAONE chevalier_n_Page_67.QC.jpg
bc9620b432f6c69553cd4b3ae1a076a2
b0502b7f547e716e99072a35814240f5d6ea95da
40037 F20101203_AAAOIG chevalier_n_Page_12.jpg
278667589c8cb7cbc93f84f01087149f
17246d9e0a2030bdfaad780ae10043ad1401f244
654155 F20101203_AAAODK chevalier_n_Page_10.jp2
bf0aaf52086a7288671cbce8d16558cc
b0c5ba3bd5e2366f84ea5a89302f1efd1ee8c8ce
2266 F20101203_AAAONF chevalier_n_Page_68thm.jpg
c3df9e967b93324641dba5dfa5c13d5d
604552690c43f7b7ab894805a94fa16fae6e99dc
49054 F20101203_AAAOIH chevalier_n_Page_14.jpg
bb0849897eab15fc650bbe1a595057ee
5d99ae58400b4f274dc04b4daf7609c04f2ae62e
6897 F20101203_AAAODL chevalier_n_Page_33thm.jpg
6aee7f3f3a1452f8b2c4f86a7b7247a5
d8486fda093bc2f003cd7febd59bc57c6ee828ee
6268 F20101203_AAAONG chevalier_n_Page_69.QC.jpg
5fa6815912933f3c5a620c2cff169d86
8126d6fce4fbdb427256a499712043eea963921a
83569 F20101203_AAAOII chevalier_n_Page_18.jpg
3c49c0b3fdf774a4d3674756cb440193
2b862a1b70454537e05cbc46aafd8fd93e76af54
F20101203_AAAODM chevalier_n_Page_47.tif
9c41adb1a02a099cf31269448ce1028b
c84193c1b9eae9012fcc937343162ec3539aca57
78560 F20101203_AAAOIJ chevalier_n_Page_20.jpg
73fa156a9611510b5cc1cfd004e5bd5a
34200bf211acdf28375dd3c0eb43f8e54633b17d
F20101203_AAAODN chevalier_n_Page_57.jp2
471dc239f75ef1f06089c1ada5c7563b
6add671dbcd53ecedf7f1faf1d4f22bca41566f7
53837 F20101203_AAAONH UFE0022274_00001.mets FULL
da75d0b904b22dd0337c2cd671529acd
577c5bc26f0760ff4e64d7cb22f3945bfea0c816
84609 F20101203_AAAOIK chevalier_n_Page_24.jpg
a67fdd1415ca1d852c6ea4ddae6bc8d1
f1c7d7fc4fc595892bbb9c932d17a11b9cb7feba
68304 F20101203_AAAODO chevalier_n_Page_06.jpg
9e7e498a48897bda8727795389be067e
a8b52b477d388e003c498f7eeff953064e26895b
85421 F20101203_AAAOIL chevalier_n_Page_25.jpg
8305e43ad164b3cb640213cb24a39cb5
a3b67cd1b16451352d5edc249e9d6c2287514b19
87911 F20101203_AAAODP chevalier_n_Page_27.jpg
3fb279995bce694d2b78bd03f15a4e02
e8ea23d59487275c782be88564feb704c3b2e459
74158 F20101203_AAAOIM chevalier_n_Page_28.jpg
319f875a2bb19b81812c2e18b4be4491
e07e769c5079ad6c9e358c0966c0ae93e362486e
35441 F20101203_AAAODQ chevalier_n_Page_48.jpg
f91a0918551a6a0821e15e6f32ba2ca8
0803cd63a3ce7391eba1141b4ba9f11090398e63
78189 F20101203_AAAOIN chevalier_n_Page_29.jpg
d4d40a3851175295d62b30252d5038c0
e0dfe8528aaa756350412bb6acd1890397c39ec1
2217 F20101203_AAAODR chevalier_n_Page_05thm.jpg
608f456adb1731a2ff01167d70a70038
49f03b459ea7ca009f0fea3b626409e22d9e7f8d
84949 F20101203_AAAOIO chevalier_n_Page_31.jpg
f2d6669a3b5672439747f2c1cba2ca27
01d75647eb37347d7a2374f39e125f4c2d8e3e46
3972 F20101203_AAAODS chevalier_n_Page_64thm.jpg
d7e630bd09cb0e36b30410a0d7a3c5d7
f1156af48c8d2e1de79697d855aa33d79ea1e319
84851 F20101203_AAAOIP chevalier_n_Page_32.jpg
a462089f386ff71d48beefd1421b4b3e
c072787362b33c8690c96947af6f906beefe5cc0
3628 F20101203_AAAODT chevalier_n_Page_42thm.jpg
a9a361de291efcc278ddaaa843f0242b
a2f5803b144206dadd910f69f3d438e371d196d7
81754 F20101203_AAAOIQ chevalier_n_Page_33.jpg
e6ac78124eec01693a30adf925984f1e
f103d0ab0c5757bf5e26b0d303a7833404d479ea
34390 F20101203_AAAOIR chevalier_n_Page_34.jpg
8c4b44c070c4dc331c434c1636cb5f4e
5a77e5f41bc6958f46c87a9b7f1f2a3023df022c
F20101203_AAAODU chevalier_n_Page_17.tif
9ad6eb207621bd2b5df6835dd4d13377
b1e73da2cab2e7166740fa97ab6ce2a47da9b079
79523 F20101203_AAAOIS chevalier_n_Page_36.jpg
8e8296e0c21cd34fe458e65704922445
66bfbe727e25e448eec558e2895a4bb2cc976ee0
3904 F20101203_AAAODV chevalier_n_Page_49thm.jpg
5eeea631b28fd36a162ce29ba5d0412b
267f6d400b4259d223d9f611b78a2024f517f57f
73805 F20101203_AAAOIT chevalier_n_Page_37.jpg
8f0d4c30690868966a089a33009de904
2506a111569682e43e8e7b875fa47f2271a5615f
F20101203_AAAODW chevalier_n_Page_08.tif
0bcb61ad1d79125085bc8004933f7c5d
90702eebb1300a43a67399f9c9e589a9d9de8959
17444 F20101203_AAAOIU chevalier_n_Page_41.jpg
cf51b875eab01d6cf8d6c23e490a9607
91c010c6db6d68c4f45d2da55dfe61c4402ad891
27501 F20101203_AAAODX chevalier_n_Page_27.QC.jpg
a5fd9bc4475e73d439db34cdd5f97105
d0bdb81bb4d007b6577ebf7a5e8371a69a639adb
74956 F20101203_AAAOIV chevalier_n_Page_51.jpg
8293bb265abc0e09f57e4316707d212b
0dd1e6df1463a959d22a3de123099e6d647731ea
573191 F20101203_AAAODY chevalier_n_Page_38.jp2
51c40a92134097f14cef355f12668df5
daa19c8e6226c97c8d9ff7c133c78788318ae3a9
88725 F20101203_AAAOIW chevalier_n_Page_52.jpg
f0f583741a83daef19caa4e117c0096a
e6f55e50dde84ac6b155964518b9975ec89265db
34618 F20101203_AAAOBA chevalier_n_Page_43.jpg
1abcca7a7f43b768f6e20150e8e60c9d
73332ccdca5a751245194e13b741c0ec987b5c90
3726 F20101203_AAAODZ chevalier_n_Page_03.QC.jpg
69cd9a70467da8b3c0b241324d8158e7
a7755e1ceb1a368390124323f3960db776d115ef
59086 F20101203_AAAOIX chevalier_n_Page_56.jpg
5a1478886661cdb5bd63afd4f97894a9
ca079b1d9e0198d555e5631675aaa373e61f954c
1051982 F20101203_AAAOBB chevalier_n_Page_21.jp2
5eeb9a0679e491f9b09fe1476f307389
a8643168eeffb894e78d2fbff614b2a967554cc4
59232 F20101203_AAAOIY chevalier_n_Page_58.jpg
9d2061e3bf55dc706003ba6a514b100a
3bcea0998e4d43e451394f49beba521dd728f7e1
F20101203_AAAOBC chevalier_n_Page_21.tif
0a7ad40154927a2bf8d4c6c0b14c2455
c9b239887f8bc35067b473adeb653a58a83fcc88
F20101203_AAAOGA chevalier_n_Page_56.tif
a2dcedebdbe4b670e916e824bff5d5a9
b5b54acb45148f6aa3df3c55911e13bf45a00156
60431 F20101203_AAAOIZ chevalier_n_Page_62.jpg
785e53b681f417c8b8aebec1155cfa5d
54b0141996c7395f8f5c0bcebe8d6698d4fc5f69
60154 F20101203_AAAOBD chevalier_n_Page_59.jpg
c25071eeab6cb7c4bd713117a4357284
179a4e005efd093fcce6dee718d4f828c38f6922
22921 F20101203_AAAOGB chevalier_n_Page_51.QC.jpg
4feaedc307807e50b9ae17ffc3951ab5
3cf28965dd7959a6a62e825b13a300bcc902a4dd
F20101203_AAAOBE chevalier_n_Page_42.tif
eb701e64a0cd6c68daf85524546ddd58
2df04b65def3259babd90644d2d2ae49b43a2b9b
86876 F20101203_AAAOGC chevalier_n_Page_16.jpg
20df150968af3317574486e15ce5dbed
38977d6ce06b795fd91d08f30915e85271c9fb11
21258 F20101203_AAAOBF chevalier_n_Page_69.jpg
758e11bbc1bc9227a1dfebdb5a8189bb
b0cc0ce416892b67551a191d25a8fd84f4571808
2347 F20101203_AAAOLA chevalier_n_Page_01thm.jpg
df7c9e682a7bbdd5003367cb8b7338d7
a6a042ec23c1c99a8bc4f5d69d152511936352af
32546 F20101203_AAAOGD chevalier_n_Page_42.jpg
03afce8736ac5fb226de707e59b0aa66
386f09b7c395a041de9f38dba051f04dc64d4de7
6519 F20101203_AAAOBG chevalier_n_Page_37thm.jpg
1d6634326e4209accc8ff6db8f6b06fc
345140f8f9bd1558b9c9ad0c4703af0f86fa9c8f
1421 F20101203_AAAOLB chevalier_n_Page_02thm.jpg
51f195bf133156a97567d3a25885b366
761989568c0dc561535962ecb194a58a9ec65a32
1051975 F20101203_AAAOGE chevalier_n_Page_62.jp2
35688d83a6a52d746b15ca1c3cbfddbd
401f4735652bbb7167ae53d753b8beb35d68a7b7
7333 F20101203_AAAOBH chevalier_n_Page_01.QC.jpg
ba8e1fed9fbaee36c941d461620dd72e
240a5eda51bcbe7414dbbf39fb69a334924a2560
27437 F20101203_AAAOLC chevalier_n_Page_04.QC.jpg
41ad71df777fa9247af68daa782bb9f0
929c22b0a63129674fda21f734b5e9d790842a26
F20101203_AAAOGF chevalier_n_Page_20.tif
3011596cdf83642f375bb28117eb30c5
042448dfde63ca176f88fec431cc5f957a42e515
F20101203_AAAOBI chevalier_n_Page_37.tif
5a465ef42731a99bf86399415db7dfa1
794272e8fde020bd47f6a7771a955b73fcb44486
6447 F20101203_AAAOLD chevalier_n_Page_05.QC.jpg
b787144fc170fd74857a9becf46c350c
eb18745e8f957ed1c65ca17ba5e1fb14c7b88e49
857308 F20101203_AAAOGG chevalier_n_Page_42.jp2
87192a93617c515e6f496daa4862a0d0
92073872aa51ea7e4140fd8df2bb6c2c40e407c2
F20101203_AAAOBJ chevalier_n_Page_02.tif
db09df853de56296255b4adef5652c72
0eddd689544e6f2379044ea8b2ca539a3266fe47
18296 F20101203_AAAOLE chevalier_n_Page_06.QC.jpg
2cf6fda00c4598529ee220c8a29a757b
50bf6e934cf062da9be36ca8397d7d43dfbc4771
33913 F20101203_AAAOGH chevalier_n_Page_45.jpg
5df0ca50ccfed26b8c65bc2628066e16
1d296078e76e41db3fd186f9e39000a83b5d7313
3125 F20101203_AAAOBK chevalier_n_Page_02.QC.jpg
be27ec2ca1aee6f923b4fd656577b5a8
b53eda6dd6aef0916e2c91a04c1a7c64a702ad26
29063 F20101203_AAAOGI chevalier_n_Page_66.QC.jpg
73a073a677a8ee1126c056c55154f2cf
caec7934a265f978976252bbfbadb1dd907b20db
35412 F20101203_AAAOBL chevalier_n_Page_46.jpg
9a45658ad86b3bdd9eaedaf53de4b0c3
79f2dfeedd6e02a21054d7eee9b27a882d8cc0d1
2262 F20101203_AAAOLF chevalier_n_Page_07thm.jpg
fd0082fe96d3a7487f8a86b558d9b995
cfe13fbc01d4b9dc6bc2e7a9be9a474e0fc8fbcd
1051986 F20101203_AAAOGJ chevalier_n_Page_61.jp2
b66666dd9ea31dc6acebdb74931eb3ae
2120124cf75dac68226be9e21e691fbf3a0a891f
7493 F20101203_AAAOBM chevalier_n_Page_54thm.jpg
75a05c604c17fb417c54c0cf091bd39b
344f7c8ffd80a3216b08c4fa42dfdb0fbf067a44
18698 F20101203_AAAOLG chevalier_n_Page_08.QC.jpg
1d40af30130c3712031863664df8c162
6615f8ced0e9e695d625b71d230be3bfb5bdd527
3965 F20101203_AAAOGK chevalier_n_Page_43thm.jpg
430f3233a811040334c095ab6e7ac7cb
648e1100591766aa361a4056698e00ff6ad79ab3
23352 F20101203_AAAOBN chevalier_n_Page_01.jpg
d0aa61bfe1d20261b0bd31a4f1df791d
c1b5e41103949a78914fa1815d955d4a156da5c5
7144 F20101203_AAAOLH chevalier_n_Page_09thm.jpg
9800faddf2fcc9457a0a07c886aca782
9a731fc465cb5b564619e6d4f389c3ff782ec50c
7535 F20101203_AAAOGL chevalier_n_Page_52thm.jpg
69f556383c3879827f969f965f7b4500
2642298125ac7bd63431c2376ad6160ad0cd7323
F20101203_AAAOBO chevalier_n_Page_09.jp2
c86c98eebe3b7e5ff2da4a3d79250311
c207c9088c1fc0a92810996dbfc7a500ed27200b
7941 F20101203_AAAOLI chevalier_n_Page_10.QC.jpg
b7ae7b049a3ac574a5e6a16f3e7d4d21
9d517a52158d5418fa75f98552829e249f6d0a97
47313 F20101203_AAAOGM chevalier_n_Page_38.jpg
2c55f70e065e939d94ed0597eec21fac
f8427bee07e0279e9695f1eb3930888ac0a23265
F20101203_AAAOBP chevalier_n_Page_01.tif
0be36ff1af471a8cde1fb0d0075aa56e
e215abb64268651fa78f6e4e2ce99ca2cb4cb58c
2445 F20101203_AAAOLJ chevalier_n_Page_10thm.jpg
83e50f68082ffad7879a3f6213cdd47b
ebd511efee7c2f375fd32eb34ac0305e0ec84454
27361 F20101203_AAAOGN chevalier_n_Page_53.QC.jpg
3a02dd1a81c7be520ffe4cc277d39bd5
88f9cb2678e76c57bc234b56df69e28e7702adbf
60669 F20101203_AAAOBQ chevalier_n_Page_57.jpg
ccbc96ee4214acc28be5e3f1043ffdd3
1e5ce826002268ccae58d973961ac3fae39862a9
4056 F20101203_AAAOLK chevalier_n_Page_11thm.jpg
7c9e3b57eb7859149132e98cc3673eec
c73bf4ddcecb253648d2763691d583479b2e39d3
86850 F20101203_AAAOGO chevalier_n_Page_53.jpg
d69e9259ecea70145d43ee31364a3753
8eb01c2adbd3f21eaff965edbbf657c38111480c
F20101203_AAAOBR chevalier_n_Page_12.tif
d50af40ba78304f285d382004ee19ed5
7e92b78ed6f4f35cd642d2980591b299ad8285c8
13536 F20101203_AAAOLL chevalier_n_Page_12.QC.jpg
27958e12979d7e733bed62548a398d19
c9646be5018aa98897cb95105fe6b2820434e2c8
F20101203_AAAOGP chevalier_n_Page_60.tif
e674210cef3db8e85fc5772db36df1d5
a0264c4573ac4b908d82a8ea63473a993d7c4baf
3398 F20101203_AAAOBS chevalier_n_Page_34thm.jpg
9776e52b2a8dcd45092af4405748632f
c3969b4a6bbf294f922eff21601b428730166416
4212 F20101203_AAAOLM chevalier_n_Page_12thm.jpg
fc09bacbe5e3628214b8398f9dd228f9
02db855eebdc62c76e315485aad0ac21ce8b01f9
F20101203_AAAOBT chevalier_n_Page_10.tif
593cc6c4ba25983746fcdef21ea226be
ef2aa452715840a93dc350a831c2a56bc69cd497
15773 F20101203_AAAOLN chevalier_n_Page_14.QC.jpg
239d4b48ada11634089a397807f0f443
5a67870093ca7771a1d59351ad1a4ff833cf1362
3600278 F20101203_AAAOGQ chevalier_n.pdf
b55a3c2c661e280973da336870fc7f2b
370870c66e4ca5ce2cf49300b5d3dc4b93255b95
15524 F20101203_AAAOBU chevalier_n_Page_19.jpg
306023563ce8fda5d0d96f91ebb1d9ea
f9fa9d61eb50f3903ffd8141d68a30ff24f4bf46
7454 F20101203_AAAOLO chevalier_n_Page_16thm.jpg
7a60ebf2804b546b16f971571ba3feaf
d28deb7516d745af23365b74c0844edac54b3d87
7293 F20101203_AAAOGR chevalier_n_Page_30thm.jpg
4f98f5abe98345939b959730067efd26
0a06574c44330f5299c499fc3970f44e65c0b188
1051873 F20101203_AAAOBV chevalier_n_Page_58.jp2
e2c3faee6bdeb7b5b91dfd5d08e556ff
0355f9b1a073083a7c196ecdcc8e15b7b074d044
25675 F20101203_AAAOLP chevalier_n_Page_18.QC.jpg
296ca7a0ae1c243e1150af6ec44fa51f
da137c41ccc324ba44008766a503455640bfdcf8
24445 F20101203_AAAOGS chevalier_n_Page_15.QC.jpg
88d9b74288376ec13d3c323677e39e91
17786f99d6611e512a721ab6dd02f228c7314520
F20101203_AAAOBW chevalier_n_Page_31.jp2
4ccf2fd24ba9d0fd89d9cda7c76e464c
1978db2adacf19d0e8849c89eace56ff453da6f4
5265 F20101203_AAAOLQ chevalier_n_Page_19.QC.jpg
d09eb3fc09ded01cb38e48e1d6f8dadc
1f0c84f054d0fdfeb299f8ff56423a644f3be0fe
F20101203_AAAOGT chevalier_n_Page_52.tif
3cee856c3f17f3b1ac19a294efe074ed
fd2c742eca8bef9a69cd405b5ef275d60d7d74da
88675 F20101203_AAAOBX chevalier_n_Page_26.jpg
93d643a8efa8e88e612a3c75eb81a917
62798f4f59fa21b7df2d2b2f689ef5680db30377
24733 F20101203_AAAOLR chevalier_n_Page_20.QC.jpg
70cc7540ff65763b0c9647f37095468c
b0c2b161ac8eb039129e7a276c7ecaae4e48ed2a
F20101203_AAAOGU chevalier_n_Page_43.tif
646f20a552d03e4dbb1cd4241f3d8bda
9d569e167f5e1e62410b1e7aae01c43a07353ddd
F20101203_AAAOBY chevalier_n_Page_35.tif
7fa786858278e17a802122e7d44e074c
f3cdb78eecf44288dfaea256c7e04aa1597c2343
6821 F20101203_AAAOLS chevalier_n_Page_20thm.jpg
5beccf77d0bec4cc4ec94e9740ebc163
92cf668ee8ce6e180b5591bd20451376cf321482
27999 F20101203_AAAOGV chevalier_n_Page_54.QC.jpg
d0529cd1c5a6735def49bb9cbb4ff6d9
9a2d8cb5006a635d7827216290068841446a9e56
3995 F20101203_AAAOBZ chevalier_n_Page_45thm.jpg
671043e33fdd61a483b9ff2f22430e3d
0d175d8800e9eedade7cbad491c0f49a4c062b34
27448 F20101203_AAAOLT chevalier_n_Page_21.QC.jpg
2c3eb5ff0c5bd8a2f523a5c7dc4e139e
9a3af206ecd5266bc59e27895f0a771937f85523
1531 F20101203_AAAOGW chevalier_n_Page_03thm.jpg
d2b48ee2fd3c5a88508a73dc15020e7e
0a55b9ed7611c8b22adfc7d92a4a3188f9a9c06c
7666 F20101203_AAAOLU chevalier_n_Page_21thm.jpg
9ab0a87925ddfc1ee5e8215342f94f81
41c255166b0a72b9c4be678edc44617d1ff7a880
11341 F20101203_AAAOGX chevalier_n_Page_44.QC.jpg
56a0de15cae5011c2ff647e8149bb07a
8e13d54108c0e84f3bc2962aa987c30864943e35
28397 F20101203_AAAOLV chevalier_n_Page_23.QC.jpg
da80064cd53811cd9ed8f118b1d7edf3
ec8e619adae56b4c91c796d94610841fa55a2fae
1051929 F20101203_AAAOGY chevalier_n_Page_52.jp2
0914d1439a8c90f4cdab9e3d3a29597e
64962afd4e42d012e48e3a3d98829dfc6444ef92
7369 F20101203_AAAOLW chevalier_n_Page_23thm.jpg
b7db3ee9a988b0947da0e1eeda0fd67f
1e433f10f0fd5a7465cea4f44993c81b05226281
82617 F20101203_AAAOEA chevalier_n_Page_30.jpg
6071c811f68aeafe0124657f17c8b569
8a64b2d5c07bc0b2df01b44eea4b9877d50216b9
5096 F20101203_AAAOGZ chevalier_n_Page_06thm.jpg
be0045b58d508f60fec24b7fc455b609
89dbd9d16752c0a83ced8fa461b30f218144f36c
7540 F20101203_AAAOLX chevalier_n_Page_24thm.jpg
a73303b73eedb99a2fdb0bfa173a384d
b88d9daeee22705b61c1974baa7b75f31a65b873
F20101203_AAAOEB chevalier_n_Page_22.tif
02f848a075db575bfb1d4c29147d1028
17237c77fcf99a58e356dc835fd7192e59ec657c
59558 F20101203_AAAOJA chevalier_n_Page_64.jpg
a80b2ac4e8102004f9f64a3d30a11ee1
83883a40078fd23aaa8db18de5c1ad671afad656
27596 F20101203_AAAOLY chevalier_n_Page_25.QC.jpg
6d8968989843f314c81c545503e0fdb1
180415cb683bf1c3eb6810b9d88c08166166ecaf
11472 F20101203_AAAOEC chevalier_n_Page_43.QC.jpg
afd87ca4773af270ce11a89dd6670888
edbe9f9724a9f7ef65d271c38d94f9559daedbb3
7523 F20101203_AAANZW chevalier_n_Page_27thm.jpg
dc27c2dd6a373621d31620a59868b323
af1824e3dbe3cc3badf880ebc21e046d809717c2
23613 F20101203_AAAOJB chevalier_n_Page_68.jpg
776f261ecbf2d4f39c27b0cd68f5c6c2
0fc0e564e7c0a49a9f11a99ada8b1956e07f5b9c
28439 F20101203_AAAOLZ chevalier_n_Page_26.QC.jpg
a24242bcb6ca9bab635df09b04ccf6a3
038262b5eb95362b5b74dd10097f8ae6732d769a
28802 F20101203_AAAOED chevalier_n_Page_09.QC.jpg
e124c9510943a2eec14d22f985878b1c
cbb05886bd5388a158a3886eba25c6ca0ce21d48
1051977 F20101203_AAANZX chevalier_n_Page_55.jp2
00e25a242406e86a32c10269079afd7e
8f94de729909a125d281a32e67ae30fe0085f70e
57555 F20101203_AAAOJC chevalier_n_Page_03.jp2
7f0ffac798c47981655695d9d1ce7980
d527334a3dc33b91569f2ca04beb07f77a0350cd
27808 F20101203_AAAOEE chevalier_n_Page_16.QC.jpg
a3608fa7f192709e2ad23c0671188e2b
7dd267e06dac88729ba10a5b3c7e718bce6f1773
514451 F20101203_AAANZY chevalier_n_Page_07.jp2
9dc6b211ea3821960b822312350bbb7c
d3c6c51604aea77f4afacddb33d963ad278276a2
F20101203_AAAOEF chevalier_n_Page_63.tif
d477e2aac4a652affdaaf7bfa8d1e1cf
e7fb8bed6f55248b977a457d22ba70193e9f008c
6481 F20101203_AAANZZ chevalier_n_Page_13thm.jpg
881586045e574ec13ff78ebd73db50c5
01a8e9a10166259eb693843bed97fbdb5195b67b
1051968 F20101203_AAAOJD chevalier_n_Page_06.jp2
baea590bdb426e33d45517d357e8051d
a9e63ab6342cb5fc3b59a2b74020fa4eb3dc5520
85423 F20101203_AAAOEG chevalier_n_Page_55.jpg
7060cf4d787bc963e7eb370f3799714c
4761fe7d463bbd4851b0eaa0b1149a001154d5b8
479284 F20101203_AAAOJE chevalier_n_Page_11.jp2
b9041db88b863c0b12ddfa53dbc369d6
f73a3923436403408c0455ab8d33bf4e46c80e76
23962 F20101203_AAAOEH chevalier_n_Page_37.QC.jpg
a261315cf3ed45791f4a1f9d4afef41c
da9ceb06a880d3e303922af6ebfde1625e673bc3
509735 F20101203_AAAOJF chevalier_n_Page_12.jp2
cffa9d87be4baa4249b8436d76adb60d
9bdb82ee8718f9f34fceacee15b11df6bd30066a
11369 F20101203_AAAOEI chevalier_n_Page_39.QC.jpg
4a6bae2c5fcb3b50a81f0b671c939145
0159816d43920c18387da491d9f8207da4b0813a
F20101203_AAAOJG chevalier_n_Page_16.jp2
41faa3b4d953575281148a36fdc0e08f
a998de9978ff14c8193cda0ab8d77303ed4303cb
5298 F20101203_AAAOEJ chevalier_n_Page_08thm.jpg
75e5ac0dd493b4df309d5566bfd22530
4b48f8133e3b1ef3c9d67579e39ce1ea0dec942c
1051971 F20101203_AAAOJH chevalier_n_Page_18.jp2
bac850c1bd41b6787e8ea22c594ffb70
5cc0e5e9e159a0c0ae14db8548032422db9c7080
895616 F20101203_AAAOEK chevalier_n_Page_44.jp2
58bde0ba2ed0fc4f29b7f0d979c4117b
a92449808c6a9e8e1d749b55665e9d635ee772a8
F20101203_AAAOJI chevalier_n_Page_25.jp2
c73acb6cf8253dd0d58ab9ddac66fa20
619aa4dcd9bf5e8d3ab310ad4c276f4ce5a6fdaf
F20101203_AAAOEL chevalier_n_Page_50.tif
0397b5f278306aacbfad0ac81b8b2bf5
7141ecbee804ee6996cc5c031f8269d2cafd667c
1051954 F20101203_AAAOJJ chevalier_n_Page_27.jp2
a4cd35ade5ab9c1f6d12aafd03d44ac8
ffe945704011df5dc3349b4421b31f7622ee65ed
922904 F20101203_AAAOEM chevalier_n_Page_43.jp2
464267f475cdcb496cb77abbd22bf75d
09468970ebdc2266823932130a8bf0a9f3ecd726
1051958 F20101203_AAAOJK chevalier_n_Page_30.jp2
fac2be01f094b1491d85b24ce636b063
84abd3aa1f13d2d755fcabdc6e25fc89293be9ba
7348 F20101203_AAAOEN chevalier_n_Page_31thm.jpg
693b24c385bbc071188a68bf7d79f3b5
62dfe3548c2d574c0dd1e49a5aeae6f445cd723c
F20101203_AAAOJL chevalier_n_Page_36.jp2
6add65629453f10260a3cc6e23461187
b14ec61536fc7f4b0ace125da9f50d69a1aa406b
F20101203_AAAOEO chevalier_n_Page_53.tif
0a570f96ccba1e8f856ad7136e310f1c
78dfa27bd1f1452d5bf352474612f612d409ceff
70686 F20101203_AAAOEP chevalier_n_Page_35.jpg
6779abf16a3a0feea939d70bdd7a7049
1d37fbe6e864ace4c82b4883eea55a09a862bd79
1019640 F20101203_AAAOJM chevalier_n_Page_37.jp2
9ef022e3923f64fe2d5d6c919107943a
d02ed1bb9fa01b6c2e12a950cdcce5242c83bb17
F20101203_AAAOEQ chevalier_n_Page_53.jp2
a9482883a2fd5e9dd9bd42d8c24ae2ac
1db8dbc175111d36f700f3fe9ebce567aa81032f
626850 F20101203_AAAOJN chevalier_n_Page_39.jp2
e999e746eb7761fe02eeac513fa2a8a9
ce62793f84d47fea5bdd7fe7a19a3422f991dbf9
F20101203_AAAOER chevalier_n_Page_13.tif
7b45efd450d1457babee63db3964ae8c
8a840387ae312ca24e6b5d2023f5b528f92be388
908939 F20101203_AAAOJO chevalier_n_Page_45.jp2
73b1c9358acf9214c4c03ca4f7d0e16f
230856ee77fd3f378db83047b8d1c985284154c9
1003465 F20101203_AAAOES chevalier_n_Page_35.jp2
9a9c6adace044ff8630d7d0dfb6aa9fb
8f55f946f865844233e5eee1c1e0104f40825b6c
961305 F20101203_AAAOJP chevalier_n_Page_46.jp2
0ab18dca0e1d30675e128cf60295c75f
6b0ac45a716a4da938762a23142985151ba9f1d6
14945 F20101203_AAAOET chevalier_n_Page_64.QC.jpg
1b16bcd20560a88f28f4d511917b9fff
18117f7a801f95fb0a41d2ed789db04f48ff0764
958211 F20101203_AAAOJQ chevalier_n_Page_47.jp2
965213494781bd78f38fe20a9c5a2fcd
fc444db93c28d532dadd2339b61c30f50f9e53f9
36022 F20101203_AAAOEU chevalier_n_Page_39.jpg
4fb24e013f144bbe9072d18ac0e71780
b895fd8e0ed10936b82a28c0b3b699f079812ce2
869079 F20101203_AAAOJR chevalier_n_Page_48.jp2
c13c77c67b78ac80b5103a40b1b57b11
bb2885a51eef1b4e1ce1b1af01b75f72143ac054
17701 F20101203_AAAOEV chevalier_n_Page_40.jpg
caa301072a58083ccaeb28300f4260fd
5db614e365b61d4f091cfd7404ebadb87c3dc0fd
917637 F20101203_AAAOJS chevalier_n_Page_49.jp2
d5022223ce9262c1c411d14fe0d3ac0d
ff8357b34cb1b035cc1f0d9315622befe6bd77e0
6861 F20101203_AAAOEW chevalier_n_Page_07.QC.jpg
835a365122f25152d12c98c8a103115d
2b6757b49339f801ed3fb550ca38ea66b9e6fcdd
897780 F20101203_AAAOJT chevalier_n_Page_50.jp2
9f5a68457b942186fa88c31836fcb384
8329ddc4fd29f3457ce8661ba661b3ab666645dc
1051979 F20101203_AAAOEX chevalier_n_Page_08.jp2
9cf07f1ecd7bc88d80d48cf6405fc282
4f90656364cc61774ce9447992b256e7f35f0ed9
1051892 F20101203_AAAOJU chevalier_n_Page_54.jp2
b605d88fe106510c9382798a4d2a67a1
cf18d475e08c8814797baddcd532b454717f342b
F20101203_AAAOEY chevalier_n_Page_64.tif
c3e649e9435c4c1e599729c33b7259f1
0225a4e2ae71a99b2e0343b3c4089330730e313f
1051910 F20101203_AAAOJV chevalier_n_Page_64.jp2
fa1f5c26079851a11cf90e04cdf0c2ce
4eabbfb0701b5dd0add136b979b85bc14d107f7d
82457 F20101203_AAAOEZ chevalier_n_Page_22.jpg
4eaca459c02b022017048cfda8318912
d3a149869684f461390cf4094a3efc98da80549d
1051931 F20101203_AAAOJW chevalier_n_Page_65.jp2
571458df0b0b8142ae11fbbe32e9e9f8
c77f5fb2c2d3f19eca62cce59ba708ffffd57f55
228652 F20101203_AAAOCA chevalier_n_Page_01.jp2
fe8b3a34fb2c232fb66d98bdd57cd94d
03490848d9fd1adaee1c3d7e0df4eaa10cbf624c
F20101203_AAAOJX chevalier_n_Page_67.jp2
13059ad0d1d649df49448e5958f33ba3
71cfdd73d73561db26a6f56d4287c0a233818d76
200119 F20101203_AAAOCB chevalier_n_Page_69.jp2
81427cca5567036aafbc095d432d8c1f
962ed8e02ddf82a4fe2e7bef16b8ded7b8ef2770
F20101203_AAAOHA chevalier_n_Page_33.tif
cc1b60a06e3f96837523499160dd89e6
f08dc788017a584b567cb4736c880d6e0dfedda3
249823 F20101203_AAAOJY chevalier_n_Page_68.jp2
02081549fd6fbd9022ce2803a3229213
956d3492923c1582e5f1224b8bb9556eb2805933
35267 F20101203_AAAOCC chevalier_n_Page_49.jpg
7d3bce88fa51cfe394471da009cfa3ff
c22eb1339917907ade5e097b31acd4b061ae456f
F20101203_AAAOJZ chevalier_n_Page_03.tif
f28c8fb46c9e8efbf45c9cc09c4cca7d
7ca4dad2088516709252976a3b24585212f0335f
23129 F20101203_AAAOCD chevalier_n_Page_28.QC.jpg
9dde3951e24cf2111f696e0d23ff1ee5
fbf8e912c9e88af73439fba52d802abf3e429c0d
F20101203_AAAOHB chevalier_n_Page_06.tif
0f1dbeccfc1af5b8cf9626cc0d1a4f37
2781e17695d8e199524b0428da3c98adafa615cd
7181 F20101203_AAAOCE chevalier_n_Page_68.QC.jpg
3719d498bf86a44e130fb5420e3d49a0
8aef58483b5def3f99e3821c42b27fc3e9c293f1
6331 F20101203_AAAOMA chevalier_n_Page_28thm.jpg
c787e8c3eea0d4edcdb2fe1c1004fa32
cacdbc6d4c21f9a3ccd1d15c816f6fdf2a53f94b
637862 F20101203_AAAOHC chevalier_n_Page_14.jp2
98bc28d2536a143b86f4351f6e4b7123
a154446e4fe190b5852b9593e87938b857ba6aaa
26181 F20101203_AAAOCF chevalier_n_Page_22.QC.jpg
4c0f16da586ade0e76c1bb05f77fd060
5d4fee29858c05b30fc00fe7fca68136c6d1d3ab
7338 F20101203_AAAOMB chevalier_n_Page_32thm.jpg
70dff72ad478616d3a2e435375280b09
7449938b9fc12ddba56194ea0b9d3587adac5998
33577 F20101203_AAAOHD chevalier_n_Page_44.jpg
891d32e3309919b4647187b974bf236a
e08bebc49871ec4c761c16d4fe4463347dd7ad13
F20101203_AAAOCG chevalier_n_Page_54.tif
d64c33fcaef5679c5c26d6fd1709a4d5
82406cd458c5a64fde602cf6cc8c4b62875833fb
25016 F20101203_AAAOMC chevalier_n_Page_33.QC.jpg
746c806482d8c57f8f10544e57e06a08
e7e6d24612fb56c1ebf0b3c812ddc05d6bccb759
F20101203_AAAOHE chevalier_n_Page_49.tif
c58e721121dd3a9a10cd4fae9014f7f2
11f85fd236168b8cfda9ad9966b5fbbc8e46fe7e
3378 F20101203_AAAOCH chevalier_n_Page_39thm.jpg
6eb5e71df9e34467a3a4ffce131d8875
aa3188a7be1a7ed076600eb2b461311396647efb
11154 F20101203_AAAOMD chevalier_n_Page_34.QC.jpg
b6d97cfb04b7e768b1df797bc4ce2f9c
83dce999011af969bc0276f48adaee89f2aa9047
59979 F20101203_AAAOHF chevalier_n_Page_60.jpg
dbb720a8fd8289fc59be24b28b030f87
e299a15c9746a083a354d3a7841db3670eea3628
7531 F20101203_AAAOCI chevalier_n_Page_66thm.jpg
ef28e88617c5b1dde1208f7de2075bee
10861d0f7820dcd4d9d7757f247867424c5c8ab0
6003 F20101203_AAAOME chevalier_n_Page_35thm.jpg
8127a26f4e8d1941748b559c61b31c0a
2dff37948611ec492edac23a4c2278aaab937f21
1051965 F20101203_AAAOHG chevalier_n_Page_66.jp2
31979a546b201563ef6e8d25707bb5ae
f202032651bb9ea1d5a40f7e50ce1461167e20a3
F20101203_AAAOCJ chevalier_n_Page_04.tif
a652fbc411fdee34b549c626c60d498f
b0075829c7de5413a4a79ed4bc2a2777264f5ba9
14495 F20101203_AAAOMF chevalier_n_Page_38.QC.jpg
be0a3bfed73b032d8ae9b5d53e1c1a84
84b775646778b8f05acdba8523af400f76b6681f
116592 F20101203_AAAOHH chevalier_n_Page_19.jp2
3062e4f356b73ff622590d1607a6f201
36bc41baea92089e8db2cdd42659fcb9bbb29484
7416 F20101203_AAAOCK chevalier_n_Page_04thm.jpg
17cbb279c684f9eeb555a27b310ccb26
2366b055d6c7e8f911e2a996d916766078eafcca
1051941 F20101203_AAAOHI chevalier_n_Page_04.jp2
8a5423a3b13d295bf933ea17c78dcc27
046d3be5583b354a9e11480a7c86e0c9623da6a9
7229 F20101203_AAAOCL chevalier_n_Page_22thm.jpg
c25ad8fc44fb2e1cc78f728fa8085383
817a054d92d78684ebad0b6d30ed0955a527638a
4012 F20101203_AAAOMG chevalier_n_Page_38thm.jpg
0c4a5e26ae9cd7cae4402b50a2d78aff
9422f2eacd1f3ad454751c6defee4f8a7103ac6b
59159 F20101203_AAAOHJ chevalier_n_Page_63.jpg
4aa6c65ea346c991acdde9004bd379cb
ed0b44f89c5aaf4c241c795f2efee4a97bfba925
6770 F20101203_AAAOCM chevalier_n_Page_36thm.jpg
03e948712ae79dcb337e948f4f820769
094afb224b6c27e97c36bd2fb9122ed894697471
6577 F20101203_AAAOMH chevalier_n_Page_40.QC.jpg
e04c779ae5184b4a84b9b263eb11d787
1f6a78ad0dc95fbd2bada7ab83e702b9fd585165
F20101203_AAAOHK chevalier_n_Page_55.tif
26407d92f923749af151394b1f03fa8e
49f2414155743a69994f57a5cf5e47b4550e97fb
13467 F20101203_AAAOCN chevalier_n_Page_11.QC.jpg
d176552b08b34052c6760a4324cb1543
7206162a0549a3858ae3f5e1697a3da9bbbd9568
1051970 F20101203_AAAOHL chevalier_n_Page_26.jp2
11fb682816d19d464cfc923e3bec819d
246659f22abb75996906084d50b28400818f3815
F20101203_AAAOCO chevalier_n_Page_45.tif
2fe75f9a654997bd06f1767a6837a4fd
751752ee47b27aa553be8d8f046ca7cb0a862ca0
2708 F20101203_AAAOMI chevalier_n_Page_40thm.jpg
67baff4b94bb9037bad1d54e9e926ae7
34d23599fd374077c073198e46ec6de3541d9c6a
88712 F20101203_AAAOHM chevalier_n_Page_23.jpg
0dce0be93144d0f1d3faed61a6ecaaf9
4f3cb621e1e7d154e597327a2ea90d4bf8a54cfb
11841 F20101203_AAAOCP chevalier_n_Page_47.QC.jpg
2e0e0376b7bc9d647f5a5a27cd852def
0ca112890b465b1ed008758e223c53715a8d3915
4019 F20101203_AAAOMJ chevalier_n_Page_44thm.jpg
5fda83113821de549584edab25b1fde1
31d80a54bcc251dfd3d2d8b7d5f86317d7c49cfd
298943 F20101203_AAAOHN chevalier_n_Page_41.jp2
ad7de24ebc89d31ec755cf5562b8de90
9bd78ecd38de2291503475de104794c22fee65f2
21219 F20101203_AAAOCQ chevalier_n_Page_35.QC.jpg
2f10c0f45a5bb8be115815a889a15650
99273c8dbac13501894bcda292b0e5dd3dd550f2
12150 F20101203_AAAOMK chevalier_n_Page_46.QC.jpg
134830a276354ce65a6ad7410e3c8979
723b4624757fd1bbfa7022a5e0aff355665a859f
1051953 F20101203_AAAOHO chevalier_n_Page_15.jp2
23a5a261d9d6a0b920fdca6dc6e3dede
bd7df3774db96f0477a1b694f868dfa45c1e2dc5
11774 F20101203_AAAOCR chevalier_n_Page_45.QC.jpg
73f6ebc874a0f65488fdda34eb7c108e
96f6cbecd8b615cf61e2070b323393abc5fb898c
4065 F20101203_AAAOML chevalier_n_Page_46thm.jpg
ed402ce3beebf680be907bdc63572496
076fd8a160042b0337094517d0cee71d741e0e42
F20101203_AAAOHP chevalier_n_Page_24.tif
be1f36649013dfd55f24c3ce97f0a76d
2b1be96677f8ad47678147c837c1f75581b386b5
F20101203_AAAOCS chevalier_n_Page_23.tif
a2f8f5e325a82e350803ba8d26e99eaf
642de16b6b0418a22eeb5d688fb8378933441630
4029 F20101203_AAAOMM chevalier_n_Page_47thm.jpg
87468995688cfdd0c8674d9991c5d2bb
84c0927d15ba20de64388b5ab498b0652d5fd030
78993 F20101203_AAAOHQ chevalier_n_Page_15.jpg
233eb477b5462b127a04711a942c33f6
bbb4a844ccdad825c97daadfcfbebe0019f822c9
24941 F20101203_AAAOCT chevalier_n_Page_17.QC.jpg
471a5e72ce73d69bc8d9b3c6b779adb8
8790780ef4559ced30f780ecfa529d770f3e2928
11830 F20101203_AAAOMN chevalier_n_Page_48.QC.jpg
084e0ab379bffa32fe6c98a5446d78c4
9d56461fcdfab75d74c3616e12b97f71b3b4710f
14783 F20101203_AAAOHR chevalier_n_Page_58.QC.jpg
aae36fa5ff8ada0e5aa89f90b6542f17
b840e46a39a32aa952fe66e72269c7efe6c33d1b
F20101203_AAAOCU chevalier_n_Page_68.tif
d720c6fae8d3e50c64643b8a535b461a
68c012bd207bbc8114bec994634d207dcae4664c
4058 F20101203_AAAOMO chevalier_n_Page_50thm.jpg
40384c92013a5c9b5cf4a2b481fd1565
e5693cae5c0b8a552bc0c91280efccbbdbd39bf7
86926 F20101203_AAAOHS chevalier_n_Page_54.jpg
6e595647f468b7fdb141dceb68073517
fd504de9d0f065ba81b5e5356f0e9404d6cf38cf
1051951 F20101203_AAAOCV chevalier_n_Page_32.jp2
c61aef7a9a55f5ec29b0b06489695c1b
fcd1fa7e117be37332c381d707dc669a50ebcb64
6677 F20101203_AAAOMP chevalier_n_Page_51thm.jpg
5259c486164d8aac8d65974b2901c251
23d501f68f29930cb7792f4f33b4c78f17c6d696
15151 F20101203_AAAOHT chevalier_n_Page_60.QC.jpg
f40603a05e83aa64cdae1040291f3b29
f94ac7331f76511a9256029ccc380f03c6b4f600
6646 F20101203_AAAOCW chevalier_n_Page_29thm.jpg
8cc9dc657b0a116c530f19995fb26754
33fd5ac1d31b7ff25fbd363f74fe621330bef1b8
7370 F20101203_AAAOMQ chevalier_n_Page_53thm.jpg
0b7fecc0b86989a648afdb1dddd5e0c0
84089e8c6e6758a1ec402371a32165ed18652c1f
6809 F20101203_AAAOHU chevalier_n_Page_15thm.jpg
095ff6e8f850643f1d7dd4439483d6ba
c8593e4c80ab995fe9210790a7869d34fa58b3ec
F20101203_AAAOCX chevalier_n_Page_44.tif
2c63d6f437bc619bb02c774576363ceb
48f35c853d6808af2367569ff41d2b1e08ab8691
27250 F20101203_AAAOMR chevalier_n_Page_55.QC.jpg
b0e22264f4d5b9bc57733588dbdfde2a
258261a0aa5fed26936ba3ec8ccae380809cde35
1051938 F20101203_AAAOHV chevalier_n_Page_24.jp2
66ea6b98ffa3b7db7dc5c43dad2bfc39
db9747bc5715e4a6a320858c5d9fc8d1ee879cf1
F20101203_AAAOCY chevalier_n_Page_33.jp2
29a568887ecb00d3c30a06c4b44d62f7
f80eec12f09909a9fc62efe61a4a493e232a845e
7429 F20101203_AAAOMS chevalier_n_Page_55thm.jpg
2fa95d664fcb40958633475ca59a13d0
8e673a774e8ae7e02bab9bfc1d74026c1a1ad973
77885 F20101203_AAAOHW UFE0022274_00001.xml
0756133867f86d997f1d5f8035b14786
55ac85a8b520951de1c80052bdea21ccad2e3faa
3947 F20101203_AAAOAA chevalier_n_Page_63thm.jpg
e467d38865a8927125d813d106e11a92
58a2b4dbcdfee85ed65f397e8c9766ee24c60819
7778 F20101203_AAAOCZ chevalier_n_Page_67thm.jpg
e8f278f5b3412f0088b33c60f91faf6d
7abb2d043bd6ff4bc3a0f0e324e5b4501c17d150
14618 F20101203_AAAOMT chevalier_n_Page_56.QC.jpg
565d398226fc2fdf912fc65196186e8e
9be33d9c2cc93b73b3b304a299df360238f7ab3d
88800 F20101203_AAAOAB chevalier_n_Page_21.jpg
7bff3ce0eab8c0dccbd9d605eb8bc9d6
6dd70100f6ca80f649971c3e21a992ec690cd52b
3960 F20101203_AAAOMU chevalier_n_Page_56thm.jpg
0a85c76d7a6d008798a7fa7b66ea7089
2b6c4a71f94609fd7d69b715da4bc2d52a4d7177
26149 F20101203_AAAOAC chevalier_n_Page_30.QC.jpg
607da4f18c86843e88d2fe8593ea7eb1
e187ce6cdd3fc36e21a5bcc73eba3b8f7d225441
15212 F20101203_AAAOMV chevalier_n_Page_57.QC.jpg
7cb57734972339a844890aaaddf39093
30dbe2a51cd1b9adfedc5e1ebb4656d0d6c88c98
10089 F20101203_AAAOHZ chevalier_n_Page_02.jpg
17a4d03f08956e08780f97201d89c7fd
b275af3e49e5d661706507edfb10dc107d14371b
62319 F20101203_AAAOAD chevalier_n_Page_61.jpg
7739f30f77aeeb02864810e3d4895563
3dbb5bcaa2f82a9e5c079c0b62d8a51cb1694ce0
3970 F20101203_AAAOMW chevalier_n_Page_58thm.jpg
c7d9328e91bd449d239c5410b3519b20
a83b2ccfab21f2497657140a28b4c5e71c476da0
300784 F20101203_AAAOFA chevalier_n_Page_40.jp2
6f26956be13f32d8ed57f5d9d451821a
d5f32bbaa554918727dc22f47638645673413439
7263 F20101203_AAAOAE chevalier_n_Page_25thm.jpg
1c3fcb10723ad5b07775265952561bc5
60c1fc143b202c149e1b1e83ca7a31caad31acc3
14985 F20101203_AAAOMX chevalier_n_Page_59.QC.jpg
5819da3b4f9925f4b932150177aea7ab
fa7657cff28889ce5cfa5528ae803a2daa0077f4
33575 F20101203_AAAOFB chevalier_n_Page_50.jpg
ea53566e02006dbc61b2c3e58b80e280
f71191668d38aa5c8e4febb0487182c76b21cbae
6934 F20101203_AAAOAF chevalier_n_Page_17thm.jpg
f300892e730521a1dd774d10394f23eb
048117efa784fafec3690d0171593d5c7c9e31a1
F20101203_AAAOKA chevalier_n_Page_05.tif
b57dde56b84d7d8303689210fd6250cd
0791e12afc0812c974b9f319709c925d7f804558
3993 F20101203_AAAOMY chevalier_n_Page_59thm.jpg
f9a8c5ca4fbcaa462515853356f5d69b
1ad887a77b92fd5652bee5e7fdc70a5befa9b226
2482 F20101203_AAAOFC chevalier_n_Page_41thm.jpg
4d26f4e42259522da44341f6f5fccaed
0055b4d58c45024e92af9b41469b25b7bdfbdfe6
F20101203_AAAOAG chevalier_n_Page_23.jp2
cfb83b06e63a82ce24774116c0b3750a
cabdbf92a6a488b26f53aed1e82af9aed3523a38
F20101203_AAAOKB chevalier_n_Page_07.tif
7962c4e5dd34b436a72dcffffb4af80f
d5fcf8ac74599ae4c37e6f1d9a68a93e1185c1bd
3998 F20101203_AAAOMZ chevalier_n_Page_60thm.jpg
153d982e0aa6f54e210b1537885b1515
20e19cf0a02e5711c538fad1d930a8fe303f8107
988093 F20101203_AAAOFD chevalier_n_Page_13.jp2
3091d9b9edeeba6c81ffe57a49dfe7bd
807a46bc794d71d3f65f6ee60276de86c62a9e08
1051962 F20101203_AAAOAH chevalier_n_Page_59.jp2
c8135274dde4b0b95abe9b1a64736ba7
1bffaacb95d1c80bb8b048a7857706ed79291a91
F20101203_AAAOKC chevalier_n_Page_09.tif
803360bb80efd8634bd41284f61e72d7
e618c382dab957025271604e633144546c641d78
F20101203_AAAOFE chevalier_n_Page_51.tif
4c34b65f2c2ef1c820d78f26f3d2d7f0
797606ee5bfca4b5cdc51525b2e6b44cb26a826a
35272 F20101203_AAAOAI chevalier_n_Page_47.jpg
4d3263ae16c9cc2a6bb30752de4e7361
9ebc29556616eef6437062531525b6e937ecac18
F20101203_AAAOKD chevalier_n_Page_11.tif
6a58306b888872e400e57183c0ee4ded
16c7b1c1e3ff43ee85a797162e7a6e34b8e37746
27540 F20101203_AAAOFF chevalier_n_Page_32.QC.jpg
3f46aad74982f06c7ba1904b3415219c
db3e804ec9cb8b42df03b15f75525ff2f8f86f19
4051 F20101203_AAAOAJ chevalier_n_Page_57thm.jpg
898bd6c576c76b5b04f43912e75285f6
8db51f5ecec6a53d231d4d0517913265f8478b17
6483 F20101203_AAAOFG chevalier_n_Page_41.QC.jpg
1b16de99fdd2f52d8c60b70e946c1b69
0a9a32699976266f43f5688167d2378a4f96db94
F20101203_AAAOAK chevalier_n_Page_69.tif
7121282f39cf270c2796660f16a53d16
43e88df4f227e77670a4a662c2a723970e164ac5
F20101203_AAAOKE chevalier_n_Page_14.tif
bc63a8c6c01392fe7627c5acb03a2dae
f2371d88614f066b687f4fa0aa803aaf12e1b409
22798 F20101203_AAAOFH chevalier_n_Page_13.QC.jpg
d8b31663d3e81e2e5aa83bd3c27e1faf
a3a8adb4c17bda1d866fec39289eba9b770b6e48
24398 F20101203_AAAOAL chevalier_n_Page_29.QC.jpg
65b6da9685d09c902602b5146aaa7398
5b02fb62c717202d2b9a562fe3a99520fbe59c36
F20101203_AAAOKF chevalier_n_Page_15.tif
d74906c0f9f2aef72e8b2e85b14eefd8
adc6804547f221380bcadc04dead8b0ae2f9bfbf
3996 F20101203_AAAOFI chevalier_n_Page_62thm.jpg
0190ca82fc13a6cadb1b83943b95b45d
65778e8069cc2fae9d0dd4da5777a84ccf90e922
27779 F20101203_AAAOAM chevalier_n_Page_52.QC.jpg
1be8c6816ff0675a57b2d17011b08a76
3b0642e0ea95522724a37ca4b420e480dc6b0c05
F20101203_AAAOKG chevalier_n_Page_16.tif
9809d74b12acd05a76a207f8c851df96
de9602162cc65cbca60c73258b3383e1aee9da59
10980 F20101203_AAAOFJ chevalier_n_Page_42.QC.jpg
e9e4fde57e4c37dc9694c85949319366
27b53a450ddc7c7b22930d48cf7a50f6dc6b4806
78414 F20101203_AAAOAN chevalier_n_Page_17.jpg
ae35ca42febaed4146d13004c73ff404
e3927851d8f624fc5c52c3c133f566b7a53fce6c
F20101203_AAAOKH chevalier_n_Page_18.tif
8bb835b1ae907a964d7620fdb42dbb4a
8d628608868d5a09d9ca2e2a94c51946402ac1a0
11680 F20101203_AAAOFK chevalier_n_Page_50.QC.jpg
021c8e1bb9e379c8d1c0596900b98fcb
bec4484f64bd9cf3c93b31384cbb4072e92f2291
1858 F20101203_AAAOAO chevalier_n_Page_19thm.jpg
b4dec962135337ee1631f433d00e041b
7a2a387a87b7bf1dc0d1b52cbdbf48a480a84052
F20101203_AAAOKI chevalier_n_Page_19.tif
56ba323a4508aaf34e5df7b4e6afbd8a
5677aebee3871938c923fc6625e1a5f9c65d581e
27590 F20101203_AAAOFL chevalier_n_Page_02.jp2
a9c99298368a4a3a2079bf39d36e5aed
dec3545b1c5d2447b24c3a858094cb8d8caf3158
11705 F20101203_AAAOAP chevalier_n_Page_49.QC.jpg
3aee51cad8f61876276632c08e3d5e32
4bce24d5c37eb96cb5952390fbdd6a0baa70e2d1
F20101203_AAAOKJ chevalier_n_Page_25.tif
f207df1f99f61f3a80ca5231f8813626
33f8fe5e2475561419ed71b2ab2719c7c0176e88
F20101203_AAAOFM chevalier_n_Page_30.tif
d330381c415b5f97ef2c2786b775e36b
999c45902ff1814d1ad7ecefffdd3192826bdfeb
F20101203_AAAOAQ chevalier_n_Page_66.tif
698bb9b5228d231a86c16ef87db6dd4d
ccf6c45d5c6c1eca246172c972c43bcc7dea637a
F20101203_AAAOKK chevalier_n_Page_26.tif
9deab56f31e098f0c81f970716989dc9
36596f872326dade6b2c3a7dd3c4be6e8320bb07
1051911 F20101203_AAAOFN chevalier_n_Page_22.jp2
58c0f33b3184795cf01f66db5b513868
a2671fb99122245eefd76cbf5af5a202091aec59
181842 F20101203_AAAOAR chevalier_n_Page_05.jp2
e7f5493e66849f077a0dfb2fe24b1353
bd5802955ee87e1669fe5e6d5938084137430622
F20101203_AAAOKL chevalier_n_Page_27.tif
166d0baf033d36671942507cbda90f50
937318dac34dab09b6968a0095a2a51f62c10f55
3986 F20101203_AAAOFO chevalier_n_Page_65thm.jpg
ef932d703d19c82c00ecb6b7ecc47adf
7a181807a79f1d5284d12541e4534b76e944c179
F20101203_AAAOAS chevalier_n_Page_34.tif
accab283ac54de6764230e921ac082f7
92421a0232733922c4c0442bf2650bc8a59726be
F20101203_AAAOKM chevalier_n_Page_28.tif
e9018bffefd790e746c7e8bde0e1ebbc
693a2540b7a7d8c516ced08413dbee755e22fc03
4124 F20101203_AAAOFP chevalier_n_Page_48thm.jpg
a6b3752befa7ec38cf2776607d016f77
2c8a782e6ff9362dca97cc94159b42da879107d8
59811 F20101203_AAAOAT chevalier_n_Page_65.jpg
e0d7b6b481c8042d52b9d9ad8102fab6
75b8244a3eb99c95d34d9507e523c8b6e0d28fed
F20101203_AAAOKN chevalier_n_Page_29.tif
719c849c5e653375ff66be6a34c121e9
f53026b00ca3361850d0260b8e9d0ef0565ea1ba
4290 F20101203_AAAOFQ chevalier_n_Page_14thm.jpg
ddf5432d78a6552453eb42da4235251e
7450ca538db30be237caaa637b0e9b6eee4abe32
420770 F20101203_AAAOAU chevalier_n_Page_34.jp2
25c01e2d987ee34555c0aa4d08505120
3181ff70f5d01cf2d9f9e88b432a6764d70a9c65
F20101203_AAAOKO chevalier_n_Page_31.tif
7f10c290f473d223f9ba407dad401fee
e0bc215fc904f282fe7e2cb5341fb91bfdeac7c7
26568 F20101203_AAAOFR chevalier_n_Page_31.QC.jpg
40ef83b5f47c632ef431ed9738101f72
fa90efbc05cb46d76c769694e548d6b2f096a6ac
F20101203_AAAOAV chevalier_n_Page_57.tif
1b238ce9c3df399fe497b28e08a57a02
417a926e40deae6b814df365cfe0d83bce2d57a1
F20101203_AAAOKP chevalier_n_Page_32.tif
b290663f7b94e7619c23e163cb286237
797c370fa389cb87114ee7f9bbbf1bf271adf116
72076 F20101203_AAAOFS chevalier_n_Page_13.jpg
a454ba3bd241c9626861ec0595b0101c
f2ba32d59e2db24bfe49a2648fcdbe1669a2cf74