Citation
Changes in Grammatical Aspects of Aphasic Discourse after Contrasting Treatments

Material Information

Title:
Changes in Grammatical Aspects of Aphasic Discourse after Contrasting Treatments
Creator:
DEL TORO, CHRISTINA M. ( Author, Primary )
Copyright Date:
2008

Subjects

Subjects / Keywords:
Aphasia ( jstor )
Discourse ( jstor )
Gestures ( jstor )
Interrogative sentences ( jstor )
Mazes ( jstor )
Nouns ( jstor )
Sentences ( jstor )
Spoken communication ( jstor )
Verbs ( jstor )
Words ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Christina M. Del Toro. Permission granted to University of Florida to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
7/24/2006
Resource Identifier:
496801928 ( OCLC )

Downloads

This item has the following downloads:


Full Text












CHANGES IN GRAMMATICAL ASPECTS OF APHASIC DISCOURSE AFTER
CONTRASTING TREATMENTS















By

CHRISTINA M. DEL TORO


A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF
FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF MASTER OF ARTS

UNIVERSITY OF FLORIDA


2006





























Copyright 2006

by

Christina M. del Toro















ACKNOWLEDGMENTS

There are many people who have helped and supported me throughout the

completion of my degree and thesis. I first must thank my committee members who have

been instrumental to my academic and professional growth.

Dr. Lori Altmann has taught me so much about aphasia and linguistics that I have

gained a new perspective on language impairments and treatments. Her encouragement,

trust, and support have been integral to my success and enthusiasm.

Dr. Diane Kendall has offered her guidance and knowledge since my undergraduate

studies when I began to develop my skills as a researcher with her help and support. She

has always offered her time to provide opportunities for research and furthering my

knowledge and interest in the field.

Dr. Bonnie Johnson has been my professor since undergraduate studies and taught

my first class in language development. She has always encouraged critical thinking in

the classroom and clinic. Her enthusiasm in the classroom and clinic has been

inspirational.

I could not have completed this study without the help and support of the Discourse

Group at the Language over the Lifespan Lab. Susan Leon, Lynn Dirk, and Elizabeth

Mikell were involved from the beginning of this project and shared in the learning

experiences of discourse analysis.

I would also like to thank Dr. Kenneth Logan who has always been available to

answer my questions about graduate school and the thesis. I thank Dr. Anna Moore for









guiding my growth as a researcher and allowing me to take time from her project to focus

on my thesis.

I thank my fellow students and friends whose support and encouragement I greatly

appreciate. I could not have enjoyed graduate school and writing a thesis without the

wonderful friendship of Michelle Troche who inspires me with her hard work and great

achievements.

Finally, I thank my family whose love always inspires me and reminds me of what

is important in life. I thank my sister and roommate Jennifer del Toro for her love and

daily support that kept me going.

I have enjoyed my time in undergraduate and graduate studies at the University of

Florida and look forward to staying for my doctoral studies to further my academic and

professional skills.
















TABLE OF CONTENTS

page

A C K N O W L E D G M E N T S ......... .................................................................................... iii

LIST OF TA BLE S ......... .... ........ .... .... ...... ..................... .. .... vii

LIST OF FIGURES ......... ...... ..................................... .. ................. viii

ABSTRACT .............. ......................................... ix

CHAPTER

1 IN TR OD U CTION ............................................... .. ......................... ..

A phasia and D discourse ................................................. ... ......... .... ............... 1
D iscourse A analysis .................. ...................................... .. ........ ..
Theoretical fram ew ork ........................................................ ............. ..2
Inform ation analyses .................................. .....................................4
Elicitation m ethods ......................................... ...................... ........ 7
M ethodologies ........................................................ ........ ......... 9
Limitations and deficiencies in current research........................................10
Purpose .................. ...... .............. ... ....................... ...............11
R research question 1............................................ ............................... 12
Research question 2............. .................................... .... .............. ......12
Research question 3 ............. .................................... .... .............. ......12
Research question 4............. .................................... .... .............. ......12

2 M E T H O D S .......................................................................................................1 3

Participants .................................................... .. ...............13
Procedures for Semantic-Phonologic Treatment and Gestural + Verbal Treatment.. 14
G ram m atical A analysis ....................... .................. ... .... .. ....... .... ... ... 15
S a m p le ................................................................................... 1 5
S corin g ...................................... ............................ .... .......... ...... 15
R eliab ility ........................................................................ 18
Statistical A nalysis................................................... 18

3 R E S U L T S ........................................................................................................1 9

Treatm ent Type: SP versus G V ...................................................................... .. .... 19



v









W ord L evel ............................................................................................ ........ 19
S en ten c e L ev el................................................................................................ 2 0
Inform ation Level .................. ............................ .. ...................... .. 21
W ord Trained: N oun or Verb .................................. ......................................21
W ord L evel ............................................................................................ ........2 1
S e n te n c e L ev e l................................................................................................ 2 1
Inform action Level .............. .. ........................ ..................... .............. 22
Gestural-Verbal Treatm ent: N versus V ........................................... ............... 23
W ord L ev el ................................................................................................... 2 3
S en ten c e L ev el................................................................................................ 2 4
Inform ation L evel ................ ...............................................25
S P N v e rsu s V ............................................................................................................. 2 5
W ord L ev el ................................................................................................... 2 5
S en ten c e L ev el................................................................................................ 2 6
Inform ation Level .... ........................................... .............................26
C o rre latio n s ........................................... ... ... ....................................2 7
S P T re atm e n t .................................................................................................. 2 7
G V T reatm ent ................................................................. ................. 28

4 D ISC U S SIO N ............................................................................... 29

Treatm ent Type......................................................... 29
Trained-W ord Type ............................................32
Trained-W ord Type within Treatm ent Type ........................................ .............. 33
SP N oun v s. V erb ............................................................33
GV Noun vs. Verb .................................. .......................... ..........34
M a z e s ............................................................................... 3 6
L im itatio n s .......................................................................................3 7
S u m m a ry ...................................................................................................... 3 8

L IST O F R E FE R E N C E S ............................................................................. 40

B IO G R A PH IC A L SK E T C H ........................................................................................ 45
















LIST OF TABLES

Table pge

1. P participant D em graphics ........................................................................................ 14

2. W ord Classes and Sentence Type Codes........ .......................................... 16
















LIST OF FIGURES


Figure pge

1. Treatment type: A comparison of the percentage of participants with increased
production of measures with significant increases.............................. ...............20

2. Word-type trained: A comparison of the percentage of participants with increased
production of measures with significant increases............................................22

3. Gestural-verbal treatment: A comparison of the percentage of participants with
increased production of measures with significant increases..............................24















Abstract of Thesis Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Master of Arts

CHANGES IN GRAMMATICAL ASPECTS OF APHASIC DISCOURSE AFTER
CONTRASTING TREATMENTS

By

Christina M. del Toro

May 2006

Chair: Lori Altmann
Major Department: Communication Sciences and Disorders

The aim of the current study is to compare the effects of two treatments on the

production of discourse of speakers with aphasia using grammatical analysis that

quantifies changes in the production of various grammatical units and forms before and

after treatment. Discourse analysis is an important method for studying the language

system because it provides a view of the language system in its natural setting of

conversation. There are several reasons for using discourse analysis as an outcome

measure of aphasia treatments. Most importantly is that impairment of discourse is the

most notable deficit and most troubling for the patient with aphasia. In addition, the goal

of any aphasia treatment should be improved discourse production; thus, the effect of

treatment on discourse is key to establishing successful treatments. Analyses of the

current study included word classes, sentence types, and information units. Analyses are

conducted to compare the two treatment types, the two conditions, and each condition

within each treatment type. Discourse data are taken from pre- and post- treatment

measures of two naming treatments. Participants were assigned to either the semantic-









phonologic treatment or the compensatory gestural-verbal treatment for aphasia; within

each treatment, participants were trained on noun access or verb access. Results indicate

greater increases in production of various word classes, sentence types, and information

units in the participants of the gestural + verbal treatment with verb training. A new

measure of information is introduced and compared to current information measures used

in discourse analysis. The significance of mazes is discussed in reference to a possible

link to lexical access. Implications for future discourse analysis are discussed.














CHAPTER 1
INTRODUCTION

Aphasia and Discourse

Aphasia is a language disorder resulting from damage to the brain significantly

affecting all levels of language production: form, content, and use. Deficits are observed

through errors of word retrieval, phonological processing, grammar, and syntax (Duffy,

1995). In the clinical and research setting, it has been very common to use standardized

tests to measure changes after treatment. However, the tasks involved in these measures,

repeating sentences, naming pictures, and following verbal directions, are not tasks

performed in everyday conversations (Boles, 1998). In Elizabeth Armstrong's (2000)

review of aphasic discourse analysis, she points out that discourse analysis is an

important method for studying the language system because it provides a view of the

language system in its natural setting of conversation compared to language tasks such as

naming.

There are several reasons for using discourse analysis as an outcome measure.

Most importantly is that impairment of discourse is the most notable deficit and most

troubling for the patient. Studying the effects of aphasia on discourse is also fundamental

to classifying the type of aphasia and developing a treatment plan, as improved

conversational skills should be the goal of any treatment. Furthermore, discourse

analysis is important from a theoretical view because all linguistic levels of language

interact in discourse providing an opportunity for developing and testing models of









normal language production, and patterns of impaired discourse can support or disprove

current models (Armstrong, 2000; Prins & Bastiaanse, 2004).

As discourse analysis has come to the forefront of aphasia research, many different

theoretical views and methods of analysis have been developed. Through the years

limitations and challenges in the use of discourse have been discussed, such as clearly

defining discourse, identifying the components of discourse to measure, and developing

elicitation techniques that yield relatively natural language samples (Armstrong, 2000).

Discourse Analysis

Elizabeth Armstrong (2000) identifies the key questions researchers aim to answer

through discourse analysis. These questions concern identification of the following: the

kinds of meanings conveyed by speakers with aphasia, the lexical and grammatical

resources used to convey meaning and identifying when meanings are no longer clear due

to lexical and grammatical deficits in aphasia. Armstrong (2000) identifies two

theoretical frameworks for defining and analyzing discourse to answer these questions:

structuralist-oriented and functionalist-oriented.

Theoretical framework

The functionalist-oriented perspective, defines discourse as language in use

(Goffman, 1981; Halliday, 1985a, 1985b). This framework is focused on the meaning of

discourse within its context. Functionalist analysis is concerned with the macrostructure

of discourse such as topic maintenance, turn-taking, appropriateness to the situation or

topic, and the ability to organize and convey meaning. Several researchers have found

that speakers with mild to moderate aphasia are able to employ the structural principles of

discourse used by normal speakers such as setting, complicating action, and resolution in

a narrative, and obligatory elements of procedural discourse, although optional elements









of procedural discourse are most often omitted (Glosser & Desser 1990; Ulatowska,

Freedman-Stem, Doyle, Macaluso-Haynes, & North, 1983; Ulatowska, North, &

Macaluso-Haynes, 1981). As the structuralist framework will be employed for the

current study, the functionalist framework will not be discussed in further detail here.

The structuralist-oriented framework defines discourse as a component of language

above the sentence (Grimes, 1975; Harris, 1963, 1988). In this framework, discourse is

analyzed through its structural and lexical components, sentences, phrases, and words,

commonly called microstructure. Lexical components have been studied from a semantic

perspective, measuring occurrence of paraphasias and non-specific lexical items, and

from a grammatical perspective, measuring types of word classes produced (Armstrong,

2000). Syntactic analysis has focused on grammatical complexity of sentences, syntactic

errors, and clause argument structure (Bird and Franklin, 1996; Brenneise-Sarshad,

Nicholas, & Brookshire, 1991; Goodglass, Christiansen, & Gallagher, 1993; Schwartz,

Saffran, Bloch, & Dell, 1994; Miceli, Silveri, Romani, & Caramazza, 1989; Roberts &

Wertz 1989; Saffran, Sloan-Berndt, & Schwartz 1989;).

Word class analysis focused on the structural difficulty in discourse of speakers

with agrammatic aphasia has found production of more nouns than verbs, and fewer

closed class words and pronouns when compared to normal speakers. Syntactic analyses

of discourse of speakers with agrammatic aphasia have found production of verbs with

the simplest argument structure and omissions of several structures including subject,

main verb, or required function words and inflections. Word class analysis of discourse

of speakers with fluent aphasia found production of more verbs than nouns (Berko-

Gleason et al., 1980). Syntactic analyses of the discourse from speakers with fluent









aphasia have found a decrease in syntactic complexity and the frequency and variety of

verbs (Bastiaanse, Edwards, & Kiss, 1996; Edwards, 1995; Edwards and Bastiaanse,

1998). The structuralist framework is adopted in the current analysis of word classes,

syntax, and information units.

Typical methods for eliciting discourse in the structuralist-oriented framework

include single picture description, retelling stories in response to a series of pictures,

retelling of previous accounts, retelling known fables, or monologues based on topics

such as family, illness, or occupation. The current discourse samples were collected

using several of these elicitation techniques.

Information analyses

An important aspect of discourse applicable to both of the above frameworks is

content and efficiency of language production (Armstrong, 2000). Measuring content or

the amount of information conveyed by the speaker, has been argued as the best

measurement for successful discourse production (Shadden, 1998). Researchers have

used several terms to refer to this unit of measurement, including 'content units' (Meyers,

1979; Yorkston & Beukelman, 1980); 'themes' (Berko-Gleason et al. 1980) 'correct

information units' (Nicholas & Brookshire, 1993a), 'main concepts' (Nicholas &

Brookshire, 1993b, 1995), 'essential information units'(Chemey & Canter, 1993; Hier,

Hagenlocker, & Shindler, 1985; Nicholas, Obler, Albert, & Helm-Estabrooks, 1985),

'propositions' (Ulatowska, North et al., 1981) 'essential and optional steps' (Terrell &

Ripich, 1989; Ulatowska, Freedman-Stern et al., 1983; Ulatowska, North et al., 1981;),

'target lexemes and thematic units' (Gleason, Goodglass, Obler, Green, Hyde, &

Weintraub, 1980), 'unscorable or nonessential content' (Tompkins, Boada, McGarry,

Jones, Rahn, & Rainer, 1993; Trupe & Hillis, 1985) and 'entire utterance' (Arbuckle,









Gold, Frank, & Motard, 1989). Although all these terms refer to measures of information

they are not equivalent.

In her chapter on information analyses, Shadden (1998) discusses the many

differences among information measures. She begins with the two approaches, 'a priori'

measurements and 'a posteriori' measurements. 'A priori' analysis refers to methods that

measure pre-determined essential components of the discourse, as in Yorkston and

Beukelman's content units (1980), and Nicholas and colleagues' essential information

units (1985), both of which used descriptions of the Cookie Theft picture (Goodglass,

Kaplan, & Barresi, 2000). An advantage to the a priori method is that results can be

compared to normative data; however, these norms are only applicable to the particular

picture used, and results cannot be generalized across pictures or across other elicitation

tasks. 'A posteriori' analysis focuses on defining measures that can be applied across

tasks and behaviors to develop computed measures with significance across these tasks.

An example is Nicholas and Brookshire's correct information unit (1993a; Shadden,

1998). Although the number of words and number of CIUs can only be compared across

identical tasks, words per minute, CIUs per minute, and percentage of CIUs can be

compared across different tasks. Another difference Shadden discusses is the idea of

measuring only the amount of information or including quality of information (1998).

Quality of information can be included in a priori measures as Myers (1979) did with

Yorkston and Beukelman's content units (1980) by dividing the content units for the

Cookie Theft picture into literal and interpretive. A third difference in information

analyses is the inclusion of efficiency in the measure. This can be done by calculating

information over time (e.g., CIU per minute, Boyle, 2004; Shadden, 1998;), number of









language units (words or syllables) per information unit, or using rating scales (e.g.,

Trupe, & Hillis, 1985) (Shadden, 1998).

Past definitions of information measurement units have specified that information

must be relevant to the topic, but have not required utterances to be coherent. Therefore,

these measures only conveyed the amount of relevant words in the discourse and not the

manner in which they were conveyed (Armstrong, 2000). For example, Nicholas and

Brookshire (1993a) considered relevance when defining correct information units (CIU)

but only in terms of the relevance of the individual words to the topic. This is

problematic, as Hasan (1985) points out, because words can be relevant to the topic, but

be incoherent in conveying information. The following utterances are given by Hasan

(1985) to illustrate this limitation, "Girl bananas two spend shopkeeper/Apples own girls

dollars grapes/Buy fifty sell cents shopkeepers/Girls fruit." Although these utterances are

relevant to the topic of shopping, they are not coherent and, therefore, are not useful for

describing the difficulties speakers of aphasia have in conveying information (Armstrong,

2000). CIUs are scored per word and not per utterance allowing each word conveying

information to be measured including auxiliary verbs, determiners, conjunctions and

other function words. However, due to differences in utterance length or number of

words produced in aphasia types, CIUs cannot be compared across aphasia types. These

scores are inflated for fluent aphasia and depressed for non-fluent aphasia, even if the

amount of information is similar. In the current study the information measure will be

referred to as 'utterances with new information' or UNIs, a measure created by the

Discourse Group at the University of Florida Language over the Lifespan Lab, and

defined as a coherent utterance providing information not previously given in the









conversation. An utterance did not have to be a grammatical sentence or include more

than one word to be counted as a UNI; the only criteria was that it provided new

information that was semantically coherent with the preceding context. The UNI is

scored at the utterance level allowing for comparison across aphasia type, unlike the CIU.

However, the UNI may be limited in detecting small increases in information as it does

not measure each word providing new information, and an utterance may include more

than one piece of new information.

Elicitation methods

Researchers have elicited various types of discourse ranging from descriptive

narrative, procedural discourse, or even role-play using a variety of methods: single

picture description, picture sequence description, retelling of a story read to the speaker,

retelling of well-known stories, recounting a memorable experience, or talking about

personal events and family members (Armstrong, 2000). Procedural discourse requires

given topics such as brushing teeth or going grocery shopping (Armstrong, 2000).

Although all of the above methods result in conversation-like verbal production, unlike

reading aloud, word or sentence repetition, or picture naming, they are not to be

considered equal in the discourse they produce and in fact produce different 'genres' of

discourse (Armstrong, 2000; Ulatowska et. al., 1981, 1990; Williams et al., 1994).

Within a genre, differences in definition have been found, such as in the narrative genre,

the most commonly used in research (Armstrong, 2000). Studies referring to their data as

narratives have included participants discussing family members, single picture

description, and retellings of personal events, fictional narrative, and known narratives

such as 'Cinderella' (Glosser & Deser, 1990). These types of language samples differ









with respect to their inclusion of the defining elements of narrative: orientation,

precipitating action and resolution (Armstrong, 2000).

In their review of discourse research, Prins and Bastiaanse (2004) make the

following divisions in defining and eliciting discourse: (1) semi-spontaneous speech from

situational pictures (e.g., Cookie Theft) or retelling a known story (e.g., Cinderella), (2)

semi-spontaneous speech elicited by role-playing, (3) spontaneous speech in a

conversation or dialogue (e.g., between patient and spouse, therapist or other person

familiar to the patient), and (4) spontaneous speech elicited through interview with open-

ended questions in an informal, natural conversational setting giving the patient ample

time to talk. These divisions may eliminate some of the differences Armstrong (2000)

points out, although even these divisions group potentially different genres together, such

as discourse from pictures and retelling a known story. Glosser, Weiner, & Kaplan

(1988), found pictures elicited less verbal complexity than spontaneous production

without pictures. However, Doyle et al. (1998), found no significant differences between

story retelling with or without pictures for measures of verbal productivity, information

content, grammatical complexity, verbal disruption, and quality of grammatical form.

Another explanation for these genre differences could be the instructions given to the

speakers. Olness (2005) suggested that the common instructions given for picture

descriptions, "Tell me everything you see going on in this picture" does not specifically

ask for temporal organization. In her study contrasting instruction type, she found

speakers more likely to produce narratives when asked for temporal sequencing, "Make

up your own story about what happened, with a beginning, a middle, and an end." As

Armstrong (2000) states, the issue of defining discourse and matching genres with









elicitation techniques remains to be resolved and finer distinction may help profile the

way speaker's with aphasia use language skills in various settings.

Methodologies

Prins and Bastiaanse (2004) review the current methods of discourse analysis and

outline the advantages and disadvantages of such analyses. They identify two methods of

analysis: (1) rating features of the language produced on a pre-determined number of

scales, such as phrase length and grammatical forms, and (2) quantifying linguistic

variables, such as mean length of utterance, and content/function words ratio. They

advocate the use of linguistic or grammatical analysis as they view aphasia as a linguistic

disorder. They state that any macrostructure impairments or pragmatic difficulties in

discourse of speakers with aphasia are most likely due to linguistic impairments. Prins

and Bastiaanse (2004) point out that discourse can be analyzed using either qualitative or

quantitative methodology.

Qualitative linguistic analyses use rating scales such as the Rating Scale Profile of

Speech Characteristics (RSPSC) and the Aphasia Severity Rating Scale (ASRS) both

used in the Boston Diagnostic Aphasia Examination (BDAE; Goodglass et al., 2000;

Prins & Bastiaanse, 2004) to classify aphasia type. The spontaneous speech portion of

the BDAE presents questions such as, "How are you today?" and "Tell me what

happened to bring you to the hospital?" and then provides the Cookie Theft picture for

the patient to describe (Goodglass et al., 2000). The rating scales are used to analyze the

patient's responses to the questions and picture description. The RSPSC uses a 7-point

scale to measure melody, phrase length, articulatory agility, grammatical form,

paraphasias, and word finding. Based on ratings on the RSPSC, a patient's aphasia type









can be identified (Prins & Bastiaanse, 2004). The ASRS is a 5-point scale measuring the

speaker's ability for verbal communication.

Quantitative analyses can be elicited from various stimuli as previously discussed,

such as story retell or picture description. A commonly used method is the Quantitative

Analysis of Agrammatic Production (QAAP; Saffran, Berndt, & Schwartz, 1989). This

approach measures variables such as proportion of closed class words, verb inflections,

well-formed sentences, and embedding index. Rochon, Saffran, Sloan-Berndt, and

Schwartz (2000) conducted reliability testing of these measures and found test-retest

variability ranging from .66 to .92 for the proportion well-formed sentences. Prins and

Bastiaanse (2004) suggest this finding is not due to instability of the QAAP, but to

instability in the performance of agrammatic speakers themselves. They also suggest that

the findings indicate a need for caution when using the QAAP to measure changes in

agrammatic speech. Thompson, Lange, Schneider, and Shapiro (1997), offer another

method of quantitative analysis of agrammatic speech, in which they analyzed verb-

argument structures. Based on their results, they conclude that agrammatic speakers

produce fewer verbs in general and produce verbs with simpler argument structure when

compared to non-brain-damaged speakers. Although this study added to the knowledge

of the underlying disorder in Broca's aphasia, it has yet to be replicated in speakers with

other aphasias. Consequently, it is not possible at this time to determine the broad

clinical implications of this finding (Prins & Bastiaanse, 2004).

Limitations and deficiencies in current research

Prins and Bastiaanse (2004) note that a primary limitation in the value and

generalizability of discourse analysis for aphasia research is the speaker with aphasia.

The stability of the production of aphasic discourse across repeated testing is unknown;









therefore, the value of language samples as representations of the speaker's ability or as

treatment outcomes is unknown. Prins and Bastiaanse (2004) present the need for more

group studies to develop norms for statistically reliable improvement, so that individual

studies could be better interpreted. In addition, as in the case of the method proposed by

Thompson et al. (1997), analyses need to be applied to various types of aphasia to

determine if research findings provide a method applicable in clinical settings.

A significant reason for the limited research available on grammatical analysis is

the nature of such analysis. It is time-consuming and demanding of resources, skills, and

knowledge not always available in clinical or academic settings. Researchers must have

significant knowledge of linguistics and aphasiology to conduct the requisite grammatical

analyses and apply the results to clinical practice (Prins & Bastiaanse, 2004).

Despite these limitations in the use of quantitative analysis, Prins and Bastiaanse

(2004) consider it an important method for investigating discourse. Like Armstrong

(2000), they advocate using discourse analysis for developing theoretical models,

designing therapy and measuring its efficacy, because it provides the best information on

a patient's everyday use of language.

Purpose

The aim of the current study is to compare the effects of two treatments on the

production of discourse using grammatical analysis that quantifies changes in the

production of various grammatical units and forms before and after treatment. Analyses

address changes at the word, sentence, and information levels. Analyses will be

conducted to compare the two treatment types, the two conditions, and each condition

within each treatment type. Participants (Raymer, Ciampitti et al., in press; Raymer,

Singletary et al., in press) were assigned to either the semantic-phonologic treatment or









the compensatory gestural-verbal treatment for aphasia; within each treatment,

participants were trained on noun access or verb access.

Research question 1

Will gestural verbal treatment have a greater effect on the production of discourse

than semantic-phonological treatment? It was hypothesized that changes in discourse

measures will be higher in the compensatory gestural-verbal treatment than the semantic-

phonologic treatment.

Research question 2

Will training one word type (e.g., nouns or verbs) lead to increased production of

that word type? It was hypothesized that word-type trained will result in increased

production of that word type.

Research question 3

Will increasing noun and verb production correlate with increases in information?

It was hypothesized that increased noun and verb production will correlate with increases

in UNIs.

Research question 4

Will production of 'good sentences' correlate with increased information? It was

hypothesized that increased production of 'good sentences' will correlate with increases

in UNIs.














CHAPTER 2
METHODS

Participants

Seventeen individuals with aphasia between the ages of 38 and 81 participated in

this study. Their demographic information is shown in Table 1. All participants

completed an experimental treatment study by Raymer, Ciampitti et al., (in press) and

Raymer, Singletary et al., (in press), contrasting a semantic-phonologic treatment (SP)

and a compensatory gestural and verbal treatment (GV). Each treatment was divided into

a noun or verb based treatment. Discourse samples elicited pre- and post- treatment were

available from five participants from the semantic-phonologic treatment and twelve

participants from the verbal-gestural treatment.

Eligibility for the treatment study was based on diagnosis of a unilateral left

hemisphere cerebrovascular accident (CVA) confirmed by structural MRIs. The CVA

must have caused aphasia that continued for more than four months prior to participation

in the current study. All participants had to demonstrate a word retrieval impairment of

less than 75% accuracy for nouns and verbs as measured by the Western Aphasia Battery

(WAB; Kertesz, 1982), the Boston Naming Test (BNT; Kaplan Goodglass, & Weintraub,

2001) and the Action Naming Test (ANT; Obler, & Albert, 1986). In addition,

participants could not have motor speech impairments greater than moderate severity,

determined by scores greater than 2.0 on the WAB repetition subtest. Participants

represented a variety of aphasia types and severity as noted on the WAB. All participants

spoke English as a first language.









In addition to these measures, participants receiving the gestural and verbal

treatment were given the Florida Apraxia Screening Test-Revised (Rothi, Raymer, &

Heilman, 1997). Results of this test showed that all participants had mild to moderate

limb apraxia except one who demonstrated a severe limb apraxia. The table below

describes the order of treatment for each participant, word type, and demographic

characteristics.

Table 1. Participant Demographics
Participant Treatment Word Type Age Education Aphasia Type WAB Score
1 Semantic Noun 38 12 Nonfluent 53.9
2 Semantic Noun 68 14 Nonfluent 44
3 Semantic Verb 74 18 Nonfluent 59.9
4 Semantic Verb 66 14 Fluent 77
5 Semantic Noun 81 12 Nonfluent 74.5
6 Gestural Verb 56 10 Nonfluent 31.6
7 Gestural Noun 70 10 Nonfluent 33
8 Gestural Noun 69 8 Nonfluent 54.5
9 Gestural Verb 73 14 Fluent 78.2
10 Gestural Noun 67 12 Nonfluent 68.7
11 Gestural Verb 40 16 Nonfluent 45
12 Gestural Verb 64 12 Fluent 58.5
13 Gestural Verb 49 12 Nonfluent 58
14 Gestural Verb 50 14 Nonfluent 38
15 Gestural Noun 70 14 Fluent 21.3
16 Gestural Verb 52 12 Fluent 43.4
17 Gestural Noun 51 12 Nonfluent 65.2
From: Raymer, Ciampitti et al., in press; Raymer, Singletary et al., in press

Procedures for Semantic-Phonologic Treatment and Gestural + Verbal Treatment

The treatment followed a multiple baseline design across participants and stimulus

sets. Training took place 2-4 times per week for a total of 10 sessions. Daily probes were

administered in 8-10 sessions. SP training sessions consisted of a word retrieval

treatment in which the examiner introduced the target picture and provided the name for

the participant to repeat three times. With the picture displayed, the examiner asked four

yes/no questions, two about semantic characteristics and two about phonologic









characteristics of the target word. This procedure was used to aid the participants in

developing a strategy that follows the normal word retrieval process.

GV training sessions consisted of four steps for each of the 20 treatment stimuli.

The treatment began with the clinician presenting the picture and verbally modeling the

target word and gesture matching the picture. The participant then repeated the target

word and gesture three times. The clinician then performed the gesture without the word

and the participant imitated three times. If necessary, the clinician aided the participant

in manipulation of the limbs. The third step required the clinician to display the target

verb and the participant to verbally repeat it three times. Then the clinician repeated the

word again, syllable by syllable if necessary. After pausing for five seconds, the clinician

prompted the participant to say the name and perform the gesture. If correct, the

participant was reinforced; if the participant was incorrect, the correct model was given.

In total, the participants attempted each target nine times per training session.

Grammatical Analysis

Sample

As part of the treatment study described above, participants produced

conversational samples in response to scripted questions before beginning treatment and

after completion. The questions between the caregiver and participant were about food,

events, and hobbies; questions between the clinician and participant focused on a set of

pictures of family members, famous people, and historic events. Conversations were

videotaped and transcribed by an examiner blind to the treatment condition.

Scoring

Conversational samples were analyzed using Systematic Analyses of Language

Transcripts (SALT; Miller & Chapman, 1991). In addition, the QAAP (Saffran, Berndt,









& Schwartz, 1989) was used to guide the choice of syntactic structures for coding. The

table below provides the word-classes and sentence types coded in this analysis.

Table 2. Word Classes and Sentence Type Codes
Word Level Code Sentence Level Code
Nouns Produced [NP] Minimal Sentence [SM]
Pronouns Produced [PRO] Good Sentence [SG]
Modifiers Produced [MP] Elliptical [SEL]
Modifying Noun [MN] One Word Response [S1]
Modifying Verb [MV] Question [SQ]
Verbs Produced [VP] Irrelevant Response [RI]
Word Level Code Sentence Level Code
Auxiliary Verbs [VA] New Information [IN]
Verb Infinitive [VN]
Intransitive Verb [VI]
Transitive Verb [VT]
Ditransitive Verb [VDI]
Automatic Speech [AS]
Phonological Error [EPH]
Semantic Error on Verb [EV]
Semantic Error on Noun [EN]


Dialectal differences such as the personal pronoun "I" pronounced as the southern

"ah" and transcribed as "ah" were coded appropriately. Contracted words such as

"wanna", "dunno", and "gonna" were coded for each word. For example "dunno" was

coded as [va][vp][vt] for "don't know." As part of the SALT program, utterances

identified as mazes were placed in parenthesis and not coded or used for word or

utterance count. Mazes include repetitions, self-corrections, incoherent words strings,

and filled pauses such as "um" and "uh." "You know" and "like" were also considered

mazes if not relevant to the ongoing statement or question. For example when "you

know" did not mean "understand" and "like" did not mean "preference" or have semantic

content. Repetitions included self-repetitions and repeating the examiner or caregiver.

Self-corrections were defined as any repetition of a phrase or word with a change. For









repetitions and self-corrections, only the last iteration of the phrase or word was coded

(Miller & Chapman, 1993).

Discourse was also scrutinized for utterances with new information (UNI), defined

as a coherent utterance providing information not previously given in the conversation.

This was used as a sentence level code but did not indicate the entire sentence as new

information. For example, a participant may have previously stated he had a son and

then stated he had a son and a daughter.

Good sentences included the following structures: noun +main verb, noun + copula

+ adjective, noun + verb + noun, and noun + copula + prepositional phrase (preposition

and noun phrase). Minimal sentences were those missing an obligatory word or

inflection, but semantically qualifying as a sentence (Saffran, Berndt, & Schwartz, 1989).

Elliptical sentences were those judged to be appropriate answers in conversation but

missing a major component; for example, "To the mall" as a response to a question,

"Where did you go yesterday?" One-word responses were any one word direct answer to

a question such as "yes", "okay", or "vanilla." Questions were defined as any request for

information that was not a repetition of a question asked by the examiner or caregiver.

Irrelevant responses were those not related to the question or topic of discussion

including neologisms. This sentence level code most often accompanied the "automatic

speech" word-level code used for statements commonly used by a participant in response

to any question or statement. "Automatic speech" could be an isolated statement or part

of a statement with relevant information. All counts were based on number of instances

per utterance, in order to control for differences in discourse length between samples.









Reliability

Reliability was conducted during development of the coding system and initial

implementation. The coding system was developed by the Discourse Group at the

University of Florida Language over the Lifespan Lab, consisting of a linguist, two

speech pathologists, and a speech pathology undergraduate student. Transcripts were

independently coded by all members. Afterward, the group discussed the current codes,

the application of codes, and any changes that were necessary until a complete coding

system with agreed upon definitions was complete. Reliability of the transcript scoring is

currently in progress.

Statistical Analysis

SPSS was used to analyze the coded transcripts. Separate analyses were conducted

for treatment type (semantic-phonologic versus gestural + verbal), trained word type

(nouns versus verbs) and trained word type within each treatment (gestural + verbal noun

versus gestural + verbal verb). Means, standard deviation, and standard error of mean

were calculated. Changes in performances were tested using the Wilcoxon Signed Ranks

Test. Correlations between change scores (e.g., post-score minus pre-score) were also

calculated.















CHAPTER 3
RESULTS

The results of the discourse analyses are reported in three ways, by the percent of

people who changed on a given measure, illustrated in Figures 1-3, using Wilcoxon Sign

Rank statistics and correlations of change scores. In many of the treatments, the number

of patients was very small, so findings with p-values of .25 and below are reported.

Treatment Type: SP versus GV

Word Level

Both SP and GV treatments had positive effects on the number of nouns per

utterance, as shown in Figure 1. Among participants from the SP treatment, 3 of 5

participants produced more nouns after treatment; however, this did not reach

significance, (Z = -.674; p > .50). Among participants in the GV treatment, 8 of 12

produced more nouns post treatment which was marginally significant, (Z= -1.804, p <

.08).

Only recipients of the GV treatment showed an increase post treatment in the

number of modifiers produced per utterance with 10 of 12 participants showing this

increase, (Z= -1.833;p = .06). In contrast, only 1 of 5 participants receiving the SP

treatment increased on this measure, (Z < 1). See Figure 1.

There were no other significant changes due to treatment type in other word level

measures: pronouns per utterance, verbs per utterance, or automatic statements per

utterance.












90


80


o 70

o
S60-

ao
50
.e Semantic-Phonologic Trt
Gestural-Verbal Trt
40

a-
S30
o
S20-


10


0
Nouns Produced Modifiers Produced One Word Response Good Sentences
Word Structure

Figure 1. Treatment Type: A comparison of the percentage of participants with increased
production of measures with significant increases.

Sentence Level

The SP treatment increased one-word responses in 4 of 5 participants, with a trend


toward significance, (Z= -1.483; p < .20). The GV treatment decreased one-word


responses in 8 of 12 participants, with marginal significance, (Z= -1.647; p <.10). The


GV treatment increased number of good sentences for 8 of 12 participants, however it


was not significant (Z= -1.098; p = .272). The SP treatment increased number of good


sentences for only 1 of 5 participants, (Z<1). See Figure 1.


There were no other significant changes due to treatment type in other sentence


level measures: number of elliptical sentences per utterance, number of good sentences









per utterance, number of questions per utterance, or number of irrelevant responses per

utterance.

Information Level

There were no significant changes attributable to differences in treatment type in

information measures: mean length of utterance in words, type token ratio, percent maze

words, or UNIs per utterance.

Word Trained: Noun or Verb

Word Level

Noun and verb based treatments had positive effects on the number of nouns and

modifiers produced in discourse, as shown in Figure 2. Noun-based treatment (SP or GV)

increased noun production for 5 of 8 participants with marginal significance, (Z= -1.660;

p = .093). Verb based treatment increased noun production for 6 of 9 participants, but it

was not significant, (Z < 1).

Production of modifiers increased for 5 of 8 participants of the noun-based

treatments with a trend toward significance, (Z= -1.260; p = .208). Verb based

treatments increased production of modifiers for 6 of 9 participants, but it was not

significant, (Z < 1). See figure 2.

There were no other significant changes due to treated word type in other word

level measures: pronouns per utterance, verbs per utterance, or automatic statements per

utterance.

Sentence Level

There were no significant changes due to treated word type in sentence level

measures: number of one-word responses per utterance, number of elliptical sentences

per utterance, number of good sentences per utterance, number of questions per utterance,








or number of irrelevant responses per utterance. However, when looking at all acceptable

responses per utterance (good sentences, one-word responses, plus elliptical sentences),

noun-based treatments increased these responses in 6 of 8 participants with a trend

toward significance, (Z= -1.260; p = .20). This is in contrast to verb-based treatments in

which acceptable responses increased in only 3 of 9 participants, (Z = 1.007; p >.20). See

Figure 2.


Nouns Produced Modifiers Produced Acceptable Maze per utterance Type Token Ratic
Responses
Word Structure
Figure 2. Word-Type Trained: A comparison of the percentage of participants with
increased production of measures with significant increases.
Information Level
Verb based treatment increased type-token ratio and percent maze words. Noun

based treatment also increased percent maze words but not type-token ratio, as shown in

Figure 2. Participants receiving verb-based treatment showed a significant increase in

type-token ratio affecting 8 of 9 participants, (Z = -2.084; p < .04). In contrast, only 3 of

8 participants of the noun-based treatments increased in type-token ratio, (Z < 1).


* Noun
* Verb


I I


o









Percent of maze words increased for 6 of 9 participants of the noun based treatment

with marginal significance, (Z= -1.859; p = .063). Verb based treatment increased

percent of maze words for 7 of 9 participants, but it was not significant, (Z= -1.011; p >

.20). See Figure 2.

There were no other significant changes attributable to differences in treated word

type in information measures: mean length of utterance in words or proportion of

utterances with UNIs.

Gestural-Verbal Treatment: N versus V

Word Level

The GV treatment, when treating nouns or verbs, had positive effects on the

number of verbs and modifiers produced in discourse, as shown in Figure 3. Verb

production increased for 6 of 7 participants of verb GV treatment, with a trend toward

significance, (Z = -1.352; p = .176). Noun based GV treatment increased production of

verbs for 3 of 5 participants, but this was not significant, (Z < 1). See Figure 3.

Treating verbs increased production of modifiers for 6 of 7 participants with a trend

toward significance, (Z= -1.352; p = .176). Treating nouns increased production of

modifiers for 4 of 5 participants with a trend toward significance, (Z= -1.483;p = .138).

There were no other significant changes due to treated word type in the GV

treatment for other word level measures: pronouns per utterance, nouns per utterance, or

automatic statements per utterance.











90

80

S70-

S60-

2 50

Produced Responses Utterance inVerb
40o

30

20

10


Verb Produced Modifiers Good Sentences Acceptable Mean Length of Type Token Ratio
Produced Responses Utterance in
Words
Word Structure

Figure 3. Gestural-Verbal Treatment: A comparison of the percentage of participants with
increased production of measures with significant increases.

Sentence Level

There were no significant changes due to treated word type in the GV treatment for


sentence level measures: number of one-word responses per utterance, number of


elliptical sentences per utterance, number of good sentences per utterance, number of


questions per utterance, or number of irrelevant responses per utterance.


Verb based GV treatment increased production of good sentences for 5 of 7


participants but it was not significant, (Z= -1.183; p = .237). Noun based GV treatment


also increased production of good sentences for 3 of 5 participants, this also was not


significant, (Z < 1). See Figure 3.


Acceptable responses per utterance (good sentences, one-word responses, elliptical


sentences), increased with noun-based GV treatment in 4 of 5 participants, but it was not









significant, (Z= -1.214; p = .225). Verb-based GV treatment increased acceptable

responses in only 3 of 7 participants, (Z< 1).

Information Level

Treating verbs in the GV treatment had positive effects on mean length of utterance

in words and type-token ratio, as shown in Figure 3. Mean length of utterance in words

increased for 5 of 7 participants of the GV verb treatment with a trend toward

significance, (Z = -1.352; p = .176). In contrast, noun based GV treatment increase mean

length of utterance in words for only 2 of 5 participants, (Z < 1).

Type-token ratio increased for 6 of 7 participants of the verb based GV treatment

with marginal significance, (Z= -1.614; p = .10). In contrast, noun based GV treatment

increased type token ratio for only 2 of 5 participants, (Z < 1).

There were no other significant changes attributable to differences in treated word

type in the GV treatment for information measures: percent maze words or proportion of

utterances with UNIs.

SP N versus V

Word Level

There were no significant changes for either of the two participants of the verb

based SP treatment or for any of the three participants of the nouns based SP treatment,

attributable to treated word type in the SP treatment for word level measures: nouns per

utterance, pronouns per utterance, modifiers per utterance, verbs per utterance, or

automatic statements per utterance.

Noun based SP treatment increased noun production for 2 of 3 participants,

however; it was not significant, (Z= -1.069; p = .285).









Sentence Level

Verb-based SP treatment did not increase production of good sentences or

acceptable responses for either of the two participants. Noun-based SP treatment

increased production of good sentences in 1 of 3 participants,,(Z <1) and increased

acceptable responses in 2 of 3 participants,(Z= -1.069; p = .285) however, these were not

significant.

There were no other significant changes due to treated word type in the SP

treatment for sentence level measures: number of one-word responses per utterance,

number of elliptical sentences per utterance, number of questions per utterance, or

number of irrelevant responses per utterance.

Information Level

Verb based SP treatment had positive effects on type token ratio and percent maze

words. Noun based SP treatment was associated with the percent of maze words

produced. Noun based SP treatment increased percent maze words for 3 of 3 participants

with a trend toward significance, (Z = -1.604; p = .109). Verb based SP treatment also

increased percent maze words for 2 of 2 participants, (Z = -1.342; p < .20).

Verb based SP treatment increased type token ratio for 2 of 2 participants with a

trend toward significance, (Z= -1.342; p < 2.0). In contrast, noun based SP treatment

increased type token ratio for only 1 of 3 participants, (Z < 1).

There were no other significant changes attributable to differences in treatment type

in information measures: mean length of utterance in words or proportion of utterances

with UNIs.

Noun based SP treatment increased proportion of utterances with UNIs for 2 of 3

participants, however; it was not significant, (Z >1).









Correlations

Correlations were calculated for changes scores in the number of nouns produced,

number of verbs produced, number of modifiers produced, UNIs produced, and number

of acceptable responses produced.

SP Treatment

Change score of the number of nouns produced was significantly correlated with

change scores of UNIs (p = .01), verbs (p = .06), modifiers (p = .10), and acceptable

responses (p = .15). As the number of nouns per utterance increased so did the number of

UNIs, modifiers, and acceptable responses. However, the number of verbs per utterance

decreased as the number of nouns increased.

Change score of the number of verbs produced was significantly correlated with

change score of nouns and UNIs (p = .20). As the number of nouns and UNIs increased

the number of verbs decreased. Change score of the number of verbs produced was not

significantly correlated with other change scores: modifiers and acceptable responses.

Change score of the number of modifiers produced was significantly correlated

with change scores of nouns, acceptable responses (p = .03), and UNIs (p = .03). As the

number of modifiers per utterance increased so did the number of nouns, acceptable

responses, and UNIs. Change score of the number of modifiers produced was not

significantly correlated with other change scores: verbs.

Change score of the number of acceptable responses produced was significantly

correlated with change scores of nouns, modifiers, and UNIs (p = .096). As the number

of acceptable responses increased, so did the number of nouns, modifiers, and UNIs.

Change score of the number of acceptable responses was not significantly correlated with

other change scores: verbs.









GV Treatment

Change score of the number of nouns produced was not significantly correlated

with other change scores: verbs, modifiers, UNIs, acceptable responses.

Change scores of the number of verbs produced was significantly correlated with

change score of acceptable responses (p = .02). As the number of verbs per utterance

increased, so did the number of acceptable responses. There were no other significant

correlations between the change score of number of verbs produced and other change

scores: nouns, modifiers, and UNIs.

Change scores of the number of modifiers produced was significantly correlated

with change scores of verbs (p =.04) and the correlation with acceptable responses

approached significance (p = .075). As the number of modifiers per utterance increased,

so did the number of verbs and acceptable responses. There were no other significant

correlations between the change score of number of modifiers produced and other change

scores: nouns and UNIs.














CHAPTER 4
DISCUSSION

This study tested the hypothesis that naming treatments for aphasia would lead to

quantifiable changes in discourse measures. More specifically, the purpose of this study

was to compare the effects of two aphasia treatments conducted by Raymer, Ciampitti, et

al.(in press) and Raymer, Singletary, et al. ( in press), on production of grammatical and

lexical aspects of discourse. There were four research predictions. First, it was predicted

that changes in grammatical units and forms would be higher in the compensatory

gestural-verbal treatment than the semantic-phonologic treatment; results of the treatment

type analysis support this hypothesis. Second, it was predicted that there would be

increases in the use of the particular word type that was trained. This effect was found in

the GV verb treatment with a trend toward significance and in the SP noun treatment

which was not significant. The third prediction was that increased noun and verb

production would correlate with increases in UNIs. This correlation was significant in the

SP treatment condition and but not in the GV treatment. The fourth and final prediction

was that increased production of 'good sentences' would correlate with increases in

UNIs. This finding was present in both GV treatments, although neither effect was

significant.

Treatment Type

Comparison of GV and SP treatment revealed more significant changes in the

discourse of the participants of the GV treatment. Participants receiving the GV

treatment increased production of both nouns and verbs; whereas, those receiving the SP









treatment only increased production of nouns. This is consistent with previous findings

from gestural therapies and in the literature. Hadar, Wenkert-Olenik, Krauss, and

Soroker, (1998) found that speakers with aphasia produced a higher number of gestures

when word finding decreased. They concluded that gestures increased lexical retrieval.

Rauscher, Krauss, and Chen (1996) found that when speakers without neurological

impairment were not allowed to use gestures, word retrieval decreased. Thus, both of

these studies provided evidence for gestures facilitating word retrieval. Treatment studies

have also found this relationship between gestures and word retrieval. Pashek (1997)

found that verbal plus gestural treatment significantly increased naming compared to the

effects of verbal-only treatment. Miranda Rose and Jacinta Douglas (2001) found iconic

gestures significantly improved object naming in participants with phonologic

impairments, but not in participants with semantic impairments. They concluded that

iconic gestures prime impaired phonological access, storage, and encoding processes, but

not impaired semantic storage. Rose and Douglas state that this finding is in support of

Krauss and Hadar's (as cited in Rose & Douglas, 2001) theory of lexical gesture and

speech production. In this theory, gestures facilitate lexical access by activating

gesturally represented features of the message. Krauss and Hadar state that this

activation occurs before articulation of the word, in Levelt's (as cited in Rose & Douglas,

2001) formulation stage. Krauss and Hadar (as cited in Rose & Douglas, 2001) imply

that priming occurs from the kinesic monitor to the formulator level which contains the

grammatical encoder and phonological encoder (Rose, Douglas, & Matyas, 2002).

However, Rose and Douglas (2001) state that the precise level at which priming occurs is

not clear in Krauss and Hadar's model. Based their findings that gestures only









facilitated speakers with phonological impairments, Rose and Douglas (2001) concluded

that gesture-related information enters the speech production system at the phonological

level. However, they also point out that gestures may facilitate word retrieval at both the

phonological and lemma level. While it is more intuitive to suggest gestures facilitating

at the lemma level or even before the formulator level at the conceptualizer, facilitation at

the phonological level explains how the GV treatment increased nouns and verbs.

Applying Rose and Douglas's theory to the current findings may explain why the

GV treatment increased nouns and verbs and the SP treatment only increased nouns.

Following Rose and Douglas's conclusions, the GV treatment activated the phonological

processes in producing nouns and verbs and not the semantic processes of nouns and

verbs that might lead to differential activation of word classes as seen in the SP treatment.

Alternately, the GV treatment may have increased production of verbs in addition to

nouns simply because the speakers were 'acting out' the verb. Performing a physical

action associated with the word may have been enough to activate retrieval of the verb.

In addition to increasing nouns and verbs, the GV treatment increased modifiers. Similar

to the possible effect on verbs, performing gestures that describe the object or verb may

have activated associated modifiers by spreading activation. As modifiers provide more

information about the topic, increased production of modifiers may have contributed to

the increase in UNIs. However, the impact of modifiers on increased production of UNIs

may not be significant, because the SP treatment also increased UNIs but not modifiers.

The SP treatment had a higher percentage of participants increase in one-word

responses. It is arguable whether this is an improvement. While it is not a quantitative

improvement as 4 of 5 participants were nonfluent, there may be a qualitative









improvement. As discussed below, the SP noun condition did increase production of

UNIs. It may be that although the participants were using more one-word responses, the

response was correct and relevant and provided more information than before treatment.

Trained-Word Type

Comparison of noun-based to verb-based treatments showed that higher

percentages of participants in the verb-based treatments increased their production of

nouns, verbs, modifiers, TTR, mazes per utterance, and UNIs. However, of all of these

measures, only the increase in TTR was significantly higher post treatment. Training

verbs may have increased TTR more than training nouns because verbs have been found

to be more difficult for speakers with aphasia to retrieve (Berndt, Burton, Haendiges, &

Mitchum, 2002; Marshall, Pring, & Chiat, 1998). In addition, Thompson et al. (1997)

found that speakers with agrammatic aphasia have been found to produce verbs with

simple argument structure. Based on the results of their study, Thompson et al.

concluded that verb argument structure is important for verb retrieval. Thus, in the

current study increasing verbs provided the speakers with access to a class of words and

argument structures that was previously difficult to retrieve, increasing the lexical

diversity of their discourse.

In the verb condition, UNIs increased as did production of nouns, verbs, modifiers,

and TTR. Thus, participants were able to increase production of content words and

convey more information using a greater variety of words. Percentage of maze words

also increased in the verb condition. This may have been due to an increased availability

of words which, in turn, led to increased attempts at verbal responses.









Trained-Word Type within Treatment Type


SP Noun vs. Verb

Results of comparing trained-word type within the SP treatment are based on three

participants in the noun condition and two participants in the verb condition, making it

difficult to achieve statistical significance for any measure. Training nouns increased

nouns, acceptable responses, and UNIs. All of these participants were nonfluent; thus, as

they were able to produce more nouns they were able to produce different types of words,

leading to production of more acceptable responses with more UNIs, but not significantly

more. However, as they were able to access more nouns and types of words, they made

more mistakes resulting in an increased percentage of maze words. It should be

emphasized that acceptable responses include one-word and elliptical responses as well

as good sentences; thus, the finding that noun training increased acceptable responses is

not necessarily contradictory to the findings stated below, which demonstrate that verb

training increases sentence production (Berndt et al., 2002; Marshall et al., 1998).

Similar to the effect in the noun condition, participants in this condition also

produced an increased percentage of maze words, presumably as a result of increases in

overall word availability, leading to more lexical intrusions. These were again

accompanied by increases in TTR, supporting the idea that there was an overall increase

in lexical availability. The increased lexical availability may have overloaded the

sentence production mechanism by providing more activated words to choose from and

organize into a sentence, thus resulting in more mistakes and mazes. Crockford (1991)

offers another explanation for finding increases in 'repair turns' accompanied by

increased functional communication in a patient with aphasia. That study revealed that

the patient's wife did not need to offer as much help during 'repair turns' because of the









patient's increased communication ability. This phenomenon was also observed in

several participants in the current study. Post-treatment transcripts contained fewer

utterances from the caregivers than pre-treatment transcripts.

Significant gains in SP noun treatment were offset by a very small N (i.e., three

participants), making any kind of statistical inference impossible. In addition, participants

of the SP noun treatment were the farthest post onset at 75, 93, and 120 months post,

(average 96 months) versus the rest of the participant population who were 5 to 62

months post onset (average 24.7 months). Therefore, the lack of effects may be the result

of the generalization effects of naming treatment to discourse being limited to a specific

time period after the onset of aphasia, rather than to inadequacy of the treatment.

In summary, noun based and verb based SP treatments increased percent maze

words in all participants. This finding may be attributable to increased attempts caused

by an increase in the availability of words. Crockford's (1991) explanation of decreased

help in 'repair turns' by the spouse can also apply in this condition. Although percent

maze words increased in all participants of SP treatment, it is difficult to identify the

cause of this effect, as only five people participated in this treatment.

GV Noun vs. Verb

Within the GV treatment, the verb condition resulted in the highest number of gains

as well as the most significant gains. The verb condition had a significantly greater effect

on production of verbs, MLU in words, and TTR. The verb condition also had higher

increases in modifiers and good sentences produced, though these were not significantly

higher than the analogous increases in the noun condition.

Increased production of good sentences in the verb condition is consistent with

findings from several studies linking verb retrieval and sentence production. Berndt,









Burton, Haendiges, and Mitchum (2002), found that speakers with greater impairment in

verb retrieval than noun retrieval also had more impaired sentence production. It follows

then that improving verb retrieval would improve sentence production (Berndt et al.,

2000; Berndt, et. al, 2002; Marshall et. al, 1998). However, there have been cases that do

not support a connection between verb retrieval and sentence production, as demonstrated

by the patient described in Berndt, Haendiges, and Wozniak (1997) who had severe

anomia characterized by significantly higher verb retrieval than noun retrieval but

impaired sentence comprehension and production. GV verb treatment also was

associated with increased MLU in words and TTR; in other words, the GV verb condition

increased the production of good sentences, the length of utterances in words, and the

lexical diversity of utterances However, this treatment did not increase maze production,

as did both variations of the SP treatment. These findings suggest that the participants

were able to retrieve the correct words in the correct order the first time and did not need

several attempts. The process of gesturing may have strengthened the neural connections

in the representations of verbs enough to decrease the need for multiple attempts to

produce the correct verb. This may have occurred because gestures do not result in the

same extent of spread of activation in the semantic system if gestures facilitate lexical

access at a post-semantic stage. This is an important observation as participants in this

treatment would then be more efficient speakers. Interestingly, although maze

production was not increased, UNIs did not increase in either GV treatment condition

either. Thus, the speakers used more types of words, had longer responses, with

increased modifiers, verbs, and good sentences, but they did not produce more utterances

with new information. The relationship between increases in the production of nouns and









verbs and increases in UNIs is difficult to clearly establish due to methodological issues.

UNIs were an utterance level code, with only one allowed per utterance; consequently, if

there were several pieces of new information in an utterance this would have been

missed. This presents the need for a more quantitative information measure combining

aspects of the UNI and CIU.

The only measure in which GV noun condition had greater effect was acceptable

responses. This coincides with the treated-word type results in which noun conditions led

to greater gains than verb conditions in acceptable responses. Acceptable responses

included one-word responses but not UNIs, as a result, this measure is not particularly

indicative of improved discourse. However, an increase in acceptable responses does

indicate the discourse is easier to understand for the communication partner even if the

amount of information conveyed has not increased.

Mazes

Production of mazes increased in noun and verb conditions in the SP and GV

treatments, but was only significant in SP noun and verb conditions. Increases of maze

production were associated with increases in TTR in GV and SP verb conditions

suggesting a link between increased lexical availability and maze production. All but one

SP participant was nonfluent; therefore, any increase in lexical access would provide the

speaker with more words to retrieve and organize into sentence form, possibly

overburdening the sentence production mechanism, leading to more mistakes or mazes.

Mazes are typically not included in analyses such as QAAP (Saffran, Berndt, &

Schwartz, 1989) and SALT (Miller & Chapman, 1991). Based on the findings of the

current study mazes should be considered in future analyses to investigate lexical access

and sentence formulation.









Limitations

In general, there were few significant gains (p < .05) due to the small N in each

treatment. When comparing treatment type, statistics were run on 12 participants in the

GV treatment and 5 participants in the SP treatment. All participants who completed the

GV treatment and all but one of the participants who completed the SP treatment were

analyzed in this study. When comparing trained-word type within each treatment,

statistical analysis was further compromised by even smaller groups: SP verb treatment

had two participants and SP noun treatment had three participants, while GV verb

treatment had seven participants and GV noun treatment five participants. Due to the

variability within and between the discourse productions of speakers with aphasia, it is

difficult to make generalizations from a small sample of speakers (Prins & Bastiaanse,

2004).

A second limitation of the treatment design was the mixture of aphasia types

included. In each treatment and condition there were participants with fluent and

nonfluent aphasia. Because nonfluent and fluent aphasias arise from different lesion sites

and result in different type and level of impairment, this may have contributed to the lack

of significant findings in the study. Specifically, analysis was completed on groups and

not individuals; thus, the mixture of aphasia types may have affected the results. The

discourse production of a speaker with nonfluent aphasia is typically characterized by

one-word responses or short phrases. The discourse of a speaker with fluent aphasia is

characterized by longer utterances with minimal information. Thus, combined analysis

of grammatical components of the discourse of nonfluent and fluent speakers may have

obscured the true treatment effects present.









A third limitation of the treatment design was the mixed genres of discourse used

as stimuli. Participants discussed favorite foods and hobbies, people and events in family

pictures, and pictures of famous people or events. Although it has not been discussed in

past studies or reviews, differences in discourse production may arise when using family

pictures and pictures of famous people not personally known to the speaker. As

Armstrong (2000) notes, some research suggests there are differences in the discourse

elicited by various stimuli such as the ones in the current study, picture description,

opened-ended questions, and discussing family members and memories. In addition,

pictures of people at an event in which the event is clear in the picture may elicit different

discourse than a picture of a single person or group of people not in an obvious setting or

event. Armstrong (2000) notes that single picture stimuli may not elicit a narrative with

orientation, precipitating action, and resolution, but elicit description of a situation. In

their categorization of discourse types, Prins and Bastiaanse (2004) separate discourse

from situational pictures and discourse elicited through interview with open-ended

questions. Differences previously found between discourse with and without picture

stimuli include: less verbal complexity with pictures (Glosser, Weiner, & Kaplan, 1998),

higher efficiency scores without pictures (Doyle et al. 1995), and higher cohesive

harmony without pictures (Armstrong, 1988). Thus, our analysis might have benefited

from being limited to one of the three discourse types used.

Summary

Despite the limitations and few significant findings of the current study, it offers

several important contributions to aphasia discourse research. The relationships among

word classes, sentence structure, and units of information found provide a strong

argument for grammatical analysis as a viable method of measuring changes in discourse.









Measuring changes in information is particularly important whether it be through use of

the UNI presented in this study or other similar measures. A comparison of the UNI and

CIU should be conducted in the future to clearly identify differences and ideal uses for

each. Mazes should be considered as part of future discourse analyses. The current study

found increases in mazes when word retrieval increased. This is important theoretically

in regards to theories of lexical access and activation. Future research should continue to

investigate the various components necessary for conveying information such as topic

relevancy and coherence. Findings of the current study also point to a need for further

research elucidating the differential effects of training nouns and verbs and of the

methods of training. As seen here, adding a component used by many speakers in daily

conversation, gestures, increased verbal output and improved communication. Although

the changes in discourse found in the current study were not, for the most part,

statistically significant, this study does provide evidence that naming treatments can lead

to changes in discourse. This is highly important as improved discourse or

communication should always be the ultimate goal of any aphasia treatment.















LIST OF REFERENCES


Arbuckle, T.Y., Gold, D., Frank, I., & Motard, D. (1989). Speech of verbose older
adults: How is it different? Paper presented at the annual meeting of the
Gerontological Society of America, Minneapolis, MN.

Armstrong, E. (2000). Review: Aphasic discourse analysis: The story so far.
Aphasiology, 14(9), 875-892.

Armstrong, E.M. (1988). A cohesion analysis ofaphasic discourse. Unpublished MA
thesis, Macquarie University, Sydney.

Bastiaanse, R., Edwards, S., & Kiss, K. (1996). Fluent aphasia in three languages:
Aspects of spontaneous speech. Aphasiology, 10, 561-575.

Berko-Gleason, J., Goodglass, H., Obler, L., Green, E., Hyde, M., & Weintraub, S.
(1980). Narrative strategies of aphasics and normal-speaking subjects. Journal of
Speech and Hearing Research, 23, 370-382.

Berndt, R.S., Burton, M.W., Haendiges, A.N., & Mitchum, C.C. (2002). Production of
nouns and verbs in aphasia: Effects of elicitation context. Aphasiology, 16(1/2),
83-106.

Berndt, R.S., Haendiges, A.N., & Wozniak, M.A. (1997). Verb retrieval and sentence
processing: Dissociation of an established association. Cortex, 33(1), 99-114.

Bird, H., & Franklin, S. (1996). Cinderella revisited: A comparison of fluent and non-
fluent aphasic speech. Journal ofNeurolinguistics, 9, 187-206.

Boles, L. (1998). Conversational discourse analysis as a method for evaluating progress
in aphasia: A case report. Journal of Communication Disorders, 31, 261-274.

Boyle, M. (2004). Semantic feature analysis treatment for anomia in two fluent aphasia
syndromes. American Journal ofSpeech-Language Pathology, 13(3), 236-249.

Brenneise-Sarshad, R., Nicholas, L.E., & Brookshire, R.H. (1991). Effects of apparent
listener knowledge and picture stimuli on aphasic and non-brain-damaged
speakers' narrative discourse. Journal of Speech and Hearing Research, 34, 168-
176.









Cherney, L.R., & Canter, G.J. (1993). Informational content in the discourse of patients
with probable Alzheimer's disease and patients with right brain damage. Clinical
Aphasiology, 21, 123-134.

Crockford, C. (1991). Assessing functional communication in aphasic adults: A
comparison of three methods. Unpublished undergraduate dissertation. Department
of Speech, University of Newcastle upon Tyke, UK.

Doyle, P.J., Goda, A.J., & Spence, K.A. (1995). The communicative informativeness and
efficiency of connected discourse by adults with aphasia under structured and
conversational sampling conditions. American Journal of Speech-Language
Pathology, 4, 130-134.

Doyle, P.J., McNeil, M.R., Spence, K.A., Goda, A.J., Cottrell, K., & Lustig, P.P. (1998).
The effects of concurrent picture presentations on retelling of orally presented
stories by adults with aphasia. Aphasiology, 12, 561-574.

Duffy, J.R. (1995). Motor speech disorders. St. Louis, MO: Mosby.

Edwards, S. (1995). Profiling fluent aphasic spontaneous speech: A comparison of two
methodologies. European Journal of Disorders of Communication, 30, 333-345.

Edwards, S., & Bastiaanse, R. (1998). Diversity in the lexical and syntactic abilities of
fluent aphasic speakers. Aphasiology, 12, 99-117.

Gleason, J.B., Goodglass, H., Obler, L., Green, E., Hyde, M.R., & Weintraub, S. (1980).
Narrative strategies of aphasic and normal-speaking subjects. Journal of Speech
and Hearing Research, 23, 370-382.

Glosser, G., & Deser, T. (1990). Patterns of discourse production among neurological
patients with fluent language disorders. Brain and Language, 40, 67-88.

Glosser, G., Weiner, M., & Kaplan, E. (1998). Variations in aphasic language behaviors.
Journal of Speech and Hearing Disorders, 53, 115-124.

Goffman, E. (1981). Forms of talk. Oxford: Basil Blackwell.

Goodglass, H., Christiansen, J.A., & Gallagher, R. (1993). Comparison of morphology
and syntax in free narrative and structured tests: Fluent vs nonfluent aphasics.
Cortex, 29, 377-407.

Goodglass, H., Kaplan, E., & Barresi, B. (2000). Boston Diagnostic Aphasia
Examination (3rd ed.). Philadelphia: Lippincott, Williams, & Wilkins.


Grimes, J.E. (1975). The thread of discourse. Paris: Mouton.









Hader, U., Wenkert-Olenik, D., Krauss, R., & Soroker, N. (1998). Gesture and the
processing of speech: Neuropsychological evidence. Brain and Language, 62, 107-
126.

Halliday, M.A.K. (1985a). An introduction to functional grammar. London: Edward
Arnold.

Halliday, M.A.K. (1985b). Context of situation. In M.A.K. Halliday, & R. Hasan (Eds.)
Language, context, and text: Aspects of language in a social-semiotic perspective
(1-14). Geelong, Victoria: Deakin University Press.

Harris, Z. (1963). Structural linguistics. Chicago: University of Chicago Press

Harris, Z. (1988). Language and information. New York: Columbia University Press.

Hasan, R. (1985). The texture of a text. In M.A.K. Halliday, & R. Hasan (Eds.)
Language, context, and text: Aspects of language in a social-semiotic perspective
(70-96). Geelong, Victoria: Deakin University Press.

Hier, D.B., Hagenlocker, K., & Shindler, A.G. (1985). Language disintegration in
dementia. Brain andLanguage, 25, 117-133.

Kaplan, E., Goodglass, H., & Weintraub, S. (2001). Boston naming test. Philadelphia,
PA: Lea and Febiger.

Kertsz, A. (1982). Western aphasia battery. New York: Grune & Stratton.

Marshall, J., Pring, T., & Chiat, S. (1998). Verb retrieval and sentence production in
aphasia. Brain and Language, 63, 159-183.

Meyers, P.S. (1979). Profiles of communication deficits in patients with right cerebral
hemisphere damage. In R.H. Brookshire (Ed.) Clinical aphasiology conference
proceedings (38-46). Minneapolis: BRK

Miceli, G., Silveri, M.C., Romani, C., & Caramazza, A. (1989). Variation in the patterns
of omissions and substitutions of grammatical morphemes in the spontaneous
speech of so-called agrammatic patients. Brain and Language, 36, 447-492.

Miller, J. F., & Chapman, R. S. (1991). SALT: A computer program for the systematic
analysis of language transcripts. Madison, WI: University of Wisconsin.

Nicholas, L.E., & Brookshire, R.H. (1993a). A system for quantifying the
informativeness and efficiency of the connected speech of adults with aphasia.
Journal of Speech and Hearing Research, 36, 338-350.

Nicholas, L.E., & Brookshire, R.H. (1993b). A system for scoring main concepts in the
discourse of non-brain-damaged and aphasic speakers. Clinical Aphasiology, 21,
87-99.









Nicholas, L.E., & Brookshire, R.H. (1995). The presence, completeness, and accuracy of
main concepts in the connected speech of non-brain-damaged and aphasic adults.
Journal of Speech and Hearing Research, 38, 145-156.

Nicholas, M., Obler L.K., Albert, M.L., & Helm-Estabrooks, N. (1985). Empty speech in
Alzheimer's disease and fluent aphasia. Journal of Speech and Hearing Research,
28, 405-410.

Obler, L.K., & Albert, M. (1986). Action naming test. Unpublished test.

Olness, G.S. (2005). Discourse genre and verb performance in picture-elicited discourse
of adults ii ilh aphasia. Presented at the Clinical Aphasiology Conference, Sanibel
Island, FL.

Pashek, G.V. (1997). A case study of gesturally cued naming in aphasia: Dominant
versus nondominant hand training. Journal of Communication Disorders, 30, 349-
366.

Prins, R., & Bastiaanse, R. (2004). Review: Analysing the spontaneous speech of
aphasic speakers. Aphasiology, 18(12), 1075-1091.

Rauscher, F.H., Krauss, R.M., & Chen, Y. (1996). Gesture, speech, and lexical access:
The role of lexical movements in speech production. Psychological Science, 7(4),
226-231.

Raymer, A.M., Ciampitti, M., Holliway, B., Singletary, F., Blonder, L.X., Ketterson, T.,
Heilman, K.M., & Gonzalez-Rothi, L.J. Semantic-phonologic treatment for noun
and verb retrieval impairments in aphasia. Manuscript submitted for publication.

Raymer, A.M., Singletary, F., Rodriguez, A., Ciampitti, M., Heilman, K.M., & Gonzalez-
Rothi, L.J. Gestural+verbal training effects for noun and verb retrieval in aphasia.
Manuscript submitted for publication.

Roberts, J, & Wertz, R.T. (1989). Comparison of spontaneous and elicited oral-
expressive language in aphasia. ClinicalAphasiology, 18, 479-488.

Rochon, E.R., Saffran, E.M., Sloan-Berndt, R.S., & Schwartz, M. (2000). Quantitative
analysis of aphasic sentence production: Further development and new data. Brain
and Language, 72, 193-218.

Rose, M., & Douglas, J. (2001). The differential facilitatory effects of gesture and
visualization processes on object naming in aphasia. Aphasiology, 15 (10/11), 977-
990.

Rose, M., Douglas, J., & Matyas, T. (2002). The comparative effectiveness of gesture and
verbal treatments for a specific phonologic naming impairment. Aphasiology,
16(10/11), 1001-1030.









Rothi, L.J.G., Raymer, A.M., & Heilman, K.M. (1997). Limb praxis assessment. In
L.J.G. Rothi & K.M. Heilman (Eds.), Apraxia: The neuropsychology of action (61-
73). Hove, East Sussex, UK: Psychology Press.

Saffran, E.M., Sloan-Berndt, R., & Schwartz, M. (1989). The quantitative analysis of
agrammatic production: Procedure and data. Brain and Language, 37, 440-479

Schwartz, M.F., Saffran, E.M., Bloch, D.D., & Dell, G.S. (1994). Disordered speech
production in aphasic and normal speakers. Brain and Language, 47, 52-88.

Shadden, B. (1998). Information analyses. In L.R. Cherney, B.B. Shadden, & C.A.
Coelho (Eds.) Analyzing discourse in communicatively impaired adults (85-114).
Gaithersburg, MD: Aspen.

Terrell, B.Y., & Ripich, D.N. (1989). Discourse competence as a variable in
intervention. Seminars in Speech and Language, 10, 282-297.

Thompson, C.K., Lange, K.L., Schneider, S.L., & Shapiro, P. (1997). Agrammatic and
non-brain damaged subjects' verb and verb argument structure production.
Aphasiology, 11, 473-490.

Tompkins, C.A., Boada, R., McGarry, K., Jones, J., Rahn, A.E., & Ranier, S. (1993).
Connected speech characteristics of right-hemisphere-damaged adults: A re-
examination. Clinical Aphasiology, 21, 113-122.

Trupe, E.H., & Hillis, A. (1985). Paucity vs. verbosity: Another analysis of right
hemisphere communication deficits. In R.H. Brookshire (Ed.), Clinical aphasiology
conference proceedings (83-96). Minneapolis: BRK.

Ulatowska, H.K., Allard, L., & Bond Chapman, S. (1990). Narrative and procedural
discourse in aphasia. In Y. Joanette, & H.H. Brownell (Eds.) Discourse ability and
brain damage: Theoretical and empiricalperspectives (180-198). New York:
Springer-Verlag.

Ulatowska, H.K., Freedman-Stern, R., Doyel, A.W., & Macaluso-Haynes, S. (1983).
Production of narrative discourse in aphasia. Brain and Language, 19, 317-334.

Ulatowska, H.K., North, A.J., & Macaluso-Haynes, S. (1981). Production of narrative
and procedural discourse in aphasia. Brain and Language, 13, 345-371.

Williams, S.E., Li, E.C., Volpe, A.D., & Ritterman, S.I. (1994). The influence of topic
and listener familiarity on aphasic discourse. Journal of Communication
Disorders, 27, 207-222.

Yorkston, K.M., & Beukelman, D.R. (1980). An analysis of connected speech Samples of
aphasic and normal speakers. Journal of Speech and Hearing Disorders, 45, 27-
36.















BIOGRAPHICAL SKETCH

Christina del Toro is a graduating master's student in the University of Florida

department of Communication Sciences and Disorders. During her master's program she

completed a master's thesis on aphasia under the mentorship ofLori Altmann, Ph.D. Ms.

del Toro received her B.A. in Communication Sciences and Disorders from the

University of Florida in May 2004. In her senior year she completed a senior honors

thesis on aphasia with Diane Kendall, Ph.D., which was accepted as a poster presentation

at the 14th NIDCD-sponsored Research Symposium. Over her four years of college she

was honored with membership into Phi Eta Sigma honor society, Golden Key honor

society, Phi Sigma Theta honor society, and Tau Sigma transfer student honor society.

She has also been on the Dean's and President's List for her GPA. While attending UF

full-time, she has worked as research assistant at the VA Brain Rehabilitation Research

Center in Gainesville, Florida. Her duties have included collecting reliability data,

developing screening forms using the Autodata software program, and most recently

study coordinator for a mild aphasia assessment protocol developed by Anna Moore,

Ph.D. In August 2006, Ms. del Toro will begin her doctoral degree in Communication

Sciences and Disorders at the University of Florida under the mentorship of Diane

Kendall, Ph.D.




Full Text

PAGE 1

CHANGES IN GRAMMATICAL ASPECT S OF APHASIC DISCOURSE AFTER CONTRASTING TREATMENTS By CHRISTINA M. DEL TORO A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Christina M. del Toro

PAGE 3

iii ACKNOWLEDGMENTS There are many people who have help ed and supported me throughout the completion of my degree and thesis. I firs t must thank my committee members who have been instrumental to my academic and professional growth. Dr. Lori Altmann has taught me so much about aphasia and lingui stics that I have gained a new perspective on language impairme nts and treatments. Her encouragement, trust, and support have been integral to my success and enthusiasm. Dr. Diane Kendall has offered her guidan ce and knowledge since my undergraduate studies when I began to develop my skills as a researcher with her help and support. She has always offered her time to provide oppor tunities for research and furthering my knowledge and interest in the field. Dr. Bonnie Johnson has been my professo r since undergraduate studies and taught my first class in language development. She has always encouraged critical thinking in the classroom and clinic. Her enthusiasm in the classroom and clinic has been inspirational. I could not have completed this study wit hout the help and sup port of the Discourse Group at the Language over the Lifespan Lab. Susan Leon, Lynn Dirk, and Elizabeth Mikell were involved from th e beginning of this project and shared in the learning experiences of discourse analysis. I would also like to thank Dr. Kenneth L ogan who has always been available to answer my questions about graduate school a nd the thesis. I tha nk Dr. Anna Moore for

PAGE 4

iv guiding my growth as a researcher and allowing me to take time from her project to focus on my thesis. I thank my fellow students and friends whose support and encouragement I greatly appreciate. I could not have enjoyed graduate school and writing a thesis without the wonderful friendship of Michelle Troche who inspires me w ith her hard work and great achievements. Finally, I thank my family whose love alwa ys inspires me and reminds me of what is important in life. I thank my sister and roommate Jennif er del Toro for her love and daily support that kept me going. I have enjoyed my time in undergraduate a nd graduate studies at the University of Florida and look forward to staying for my doctoral studies to further my academic and professional skills.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iii LIST OF TABLES............................................................................................................vii LIST OF FIGURES.........................................................................................................viii ABSTRACT....................................................................................................................... ix CHAPTER 1 INTRODUCTION........................................................................................................1 Aphasia and Discourse.................................................................................................1 Discourse Analysis................................................................................................2 Theoretical framework...................................................................................2 Information analyses......................................................................................4 Elicitation methods.........................................................................................7 Methodologies................................................................................................9 Limitations and deficienci es in current research..........................................10 Purpose................................................................................................................11 Research question 1......................................................................................12 Research question 2......................................................................................12 Research question 3......................................................................................12 Research question 4......................................................................................12 2 METHODS.................................................................................................................13 Participants.................................................................................................................13 Procedures for Semantic-Phonologic Treatme nt and Gestural + Verbal Treatment..14 Grammatical Analysis................................................................................................15 Sample.................................................................................................................15 Scoring.................................................................................................................15 Reliability............................................................................................................18 Statistical Analysis......................................................................................................18 3 RESULTS...................................................................................................................19 Treatment Type: SP versus GV..................................................................................19

PAGE 6

vi Word Level..........................................................................................................19 Sentence Level.....................................................................................................20 Information Level................................................................................................21 Word Trained: Noun or Verb.....................................................................................21 Word Level..........................................................................................................21 Sentence Level.....................................................................................................21 Information Level................................................................................................22 Gestural-Verbal Treatment: N versus V....................................................................23 Word Level..........................................................................................................23 Sentence Level.....................................................................................................24 Information Level................................................................................................25 SP N versus V.............................................................................................................25 Word Level..........................................................................................................25 Sentence Level.....................................................................................................26 Information Level................................................................................................26 Correlations.................................................................................................................27 SP Treatment.......................................................................................................27 GV Treatment......................................................................................................28 4 DISCUSSION.............................................................................................................29 Treatment Type...........................................................................................................29 Trained-Word Type....................................................................................................32 Trained-Word Type within Treatment Type..............................................................33 SP Noun vs. Verb................................................................................................33 GV Noun vs. Verb...............................................................................................34 Mazes.......................................................................................................................... 36 Limitations..................................................................................................................37 Summary.....................................................................................................................38 LIST OF REFERENCES...................................................................................................40 BIOGRAPHICAL SKETCH.............................................................................................45

PAGE 7

vii LIST OF TABLES Table page 1. Participant Demographics..............................................................................................14 2. Word Classes and Sentence Type Codes.......................................................................16

PAGE 8

viii LIST OF FIGURES Figure page 1. Treatment type: A comparison of the per centage of participants with increased production of measures with significant increases...................................................20 2. Word-type trained: A comparison of the pe rcentage of participants with increased production of measures with significant increases...................................................22 3. Gestural-verbal treatment: A comparison of the percentage of participants with increased production of measures with significant increases...................................24

PAGE 9

ix Abstract of Thesis Presented to the Graduate School of th e University of Florida in Partial Fulfillment of the Requirement s for the Degree of Master of Arts CHANGES IN GRAMMATICAL ASPECT S OF APHASIC DISCOURSE AFTER CONTRASTING TREATMENTS By Christina M. del Toro May 2006 Chair: Lori Altmann Major Department: Communica tion Sciences and Disorders The aim of the current study is to compar e the effects of two treatments on the production of discourse of speakers with a phasia using grammatical analysis that quantifies changes in the produc tion of various grammatical units and forms before and after treatment. Discourse analysis is an important method for studying the language system because it provides a view of the la nguage system in its natural setting of conversation. There are seve ral reasons for using discourse analysis as an outcome measure of aphasia treatments. Most important ly is that impairment of discourse is the most notable deficit and most troubling for the patient with aphasia. In addition, the goal of any aphasia treatment shoul d be improved discourse produc tion; thus, the effect of treatment on discourse is key to establishing successful treatments. Analyses of the current study included word classes, sentence types, and information units. Analyses are conducted to compare the two treatment type s, the two conditions, and each condition within each treatment type. Discourse data are taken from preand posttreatment measures of two naming treatments. Participan ts were assigned to either the semantic-

PAGE 10

x phonologic treatment or the compensatory gestur al-verbal treatment for aphasia; within each treatment, participants were trained on noun access or verb access. Results indicate greater increases in production of various wo rd classes, sentence types, and information units in the participants of the gestural + verbal treatment with verb training. A new measure of information is introduced and compared to current information measures used in discourse analysis. The significance of m azes is discussed in reference to a possible link to lexical access. Implications for futu re discourse analysis are discussed.

PAGE 11

1 CHAPTER 1 INTRODUCTION Aphasia and Discourse Aphasia is a language disorder resulting fr om damage to the brain significantly affecting all levels of language production: form, content, an d use. Deficits are observed through errors of word retrie val, phonological processing, grammar, and syntax (Duffy, 1995). In the clinical and research setting, it has been very common to use standardized tests to measure changes after treatment. Ho wever, the tasks involved in these measures, repeating sentences, naming pictures, and fo llowing verbal direc tions, are not tasks performed in everyday conversations (Boles 1998). In Elizabeth ArmstrongÂ’s (2000) review of aphasic discourse analysis, she points out that discourse analysis is an important method for studying the language sy stem because it provides a view of the language system in its natural setting of conve rsation compared to language tasks such as naming. There are several reasons for using discour se analysis as an outcome measure. Most importantly is that impairment of disc ourse is the most notable deficit and most troubling for the patient. Studying the effects of aphasia on discourse is also fundamental to classifying the type of aphasia and developing a treatment plan, as improved conversational skills should be the goal of any treatment. Furthermore, discourse analysis is important from a theoretical view because all linguistic levels of language interact in discourse provid ing an opportunity for devel oping and testing models of

PAGE 12

2 normal language production, and patterns of impa ired discourse can support or disprove current models (Armstrong, 2000; Prins & Bastiaanse, 2004). As discourse analysis has come to the fo refront of aphasia research, many different theoretical views and methods of analysis have been developed. Through the years limitations and challenges in th e use of discourse have been discussed, such as clearly defining discourse, identifying the components of discourse to measure, and developing elicitation techniques that yiel d relatively natural language sa mples (Armstrong, 2000). Discourse Analysis Elizabeth Armstrong (2000) identifies the ke y questions researchers aim to answer through discourse analysis. These questions concern identification of the following: the kinds of meanings conveyed by speakers w ith aphasia, the lexical and grammatical resources used to convey meaning and identify ing when meanings are no longer clear due to lexical and grammatical deficits in aphasia. Ar mstrong (2000) identifies two theoretical frameworks for defining and anal yzing discourse to answer these questions: structuralist-oriented and functionalist-oriented. Theoretical framework The functionalist-oriented perspective, defines discourse as language in use (Goffman, 1981; Halliday, 1985a, 1985b). This framework is focused on the meaning of discourse within its context. Functionalist an alysis is concerned with the macrostructure of discourse such as topic maintenance, tu rn-taking, appropriateness to the situation or topic, and the ability to or ganize and convey meaning. Several researchers have found that speakers with mild to moderate aphasia ar e able to employ the stru ctural principles of discourse used by normal speakers such as se tting, complicating ac tion, and resolution in a narrative, and obligatory elements of pro cedural discourse, although optional elements

PAGE 13

3 of procedural discourse are most ofte n omitted (Glosser & Desser 1990; Ulatowska, Freedman-Stern, Doyle, Macaluso-Haynes, & North, 1983; Ulatowska, North, & Macaluso-Haynes, 1981). As the structuralist framework will be employed for the current study, the functionalist framework will not be discussed in further detail here. The structuralist-oriented framework defi nes discourse as a co mponent of language above the sentence (Grimes, 1975; Harris, 1963 1988). In this framework, discourse is analyzed through its structural and lexical components, sentences, phrases, and words, commonly called microstructure. Lexical components have b een studied from a semantic perspective, measuring occurrence of para phasias and non-specific lexical items, and from a grammatical perspective, measuring types of word classes produced (Armstrong, 2000). Syntactic analysis has focused on gra mmatical complexity of sentences, syntactic errors, and clause argument structure (Bir d and Franklin, 1996; Brenneise-Sarshad, Nicholas, & Brookshire, 1991; Goodglass, Chri stiansen, & Gallagher, 1993; Schwartz, Saffran, Bloch, & Dell, 1994; Miceli, Silver i, Romani, & Caramazza, 1989; Roberts & Wertz 1989; Saffran, Sloan-Ber ndt, & Schwartz 1989;). Word class analysis focused on the structur al difficulty in discourse of speakers with agrammatic aphasia has found production of more nouns than verbs, and fewer closed class words and pronouns when compared to normal speakers. Syntactic analyses of discourse of speakers with agrammatic a phasia have found production of verbs with the simplest argument structur e and omissions of several st ructures including subject, main verb, or required function words and infl ections. Word class an alysis of discourse of speakers with fluent aphasia found pr oduction of more ve rbs than nouns (BerkoGleason et al., 1980). Syntactic analyses of the discourse from speakers with fluent

PAGE 14

4 aphasia have found a decrease in syntactic co mplexity and the frequency and variety of verbs (Bastiaanse, Edwards, & Kiss, 1996; Edwards, 1995; Edwards and Bastiaanse, 1998). The structuralist framework is adopted in the current analysis of word classes, syntax, and information units. Typical methods for eliciting discourse in the structuralist-oriented framework include single picture descrip tion, retelling stories in respon se to a series of pictures, retelling of previous accounts, retelli ng known fables, or monologues based on topics such as family, illness, or occupation. The current discourse samples were collected using several of these el icitation techniques. Information analyses An important aspect of discourse applicab le to both of the above frameworks is content and efficiency of language producti on (Armstrong, 2000). Measuring content or the amount of information conveyed by the speaker, has been argued as the best measurement for successful discourse produc tion (Shadden, 1998). Researchers have used several terms to refer to this unit of measurement, including ‘content units’ (Meyers, 1979; Yorkston & Beukelman, 1980); ‘themes’ (Berko-Gleason et al. 1980) ‘correct information units’ (Nicholas & Brookshi re, 1993a), ‘main concepts’ (Nicholas & Brookshire, 1993b, 1995), ‘essential informati on units’(Cherney & Canter, 1993; Hier, Hagenlocker, & Shindler, 1985; Nicholas, Obler, Albert, & Helm-Estabrooks, 1985), ‘propositions’ (Ulatowska, North et al., 1981) ‘essential and optiona l steps’ (Terrell & Ripich, 1989; Ulatowska, Freed man-Stern et al., 1983; Ulat owska, North et al., 1981;), ‘target lexemes and thematic units’ (G leason, Goodglass, Obler, Green, Hyde, & Weintraub, 1980), ‘unscorable or nonessentia l content’ (Tompkins, Boada, McGarry, Jones, Rahn, & Rainer, 1993; Trupe & Hillis, 1985) and ‘entire utterance’ (Arbuckle,

PAGE 15

5 Gold, Frank, & Motard, 1989). A lthough all these terms refer to measures of information they are not equivalent. In her chapter on information analys es, Shadden (1998) discusses the many differences among information measures. She be gins with the two approaches, ‘a priori’ measurements and ‘a posteriori’ measurements. ‘A priori’ analysis refers to methods that measure pre-determined esse ntial components of the disc ourse, as in Yorkston and Beukelman’s content units (1980), and Nichol as and colleagues’ e ssential information units (1985), both of which us ed descriptions of the C ookie Theft picture (Goodglass, Kaplan, & Barresi, 2000). An advantage to th e a priori method is that results can be compared to normative data; however, these no rms are only applicable to the particular picture used, and results cannot be generalized across pictures or across other elicitation tasks. ‘A posteriori’ analysis focuses on defining measures that can be applied across tasks and behaviors to develop computed meas ures with significance across these tasks. An example is Nicholas and Brookshire’s correct information unit (1993a; Shadden, 1998). Although the number of words and numbe r of CIUs can only be compared across identical tasks, words per minute, CIUs pe r minute, and percentage of CIUs can be compared across different tasks. Another di fference Shadden discusses is the idea of measuring only the amount of information or including quality of information (1998). Quality of information can be included in a priori measures as Myers (1979) did with Yorkston and Beukelman’s content units ( 1980) by dividing the c ontent units for the Cookie Theft picture into literal and interp retive. A third difference in information analyses is the inclusion of efficiency in the measure. This can be done by calculating information over time (e.g., CIU per minute, Boyle, 2004; Shadden, 1998;), number of

PAGE 16

6 language units (words or syllables) per info rmation unit, or using rating scales (e.g., Trupe, & Hillis, 1985) (Shadden, 1998). Past definitions of information measuremen t units have specified that information must be relevant to the topic, but have not re quired utterances to be coherent. Therefore, these measures only conveyed the amount of re levant words in the discourse and not the manner in which they were conveyed (Arm strong, 2000). For example, Nicholas and Brookshire (1993a) considered relevance when defining correct information units (CIU) but only in terms of the relevance of the individual words to the topic. This is problematic, as Hasan (1985) points out, because words can be relevant to the topic, but be incoherent in conveying information. The following utterances are given by Hasan (1985) to illustrate this limitation, “Girl bananas two spend shopkeeper/Apples own girls dollars grapes/Buy fifty sell cen ts shopkeepers/Girls fruit.” Although these utterances are relevant to the topic of shopping, they are not coherent and, therefore, are not useful for describing the difficulties speakers of aphasi a have in conveying information (Armstrong, 2000). CIUs are scored per word and not per utterance allowing each word conveying information to be measured including auxili ary verbs, determiners, conjunctions and other function words. However, due to diffe rences in utterance length or number of words produced in aphasia types, CIUs cannot be compared across aphasia types. These scores are inflated for fluent aphasia and depressed for non-fl uent aphasia, even if the amount of information is similar. In the current study the information measure will be referred to as ‘utterances with new inform ation’ or UNIs, a measure created by the Discourse Group at the University of Fl orida Language over the Lifespan Lab, and defined as a coherent utte rance providing information not previously given in the

PAGE 17

7 conversation. An utterance did not have to be a grammatical sentence or include more than one word to be counted as a UNI; th e only criteria was that it provided new information that was semantically coherent with the preceding context. The UNI is scored at the utterance level allowing for co mparison across aphasia type, unlike the CIU. However, the UNI may be limited in detecting small increases in information as it does not measure each word providing new information, and an utterance may include more than one piece of new information. Elicitation methods Researchers have elicited various types of discour se ranging from descriptive narrative, procedural discourse, or even role -play using a variety of methods: single picture description, picture seque nce description, retelling of a story read to the speaker, retelling of well-known storie s, recounting a memorable ex perience, or talking about personal events and family members (Armst rong, 2000). Procedural discourse requires given topics such as brushing teeth or going grocery shopping (Armstrong, 2000). Although all of the above methods result in conversation-like verb al production, unlike reading aloud, word or sentence repetition, or picture naming, they are not to be considered equal in the discourse they produ ce and in fact produce di fferent ‘genres’ of discourse (Armstrong, 2000; Ulatowska et. al., 1981, 1990; Williams et al., 1994). Within a genre, differences in definition have been found, such as in the narrative genre, the most commonly used in research (Armstr ong, 2000). Studies referring to their data as narratives have included participants disc ussing family members, single picture description, and retellings of personal even ts, fictional narrative, and known narratives such as ‘Cinderella’ (Glosser & Deser, 1990) These types of language samples differ

PAGE 18

8 with respect to their inclusion of the de fining elements of narrative: orientation, precipitating action and resolution (Armstrong, 2000). In their review of discourse research Prins and Bastiaanse (2004) make the following divisions in defining and eliciting discourse: (1) semi-spontaneous speech from situational pictures (e.g., Cookie Theft) or retelling a known stor y (e.g., Cinderella), (2) semi-spontaneous speech elicited by role -playing, (3) spontaneous speech in a conversation or dialogue (e.g., between patient and spouse, therapist or other person familiar to the patient), and (4) spontaneous speech elicited through interview with openended questions in an informal, natural c onversational setting giving the patient ample time to talk. These divisions may eliminat e some of the differences Armstrong (2000) points out, although even these divisions group po tentially different genres together, such as discourse from pictures and retelling a known story. Glosser, Weiner, & Kaplan (1988), found pictures elicited less verbal complexity than spontaneous production without pictures. However, Doyle et al. (1998), found no significant differences between story retelling with or without pictures for measures of ve rbal productivity, information content, grammatical complexity, verbal di sruption, and quality of grammatical form. Another explanation for these genre differences could be the instru ctions given to the speakers. Olness (2005) suggested that th e common instructions given for picture descriptions, “Tell me everything you see going on in this picture” does not specifically ask for temporal organization. In her st udy contrasting instru ction type, she found speakers more likely to produce narratives when asked for temporal sequencing, “Make up your own story about what happened, with a beginning, a middle, and an end.” As Armstrong (2000) states, the issue of defi ning discourse and matching genres with

PAGE 19

9 elicitation techniques remains to be resolved and finer distinction may help profile the way speaker’s with aphasi a use language skills in various settings. Methodologies Prins and Bastiaanse (2004) review the cu rrent methods of disc ourse analysis and outline the advantages and disadvantages of such analyses. They identify two methods of analysis: (1) rating features of the language produced on a pre-determined number of scales, such as phrase length and grammati cal forms, and (2) quantifying linguistic variables, such as mean length of utteran ce, and content/function words ratio. They advocate the use of linguistic or grammatical an alysis as they view aphasia as a linguistic disorder. They state that any macrostructu re impairments or pragmatic difficulties in discourse of speakers with aphasia are most lik ely due to linguistic impairments. Prins and Bastiaanse (2004) point out that discourse can be analyzed using either qualitative or quantitative methodology. Qualitative linguistic analyses use rating scal es such as the Rating Scale Profile of Speech Characteristics (RSPSC) and the Aphasia Severity Rating Scale (ASRS) both used in the Boston Diagnostic Aphasia Examination (BDAE; Goodglass et al., 2000; Prins & Bastiaanse, 2004) to classify aphasi a type. The spontaneous speech portion of the BDAE presents questions such as, “How are you today?” and “Tell me what happened to bring you to the hospital?” and then provides the Cookie Theft picture for the patient to describe (Goodglass et al., 2000). The rating scales are used to analyze the patient’s responses to the questions and pict ure description. The RSPSC uses a 7-point scale to measure melody, phrase length, articulatory agility, grammatical form, paraphasias, and word finding. Based on ratings on the RSPSC, a patient’s aphasia type

PAGE 20

10 can be identified (Prins & Bastiaanse, 2004). The ASRS is a 5-point scale measuring the speakerÂ’s ability for verbal communication. Quantitative analyses can be elicited from various stimuli as previously discussed, such as story retell or picture description. A commonly used met hod is the Quantitative Analysis of Agrammatic Production (QAAP; Saffran, Berndt, & Schwartz, 1989). This approach measures variables such as proporti on of closed class words, verb inflections, well-formed sentences, and embedding i ndex. Rochon, Saffran, Sloan-Berndt, and Schwartz (2000) conducted reliability testi ng of these measures and found test-retest variability ranging from .66 to .92 for the pr oportion well-formed sentences. Prins and Bastiaanse (2004) suggest this finding is not due to instability of the QAAP, but to instability in the performance of agrammatic speakers themselves. They also suggest that the findings indicate a need for caution wh en using the QAAP to measure changes in agrammatic speech. Thompson, Lange, Schneider, and Shapiro (1997), offer another method of quantitative analysis of agrammatic speech, in which they analyzed verbargument structures. Based on their results, they conclude that agrammatic speakers produce fewer verbs in general and produce ve rbs with simpler argument structure when compared to non-brain-damaged speakers. Although this study added to the knowledge of the underlying disorder in BrocaÂ’s aphasia, it has yet to be replicated in speakers with other aphasias. Consequently, it is not po ssible at this time to determine the broad clinical implications of this fi nding (Prins & Bastiaanse, 2004). Limitations and deficiencies in current research Prins and Bastiaanse (20 04) note that a primary limitation in the value and generalizability of discourse analysis for aphasia research is the speaker with aphasia. The stability of the production of aphasic di scourse across repeated testing is unknown;

PAGE 21

11 therefore, the value of language samples as representations of the speakerÂ’s ability or as treatment outcomes is unknown. Prins and Bast iaanse (2004) present the need for more group studies to develop norms for statistically reliable impr ovement, so that individual studies could be better interpreted. In a ddition, as in the case of the method proposed by Thompson et al. (1997), analyses need to be applied to various types of aphasia to determine if research findings provide a me thod applicable in c linical settings. A significant reason for the limited research available on grammatical analysis is the nature of such analysis. It is time-c onsuming and demanding of resources, skills, and knowledge not always available in clinical or academic settings. Researchers must have significant knowledge of linguistics and aphasi ology to conduct the requisite grammatical analyses and apply the results to clinical pract ice (Prins & Bastiaanse, 2004). Despite these limitations in the use of quantitative analysis, Prins and Bastiaanse (2004) consider it an important method for investigating discours e. Like Armstrong (2000), they advocate using discourse analys is for developing theoretical models, designing therapy and measuring its efficacy, because it provi des the best information on a patientÂ’s everyday use of language. Purpose The aim of the current study is to compar e the effects of two treatments on the production of discourse using grammatical analysis that quantifies changes in the production of various grammatical units and fo rms before and after treatment. Analyses address changes at the word, sentence, a nd information levels. Analyses will be conducted to compare the two treatment type s, the two conditions, and each condition within each treatment type. Participants (Raymer, Ciampitti et al., in press; Raymer, Singletary et al., in press) were assigned to either the semantic-phonologic treatment or

PAGE 22

12 the compensatory gestural-verbal treatmen t for aphasia; within each treatment, participants were trained on noun access or verb access. Research question 1 Will gestural verbal treatment have a greater effect on the production of discourse than semantic-phonological treatment? It was hypothesized that changes in discourse measures will be higher in the compensatory gestural-verbal treatment than the semanticphonologic treatment. Research question 2 Will training one word type (e.g., nouns or verbs) lead to increased production of that word type? It was hypothesized that word-type trained will result in increased production of that word type. Research question 3 Will increasing noun and verb production correla te with increases in information? It was hypothesized that increased noun and ve rb production will correlate with increases in UNIs. Research question 4 Will production of ‘good sentences’ correlate with increased information? It was hypothesized that increased production of ‘good sentences’ will correlate with increases in UNIs.

PAGE 23

13 CHAPTER 2 METHODS Participants Seventeen individuals with aphasia between the ages of 38 and 81 participated in this study. Their demographic information is shown in Table 1. All participants completed an experimental treatment study by Raymer, Ciampitti et al., (in press) and Raymer, Singletary et al., (i n press), contrasting a sema ntic-phonologic treatment (SP) and a compensatory gestural and verbal treatme nt (GV). Each treatment was divided into a noun or verb based treatment. Discourse samp les elicited preand posttreatment were available from five participants from the semantic-phonologic treatment and twelve participants from the verbal-gestural treatment. Eligibility for the treatment study was based on dia gnosis of a unilateral left hemisphere cerebrovascular accident (CVA) confirmed by structural MRIs. The CVA must have caused aphasia that continued for mo re than four months prior to participation in the current study. All partic ipants had to demonstrate a word retrieval impairment of less than 75% accuracy for nouns and verbs as measured by the Western Aphasia Battery (WAB; Kertesz, 1982), the Boston Naming Test (BNT; Kaplan Goodglass, & Weintraub, 2001) and the Action Naming Test (ANT; Obler, & Albert, 1 986). In add ition, participants could not have motor speech im pairments greater than moderate severity, determined by scores greater than 2.0 on th e WAB repetition subtest. Participants represented a variety of aphasia types and severity as noted on the WAB. All participants spoke English as a first language.

PAGE 24

14 In addition to these measures, particip ants receiving the gestural and verbal treatment were given the Florida Apraxia Screening Test-Revised (Rothi, Raymer, & Heilman, 1997). Results of this test showed that all participants had mild to moderate limb apraxia except one who demonstrated a severe limb apraxia. The table below describes the order of treatment for each participant, word type, and demographic characteristics. Table 1. Participant Demographics Participant Treatment Word Type AgeEducationAphasia Type WAB Score 1 Semantic Noun 38 12 Nonfluent 53.9 2 Semantic Noun 68 14 Nonfluent 44 3 Semantic Verb 74 18 Nonfluent 59.9 4 Semantic Verb 66 14 Fluent 77 5 Semantic Noun 81 12 Nonfluent 74.5 6 Gestural Verb 56 10 Nonfluent 31.6 7 Gestural Noun 70 10 Nonfluent 33 8 Gestural Noun 69 8 Nonfluent 54.5 9 Gestural Verb 73 14 Fluent 78.2 10 Gestural Noun 67 12 Nonfluent 68.7 11 Gestural Verb 40 16 Nonfluent 45 12 Gestural Verb 64 12 Fluent 58.5 13 Gestural Verb 49 12 Nonfluent 58 14 Gestural Verb 50 14 Nonfluent 38 15 Gestural Noun 70 14 Fluent 21.3 16 Gestural Verb 52 12 Fluent 43.4 17 Gestural Noun 51 12 Nonfluent 65.2 From: Raymer, Ciampitti et al., in pre ss; Raymer, Singletary et al., in press Procedures for Semantic-Phonologic Trea tment and Gestural + Verbal Treatment The treatment followed a multiple baseline design across participants and stimulus sets. Training took place 2-4 times per week for a total of 10 sessions. Daily probes were administered in 8-10 sessions. SP traini ng sessions consisted of a word retrieval treatment in which the examiner introduced the target picture and provided the name for the participant to repeat three times. With the picture displayed, the examiner asked four yes/no questions, two about semantic ch aracteristics and two about phonologic

PAGE 25

15 characteristics of the target word. This pr ocedure was used to aid the participants in developing a strategy that follows the normal word retrieval process. GV training sessions consisted of four steps for each of the 20 treatment stimuli. The treatment began with the clinician pres enting the picture and verbally modeling the target word and gesture matchi ng the picture. The participan t then repeated the target word and gesture three times. The clinician then performed the gesture without the word and the participant imitated three times. If necessary, the clinician aided the participant in manipulation of the limbs. The third step required the clinician to display the target verb and the participant to verbally repeat it three times. Then the clinician repeated the word again, syllable by syllable if necessary. After pausing for five seconds, the clinician prompted the participant to say the name and perform the gestur e. If correct, the participant was reinforced; if the participan t was incorrect, the corr ect model was given. In total, the participants a ttempted each target nine times per training session. Grammatical Analysis Sample As part of the treatment study desc ribed above, participants produced conversational samples in response to scripted questions before beginning treatment and after completion. The questions between the caregiver and participant were about food, events, and hobbies; questions be tween the clinician and participant focused on a set of pictures of family members, famous people, and historic events. Conversations were videotaped and transcribed by an examin er blind to the tr eatment condition. Scoring Conversational samples were analyzed using Systematic Analyses of Language Transcripts (SALT; Miller & Chapman, 1991). In addition, the QAAP (Saffran, Berndt,

PAGE 26

16 & Schwartz, 1989) was used to guide the choi ce of syntactic struct ures for coding. The table below provides the word -classes and sentence types coded in this analysis. Table 2. Word Classes and Sentence Type Codes Word Level Code Sentence Level Code Nouns Produced [NP] Minimal Sentence [SM] Pronouns Produced [PRO] Good Sentence [SG] Modifiers Produced [MP] Elliptical [SEL] Modifying Noun [MN] One Word Response [S1] Modifying Verb [MV] Question [SQ] Verbs Produced [VP] Irrelevant Response [RI] Word Level Code Sentence Level Code Auxiliary Verbs [VA] New Information [IN] Verb Infinitive [VN] Intransitive Verb [VI] Transitive Verb [VT] Ditransitive Verb [VDI] Automatic Speech [AS] Phonological Error [EPH] Semantic Error on Verb [EV] Semantic Error on Noun [EN] Dialectal differences such as the persona l pronoun “I” pronounced as the southern “ah” and transcribed as “ah” were coded a ppropriately. Contracted words such as “wanna”, “dunno”, and “gonna” were coded fo r each word. For example “dunno” was coded as [va][vp][vt] for “don’t know.” As part of the SALT program, utterances identified as mazes were placed in parent hesis and not coded or used for word or utterance count. Mazes include repetitions, self-corrections incoherent words strings, and filled pauses such as “um” and “uh.” “You know” and “like” were also considered mazes if not relevant to the ongoing statem ent or question. For example when “you know” did not mean “understand” and “like” did not mean “preference” or have semantic content. Repetitions included self-repetitions and re peating the examiner or caregiver. Self-corrections were defined as any repetiti on of a phrase or word with a change. For

PAGE 27

17 repetitions and self-corrections only the last iter ation of the phrase or word was coded (Miller & Chapman, 1993). Discourse was also scrutinized for utteranc es with new information (UNI), defined as a coherent utterance providi ng information not previously given in the conversation. This was used as a sentence level code but did not indicate the entire sentence as new information. For example, a participant may have previously stated he had a son and then stated he had a son and a daughter. Good sentences included the following structures: noun +main verb, noun + copula + adjective, noun + verb + noun, and noun + c opula + prepositional phrase (preposition and noun phrase). Minimal sentences were those missing an obligatory word or inflection, but semantically qualifying as a se ntence (Saffran, Berndt, & Schwartz, 1989). Elliptical sentences were those judged to be appropriate answers in conversation but missing a major component; for example, “T o the mall” as a response to a question, “Where did you go yesterday?” One-word responses were any one word direct answer to a question such as “yes”, “okay”, or “vanilla.” Questions were defi ned as any request for information that was not a re petition of a question asked by the examiner or caregiver. Irrelevant responses were those not relate d to the question or topic of discussion including neologisms. This sentence level code most ofte n accompanied the “automatic speech” word-level code used for statements commonly used by a participant in response to any question or statement. “Automatic speec h” could be an isolated statement or part of a statement with relevant information. All counts were based on number of instances per utterance, in order to c ontrol for differences in discourse length between samples.

PAGE 28

18 Reliability Reliability was conducted during developm ent of the coding system and initial implementation. The coding system was developed by the Discourse Group at the University of Florida Language over the Li fespan Lab, consisting of a linguist, two speech pathologists, and a speech pathology und ergraduate student. Transcripts were independently coded by all members. Afterw ard, the group discussed the current codes, the application of codes, and any changes th at were necessary until a complete coding system with agreed upon definitions was complete Reliability of the transcript scoring is currently in progress. Statistical Analysis SPSS was used to analyze the coded transcri pts. Separate anal yses were conducted for treatment type (semantic-phonologic versus gestural + verbal), trained word type (nouns versus verbs) and trained word type within each treatment (gestural + verbal noun versus gestural + verbal verb). Means, st andard deviation, and st andard error of mean were calculated. Changes in performances were tested using the Wilcoxon Signed Ranks Test. Correlations between change scores (e .g., post-score minus pre-score) were also calculated.

PAGE 29

19 CHAPTER 3 RESULTS The results of the discourse analyses are reported in three ways, by the percent of people who changed on a given measure, illu strated in Figures 1-3, using Wilcoxon Sign Rank statistics and correlations of change sc ores. In many of the treatments, the number of patients was very sma ll, so findings with p -values of .25 and below are reported. Treatment Type: SP versus GV Word Level Both SP and GV treatments had positive effects on the number of nouns per utterance, as shown in Figure 1. Among pa rticipants from the SP treatment, 3 of 5 participants produced more nouns after treatment; however, this did not reach significance, ( Z = -.674; p > .50). Among participants in the GV treatment, 8 of 12 produced more nouns post treatment wh ich was marginally significant, ( Z = -1.804, p < .08). Only recipients of the GV treatment showed an increase post treatment in the number of modifiers produced per utterance with 10 of 12 participants showing this increase, ( Z = -1.833; p = .06). In contrast, only 1 of 5 participants receiving the SP treatment increased on this measure, ( Z < 1). See Figure 1. There were no other significant changes due to treatment type in other word level measures: pronouns per utterance, verbs per utterance, or automatic statements per utterance.

PAGE 30

20 0 10 20 30 40 50 60 70 80 90Nouns ProducedModifiers ProducedOne Word ResponseGood SentencesWord Structure% of Participants with Increased Production Semantic-Phonologic Trt Gestural-Verbal Trt Figure 1. Treatment Type: A comparison of the pe rcentage of participants with increased production of measures with significant increases. Sentence Level The SP treatment increased one-word responses in 4 of 5 participants, with a trend toward significance, ( Z = -1.483; p < .20). The GV treatment decreased one-word responses in 8 of 12 participants with marginal significance, ( Z = -1.647; p < .10). The GV treatment increased number of good sentences for 8 of 12 participants, however it was not significant ( Z = -1.098; p = .272). The SP treatment increased number of good sentences for only 1 of 5 participants, ( Z <1). See Figure 1. There were no other significant changes due to treatment type in other sentence level measures: number of elliptical sentence s per utterance, numb er of good sentences

PAGE 31

21 per utterance, number of questions per uttera nce, or number of irre levant responses per utterance. Information Level There were no significant cha nges attributable to differenc es in treatment type in information measures: mean length of uttera nce in words, type token ratio, percent maze words, or UNIs per utterance. Word Trained: Noun or Verb Word Level Noun and verb based treatments had positive effects on the number of nouns and modifiers produced in discourse, as shown in Figure 2. Noun-based treatment (SP or GV) increased noun production for 5 of 8 partic ipants with margin al significance, ( Z = -1.660; p = .093). Verb based treatment increased noun production for 6 of 9 participants, but it was not significant, ( Z < 1). Production of modifiers increased for 5 of 8 participants of the noun-based treatments with a trend toward significance, ( Z = -1.260; p = .208). Verb based treatments increased producti on of modifiers for 6 of 9 participants, but it was not significant, ( Z < 1). See figure 2. There were no other significant changes due to treated word type in other word level measures: pronouns per utterance, verbs pe r utterance, or automatic statements per utterance. Sentence Level There were no significant changes due to treated word type in sentence level measures: number of one-word responses per utterance, number of elliptical sentences per utterance, number of good sentences per utte rance, number of questions per utterance,

PAGE 32

22 or number of irrelevant responses per utteran ce. However, when looking at all acceptable responses per utterance (good sentences, one-w ord responses, plus elliptical sentences), noun-based treatments increased these responses in 6 of 8 participants with a trend toward significance, ( Z = -1.260; p = .20). This is in contrast to verb-based treatments in which acceptable responses increased in only 3 of 9 participants, ( Z = 1.007; p >.20). See Figure 2. 0 10 20 30 40 50 60 70 80 90 100Nouns ProducedModifiers ProducedAcceptable Responses Maze per utteranceType Token RatioWord Structure% Participants with Increased Productio n Noun Verb Figure 2. Word-Type Trained: A comparison of the percentage of participants with increased production of measures with significant increases. Information Level Verb based treatment increased type-toke n ratio and percent maze words. Noun based treatment also increased percent maze words but not type-token ratio, as shown in Figure 2. Participants receiving verb-based treatment showed a significant increase in type-token ratio affecting 8 of 9 participants, ( Z = -2.084; p < .04). In contrast, only 3 of 8 participants of the noun-based treatm ents increased in type-token ratio, ( Z < 1).

PAGE 33

23 Percent of maze words increased for 6 of 9 participants of the noun based treatment with marginal significance, ( Z = -1.859; p = .063). Verb base d treatment increased percent of maze words for 7 of 9 part icipants, but it was not significant, ( Z = -1.011; p > .20). See Figure 2. There were no other significant changes attributable to differences in treated word type in information measures: mean lengt h of utterance in words or proportion of utterances with UNIs. Gestural-Verbal Treatment: N versus V Word Level The GV treatment, when treating nouns or verbs, had positive effects on the number of verbs and modifiers produced in discourse, as shown in Figure 3. Verb production increased for 6 of 7 participants of verb GV treatment, with a trend toward significance, ( Z = -1.352; p = .176). Noun based GV treatme nt increased production of verbs for 3 of 5 participants, but this was not significant, ( Z < 1). See Figure 3. Treating verbs increased producti on of modifiers for 6 of 7 participants with a trend toward significance, ( Z = -1.352; p = .176). Treating nouns in creased production of modifiers for 4 of 5 participants with a trend toward significance, ( Z = -1.483; p = .138). There were no other significant changes due to treated word type in the GV treatment for other word level measures: pr onouns per utterance, nouns per utterance, or automatic statements per utterance.

PAGE 34

24 0 10 20 30 40 50 60 70 80 90Verb ProducedModifiers Produced Good SentencesAcceptable Responses Mean Length of Utterance in Words Type Token Ratio Word Structure% Participants with Increased Productio n Noun Verb Figure 3. Gestural-Verbal Treatment: A comparison of the percentage of participants with increased production of measures with significant increases. Sentence Level There were no significant changes due to tr eated word type in the GV treatment for sentence level measures: number of one-w ord responses per utterance, number of elliptical sentences per utterance, number of good sentences per utterance, number of questions per utterance, or number of i rrelevant responses per utterance. Verb based GV treatment increased pr oduction of good sentences for 5 of 7 participants but it was not significant, ( Z = -1.183; p = .237). Noun based GV treatment also increased production of good sentences fo r 3 of 5 participants, this also was not significant, ( Z < 1). See Figure 3. Acceptable responses per utterance (good se ntences, one-word responses, elliptical sentences), increased with nounbased GV treatment in 4 of 5 participants, but it was not

PAGE 35

25 significant, ( Z = -1.214; p = .225). Verb-based GV tr eatment increased acceptable responses in only 3 of 7 participants, ( Z < 1). Information Level Treating verbs in the GV treatment had posit ive effects on mean length of utterance in words and type-token ratio, as shown in Fi gure 3. Mean length of utterance in words increased for 5 of 7 participants of th e GV verb treatment with a trend toward significance, ( Z = -1.352; p = .176). In contrast, noun based GV treatment increase mean length of utterance in words fo r only 2 of 5 participants, ( Z < 1). Type-token ratio increased for 6 of 7 part icipants of the verb based GV treatment with marginal significance, ( Z = -1.614; p = .10). In contrast noun based GV treatment increased type token ratio fo r only 2 of 5 participants, ( Z < 1). There were no other significant changes attributable to differences in treated word type in the GV treatment for information m easures: percent maze words or proportion of utterances with UNIs. SP N versus V Word Level There were no significant changes for either of the two participants of the verb based SP treatment or for any of the three pa rticipants of the nouns based SP treatment, attributable to treated word type in the SP treatment for word level measures: nouns per utterance, pronouns per utterance, modifiers per u tterance, verbs per utterance, or automatic statements per utterance. Noun based SP treatment increased noun production for 2 of 3 participants, however; it was not significant, ( Z = -1.069; p = .285).

PAGE 36

26 Sentence Level Verb-based SP treatment did not in crease production of good sentences or acceptable responses for either of the two participants. Noun-based SP treatment increased production of good senten ces in 1 of 3 participants,,( Z <1) and increased acceptable responses in 2 of 3 participants,( Z = -1.069; p = .285) however, these were not significant. There were no other significant changes due to treated word type in the SP treatment for sentence level measures: numbe r of one-word responses per utterance, number of elliptical sentences per utterance, number of questions per utterance, or number of irrelevant responses per utterance. Information Level Verb based SP treatment had positive eff ects on type token ratio and percent maze words. Noun based SP treatment was associ ated with the percent of maze words produced. Noun based SP treatment increased pe rcent maze words for 3 of 3 participants with a trend toward significance, (Z = 1.604; p = .109). Verb based SP treatment also increased percent maze words for 2 of 2 participants, (Z = -1.342; p < .20). Verb based SP treatment increased type toke n ratio for 2 of 2 pa rticipants with a trend toward significance, ( Z = -1.342; p < 2.0). In contrast, noun based SP treatment increased type token ratio fo r only 1 of 3 participants, ( Z < 1). There were no other significant changes attrib utable to differences in treatment type in information measures: mean length of utte rance in words or proportion of utterances with UNIs. Noun based SP treatment increased proportion of utterances with UNIs for 2 of 3 participants, however; it was not significant, ( Z >1).

PAGE 37

27 Correlations Correlations were calculated for changes scores in the number of nouns produced, number of verbs produced, number of modi fiers produced, UNIs produced, and number of acceptable responses produced. SP Treatment Change score of the number of nouns pr oduced was significantly correlated with change scores of UNIs ( p = .01), verbs ( p = .06), modifiers ( p = .10), and acceptable responses ( p = .15). As the number of nouns per utte rance increased so did the number of UNIs, modifiers, and acceptable responses. Ho wever, the number of verbs per utterance decreased as the number of nouns increased. Change score of the number of verbs pr oduced was significantly correlated with change score of nouns and UNIs ( p = .20). As the number of nouns and UNIs increased the number of verbs decreased. Change score of the number of verbs produced was not significantly correlated with other change scor es: modifiers and acceptable responses. Change score of the number of modifier s produced was significantly correlated with change scores of nouns, acceptable responses ( p = .03), and UNIs ( p = .03). As the number of modifiers per utterance incr eased so did the number of nouns, acceptable responses, and UNIs. Change score of the number of modifiers produced was not significantly correlated w ith other change scores: verbs. Change score of the number of acceptabl e responses produced was significantly correlated with change scores of nouns, modifiers, and UNIs ( p = .096). As the number of acceptable responses increased, so did th e number of nouns, modifiers, and UNIs. Change score of the number of acceptable res ponses was not significantly correlated with other change scores: verbs.

PAGE 38

28 GV Treatment Change score of the number of nouns pr oduced was not significantly correlated with other change scores: verbs, m odifiers, UNIs, acceptable responses. Change scores of the number of verbs pr oduced was significantly correlated with change score of acceptable responses ( p = .02). As the number of verbs per utterance increased, so did the number of acceptable responses. There were no other significant correlations between the change score of nu mber of verbs produced and other change scores: nouns, modifiers, and UNIs. Change scores of the number of modifier s produced was significantly correlated with change scores of verbs ( p =.04) and the correlation with acceptable responses approached significance ( p = .075). As the number of modifiers per utterance increased, so did the number of verbs and acceptable responses. There were no other significant correlations between the change score of number of modifi ers produced and other change scores: nouns and UNIs.

PAGE 39

29 CHAPTER 4 DISCUSSION This study tested the hypothesis that nami ng treatments for aphasia would lead to quantifiable changes in discourse measures. Mo re specifically, the purpose of this study was to compare the effects of two aphasia treatments conducted by Raymer, Ciampitti, et al.(in press) and Raymer, Si ngletary, et al. ( in press), on production of grammatical and lexical aspects of discourse. There were four research predictions. Fi rst, it was predicted that changes in grammatical units and fo rms would be higher in the compensatory gestural-verbal treatment than the semantic -phonologic treatment; results of the treatment type analysis support this hypothesis. S econd, it was predicted th at there would be increases in the use of the part icular word type that was tr ained. This effect was found in the GV verb treatment with a trend towa rd significance and in the SP noun treatment which was not significant. The third pr ediction was that increased noun and verb production would correlate with increases in UN Is. This correlation was significant in the SP treatment condition and but not in the GV trea tment. The fourth and final prediction was that increased production of ‘good sentences’ would correlate with increases in UNIs. This finding was present in both GV treatments, although neither effect was significant. Treatment Type Comparison of GV and SP treatment reveal ed more significant changes in the discourse of the participants of the GV treatment. Participants receiving the GV treatment increased producti on of both nouns and verbs; wh ereas, those receiving the SP

PAGE 40

30 treatment only increased production of nouns. This is consistent with previous findings from gestural therapies and in the literat ure. Hadar, Wenkert-Olenik, Krauss, and Soroker, (1998) found that sp eakers with aphasia produced a higher number of gestures when word finding decreased. Th ey concluded that gestures in creased lexical retrieval. Rauscher, Krauss, and Chen (1996) found th at when speakers without neurological impairment were not allowed to use gestures word retrieval decreased. Thus, both of these studies provided evidence for gestures faci litating word retrieval. Treatment studies have also found this relationship between ge stures and word retrieval. Pashek (1997) found that verbal plus gestural treatment si gnificantly increased na ming compared to the effects of verbal-only treatment. Miranda Rose and Jacinta Douglas (2001) found iconic gestures significantly improved object naming in participants with phonologic impairments, but not in participants with se mantic impairments. They concluded that iconic gestures prime impaired phonological acce ss, storage, and enc oding processes, but not impaired semantic storage. Rose and D ouglas state that this finding is in support of Krauss and HadarÂ’s (as cited in Rose & D ouglas, 2001) theory of lexical gesture and speech production. In this theory, gestur es facilitate lexical access by activating gesturally represented featur es of the message. Krauss and Hadar state that this activation occurs before articu lation of the word, in LeveltÂ’s (as cited in Rose & Douglas, 2001) formulation stage. Krauss and Hadar (as cited in Rose & Douglas, 2001) imply that priming occurs from the kinesic monitor to the formulator level which contains the grammatical encoder and phonological encoder (Rose, Douglas, & Matyas, 2002). However, Rose and Douglas (2001) state that th e precise level at which priming occurs is not clear in Krauss and HadarÂ’s model. Based their findings that gestures only

PAGE 41

31 facilitated speakers with phonological impairme nts, Rose and Douglas (2001) concluded that gesture-related information enters th e speech production system at the phonological level. However, they also point out that gest ures may facilitate word retrieval at both the phonological and lemma level. While it is more intuitive to suggest gestures facilitating at the lemma level or even befo re the formulator level at the conceptualizer, facilitation at the phonological level explains how the GV treatment increased nouns and verbs. Applying Rose and Douglas’s theory to the current findings may explain why the GV treatment increased nouns and verbs a nd the SP treatment only increased nouns. Following Rose and Douglas’s conclusions, the GV treatment activated the phonological processes in producing nouns and verbs and not the semantic processes of nouns and verbs that might lead to differe ntial activation of word classes as seen in the SP treatment. Alternately, the GV treatment may have incr eased production of verbs in addition to nouns simply because the speakers were ‘act ing out’ the verb. Performing a physical action associated with the word may have b een enough to activate retrieval of the verb. In addition to increasing nouns and verbs, the GV treatment increased modifiers. Similar to the possible effect on verbs, performing gest ures that describe the object or verb may have activated associated modifiers by spread ing activation. As m odifiers provide more information about the topic, increased producti on of modifiers may have contributed to the increase in UNIs. However, the impact of modifiers on increas ed production of UNIs may not be significant, because the SP treatmen t also increased UNIs but not modifiers. The SP treatment had a higher percentage of participants increase in one-word responses. It is arguable whet her this is an improvement. While it is not a quantitative improvement as 4 of 5 participants we re nonfluent, there may be a qualitative

PAGE 42

32 improvement. As discussed below, the SP noun condition did increase production of UNIs. It may be that although the participan ts were using more one-word responses, the response was correct and relevant and provided more information than before treatment. Trained-Word Type Comparison of noun-based to verb-bas ed treatments showed that higher percentages of particip ants in the verb-based treatmen ts increased their production of nouns, verbs, modifiers, TTR, mazes per utteranc e, and UNIs. However, of all of these measures, only the increase in TTR was signi ficantly higher post treatment. Training verbs may have increased TTR more than tr aining nouns because verbs have been found to be more difficult for speakers with aphasi a to retrieve (Berndt, Burton, Haendiges, & Mitchum, 2002; Marshall, Pring, & Chiat, 1998) In addition, Thompson et al. (1997) found that speakers with agrammatic aphasia have been found to produce verbs with simple argument structure. Based on the results of their stud y, Thompson et al. concluded that verb argument structure is important for verb retrieval. Thus, in the current study increasing verbs pr ovided the speakers with access to a class of words and argument structures that was previously di fficult to retrieve, increasing the lexical diversity of their discourse. In the verb condition, UNIs increased as di d production of nouns, verbs, modifiers, and TTR. Thus, participants were able to increase produc tion of content words and convey more information using a greater vari ety of words. Percentage of maze words also increased in the verb condition. This may have been due to an increased availability of words which, in turn, led to increas ed attempts at verbal responses.

PAGE 43

33 Trained-Word Type within Treatment Type SP Noun vs. Verb Results of comparing trained-word type within the SP treatment are based on three participants in the noun conditi on and two participants in the verb condition, making it difficult to achieve statistic al significance for any meas ure. Training nouns increased nouns, acceptable responses, and UNIs. All of these participants were nonfluent; thus, as they were able to produce more nouns they were able to produce different types of words, leading to production of more acceptable respons es with more UNIs, but not significantly more. However, as they were able to access more nouns and types of words, they made more mistakes resulting in an increased pe rcentage of maze words. It should be emphasized that acceptable responses include one-word and elliptical responses as well as good sentences; thus, the finding that noun training increased acceptable responses is not necessarily contradictory to the findings stated below, which demonstrate that verb training increases sentence production (Berndt et al., 2002; Marshall et al., 1998). Similar to the effect in the noun conditi on, participants in this condition also produced an increased percentage of maze word s, presumably as a result of increases in overall word availability, leading to more lexical in trusions. These were again accompanied by increases in TTR, supporting the idea that there was an overall increase in lexical availability. The increased le xical availability may have overloaded the sentence production mechanism by providing mo re activated words to choose from and organize into a sentence, thus resulting in more mistakes and mazes. Crockford (1991) offers another explanation for finding in creases in ‘repair turns’ accompanied by increased functional comm unication in a patient with aphasi a. That study revealed that the patient’s wife did not need to offer as mu ch help during ‘repair turns’ because of the

PAGE 44

34 patient’s increased communication ability. This phenomenon was also observed in several participants in the current study. Post-treatment transcripts contained fewer utterances from the caregivers than pre-treatment transcripts. Significant gains in SP noun treatment were offset by a very small N (i.e., three participants), making any kind of statistical inference impossibl e. In addition, participants of the SP noun treatment were the farthest post onset at 75, 93, and 120 months post, (average 96 months) versus the rest of the participant population who were 5 to 62 months post onset (average 24.7 months). Theref ore, the lack of effects may be the result of the generalization effects of naming treatment to discourse being limited to a specific time period after the onset of aphasia, rather th an to inadequacy of the treatment. In summary, noun based and verb based SP treatments increased percent maze words in all participants. This finding may be attributable to increased attempts caused by an increase in the availability of words. Crockford’s (1991) explanation of decreased help in ‘repair turns’ by th e spouse can also apply in th is condition. Although percent maze words increased in all participants of SP treatment, it is difficult to identify the cause of this effect, as only five peopl e participated in this treatment. GV Noun vs. Verb Within the GV treatment, the verb condition resulted in the highest number of gains as well as the most significant gains. The ve rb condition had a significantly greater effect on production of verbs, MLU in words, a nd TTR. The verb condition also had higher increases in modifiers and good sentences pr oduced, though these were not significantly higher than the analogous increases in the noun condition. Increased production of good sentences in th e verb condition is consistent with findings from several studies linking verb retrieval and sentence production. Berndt,

PAGE 45

35 Burton, Haendiges, and Mitchum (2002), found th at speakers with grea ter impairment in verb retrieval than noun retrie val also had more impaired sentence production. It follows then that improving verb retrieval would improve sentence production (Berndt et al., 2000; Berndt, et. al, 2002; Marsha ll et. al, 1998). However, th ere have been cases that do not support a connection between verb retrieva l and sentence production, as demonstrated by the patient described in Berndt, Haendi ges, and Wozniak (1997) who had severe anomia characterized by significantly highe r verb retrieval th an noun retrieval but impaired sentence comprehension and pr oduction. GV verb treatment also was associated with increased MLU in words a nd TTR; in other words, the GV verb condition increased the production of good sentences, the length of utterances in words, and the lexical diversity of utterances However, this treatment did not increase maze production, as did both variations of the SP treatment. These findings suggest th at the participants were able to retrieve the correct words in th e correct order the first time and did not need several attempts. The process of gesturing may have strengthened the neural connections in the representations of verbs enough to decrease the need for multiple attempts to produce the correct verb. This may have occu rred because gestures do not result in the same extent of spread of activ ation in the semantic system if gestures facilitate lexical access at a post-semantic stage. This is an important observation as participants in this treatment would then be more efficien t speakers. Interestingly, although maze production was not increased, UNIs did not in crease in either GV treatment condition either. Thus, the speakers used more type s of words, had longer responses, with increased modifiers, verbs, and good sentences, but they did not produce more utterances with new information. The relationship betw een increases in the production of nouns and

PAGE 46

36 verbs and increases in UNIs is difficult to cl early establish due to methodological issues. UNIs were an utterance level code, with only one allowed per utteranc e; consequently, if there were several pieces of new informati on in an utterance this would have been missed. This presents the need for a more quantitative information measure combining aspects of the UNI and CIU. The only measure in which GV noun conditi on had greater effect was acceptable responses. This coincides with the treated-w ord type results in which noun conditions led to greater gains than verb conditions in acceptable responses. Acceptable responses included one-word responses but not UNIs, as a result, this measure is not particularly indicative of improved discourse. However, an increase in acceptable responses does indicate the discourse is eas ier to understand for the comm unication partner even if the amount of information conveyed has not increased. Mazes Production of mazes increased in noun and verb conditions in the SP and GV treatments, but was only significant in SP noun and verb conditions. Increases of maze production were associated with increases in TTR in GV and SP verb conditions suggesting a link between increased lexical av ailability and maze pr oduction. All but one SP participant was nonfluent; therefore, any in crease in lexical access would provide the speaker with more words to retrieve a nd organize into sentence form, possibly overburdening the sentence production mechanism, leading to more mistakes or mazes. Mazes are typically not included in anal yses such as QAAP (Saffran, Berndt, & Schwartz, 1989) and SALT (Miller & Chapma n, 1991). Based on the findings of the current study mazes should be considered in fu ture analyses to investigate lexical access and sentence formulation.

PAGE 47

37 Limitations In general, there were few significant gains ( p < .05) due to the small N in each treatment. When comparing tr eatment type, statistics were run on 12 participants in the GV treatment and 5 participants in the SP treatment. All participants who completed the GV treatment and all but one of the part icipants who completed the SP treatment were analyzed in this study. When comparing trained-word type within each treatment, statistical analysis was further compromise d by even smaller groups: SP verb treatment had two participants and SP noun treatment had three participants, while GV verb treatment had seven participants and GV noun tr eatment five participants. Due to the variability within and between the discourse productions of speakers with aphasia, it is difficult to make generalizations from a sma ll sample of speakers (Prins & Bastiaanse, 2004). A second limitation of the treatment de sign was the mixture of aphasia types included. In each treatment and condition th ere were participants with fluent and nonfluent aphasia. Because nonfluent and fluent aphasias arise from different lesion sites and result in different type a nd level of impairment, this may have contributed to the lack of significant findings in the study. Specifica lly, analysis was completed on groups and not individuals; thus, the mixture of aphasia types may ha ve affected the results. The discourse production of a speaker with nonflu ent aphasia is typical ly characterized by one-word responses or short phrases. The disc ourse of a speaker with fluent aphasia is characterized by longer utterances with mini mal information. Thus, combined analysis of grammatical components of the discourse of nonfluent and fluent speakers may have obscured the true treatment effects present.

PAGE 48

38 A third limitation of the treatment design was the mixed genres of discourse used as stimuli. Participants discussed favorite foods and hobbies, people and events in family pictures, and pictures of fam ous people or events. Although it has not been discussed in past studies or reviews, differences in di scourse production may arise when using family pictures and pictures of famous people not personally known to the speaker. As Armstrong (2000) notes, some research suggest s there are differences in the discourse elicited by various stimuli such as the ones in the current study, picture description, opened-ended questions, and discussing fam ily members and memories. In addition, pictures of people at an event in which the ev ent is clear in the picture may elicit different discourse than a picture of a single person or group of people not in an obvious setting or event. Armstrong (2000) notes that single picture stimuli ma y not elicit a narrative with orientation, precipitating action, and resolution, but elicit desc ription of a situation. In their categorization of discour se types, Prins and Bastiaan se (2004) separate discourse from situational pictures and discourse elicited through interview with open-ended questions. Differences previously found be tween discourse with and without picture stimuli include: less verbal complexity with pictures (Glosser, Weiner, & Kaplan, 1998), higher efficiency scores without pictures (Doyle et al. 1995), and higher cohesive harmony without pictures (Armstrong, 1988). T hus, our analysis might have benefited from being limited to one of the three discourse types used. Summary Despite the limitations and few significan t findings of the current study, it offers several important contributions to aphasia di scourse research. The relationships among word classes, sentence structure, and un its of information found provide a strong argument for grammatical analysis as a viable method of measuring changes in discourse.

PAGE 49

39 Measuring changes in information is particul arly important whether it be through use of the UNI presented in this study or other sim ilar measures. A comparison of the UNI and CIU should be conducted in the future to clea rly identify differences and ideal uses for each. Mazes should be considered as part of future discourse analys es. The current study found increases in mazes when word retrieval increased. This is important theoretically in regards to theories of lexi cal access and activation. Future research should continue to investigate the various components necessary for conveying information such as topic relevancy and coherence. Findings of the cu rrent study also point to a need for further research elucidating the differential eff ects of training nouns and verbs and of the methods of training. As seen here, adding a component used by many speakers in daily conversation, gestures, increased verbal output and improved co mmunication. Although the changes in discourse f ound in the current study were not, for the most part, statistically significant, this study does provide evidence th at naming treatments can lead to changes in discourse. This is high ly important as improved discourse or communication should always be the ultim ate goal of any aphasia treatment.

PAGE 50

40 LIST OF REFERENCES Arbuckle, T.Y., Gold, D., Fra nk, I., & Motard, D. (1989). Speech of verbose older adults: How is it different? Paper presented at the annual meeting of the Gerontological Society of America, Minneapolis, MN. Armstrong, E. (2000). Review: Aphasic disc ourse analysis: The story so far. Aphasiology, 14 (9), 875-892. Armstrong, E.M. (1988). A cohesion analysis of aphasic discourse Unpublished MA thesis, Macquarie University, Sydney. Bastiaanse, R., Edwards, S., & Kiss, K. ( 1996). Fluent aphasia in three languages: Aspects of spontaneous speech. Aphasiology, 10, 561-575. Berko-Gleason, J., Goodglass, H., Obler, L., Green, E., Hyde, M., & Weintraub, S. (1980). Narrative strategies of apha sics and normal-speaking subjects. Journal of Speech and Hearing Research, 23, 370-382. Berndt, R.S., Burton, M.W., Haendiges, A. N., & Mitchum, C.C. (2002). Production of nouns and verbs in aphasia: Eff ects of elicitation context. Aphasiology, 16 (1/2), 83-106. Berndt, R.S., Haendiges, A.N., & Wozniak, M. A. (1997). Verb retrieval and sentence processing: Dissociation of an established association. Cortex, 33 (1), 99-114. Bird, H., & Franklin, S. (1996) Cinderella revisited: A comparison of fluent and nonfluent aphasic speech. Journal of Neurolinguistics, 9, 187-206. Boles, L. (1998). Conversational discourse analysis as a method for evaluating progress in aphasia: A case report. Journal of Communication Disorders, 31, 261-274 Boyle, M. (2004). Semantic f eature analysis treatment for an omia in two fluent aphasia syndromes. American Journal of Speech -Language Pathology, 13 (3), 236-249. Brenneise-Sarshad, R., Nicholas L.E., & Brookshire, R.H. (1991). Effects of apparent listener knowledge and picture stim uli on aphasic and non-brain-damaged speakersÂ’ narrative discourse. Journal of Speech and Hearing Research, 34, 168176.

PAGE 51

41 Cherney, L.R., & Canter, G.J. (1993). Informa tional content in the di scourse of patients with probable AlzheimerÂ’s disease and patients with right brain damage. Clinical Aphasiology, 21, 123-134. Crockford, C. (1991). Assessing functiona l communication in aphasic adults: A comparison of three methods. Unpublished undergraduate dissertation. Department of Speech, University of Newcastle upon Tyke, UK. Doyle, P.J., Goda, A.J., & Spence, K.A. (1995). The communicative informativeness and efficiency of connected discourse by a dults with aphasia under structured and conversational sampling conditions. American Journal of Speech-Language Pathology, 4, 130-134. Doyle, P.J., McNeil, M.R., Spence, K.A., Goda A.J., Cottrell, K., & Lustig, P.P. (1998). The effects of concurrent picture presenta tions on retelling of orally presented stories by adults with aphasia. Aphasiology, 12, 561-574. Duffy, J.R. (1995). Motor speech disorders. St. Louis, MO: Mosby. Edwards, S. (1995). Profiling fluent aphasi c spontaneous speech: A comparison of two methodologies. European Journal of Disorders of Communication, 30, 333-345. Edwards, S., & Bastiaanse, R. (1998). Diversit y in the lexical and syntactic abilities of fluent aphasic speakers. Aphasiology, 12, 99-117. Gleason, J.B., Goodglass, H., Obler, L., Green, E., Hyde, M.R., & Weintraub, S. (1980). Narrative strategies of aphasic and normal-speaking subjects. Journal of Speech and Hearing Research, 23, 370-382. Glosser, G., & Deser, T. (1990). Patterns of discourse production among neurological patients with fluent language disorders. Brain and Language, 40, 67-88. Glosser, G., Weiner, M., & Kaplan, E. (1998). Variations in aphasic language behaviors. Journal of Speech and Hearing Disorders, 53, 115-124. Goffman, E. (1981). Forms of talk. Oxford: Basil Blackwell. Goodglass, H., Christiansen, J.A., & Galla gher, R. (1993). Comparison of morphology and syntax in free narrative and structured tests: Fluent vs nonfluent aphasics. Cortex, 29, 377-407. Goodglass, H., Kaplan, E ., & Barresi, B. (2000). Boston Diagnostic Aphasia Examination (3rd ed.). Philadelphia: Lippinco tt, Williams, & Wilkins. Grimes, J.E. (1975). The thread of discourse. Paris: Mouton.

PAGE 52

42 Hader, U., Wenkert-Olenik, D., Krauss, R ., & Soroker, N. (1998). Gesture and the processing of speech: Neuropsychological evidence. Brain and Language, 62, 107126. Halliday, M.A.K. (1985a). An introduction to functional grammar. London: Edward Arnold. Halliday, M.A.K. (1985b). Context of situation. In M.A.K. Halliday, & R. Hasan (Eds.) Language, context, and text: Aspects of language in a social-semiotic perspective (1-14). Geelong, Victoria: Deakin University Press. Harris, Z. (1963). Structural linguistics. Chicago: University of Chicago Press Harris, Z. (1988). Language and information New York: Columbia University Press. Hasan, R. (1985). The texture of a text. In M.A.K. Halliday, & R. Hasan (Eds.) Language, context, and text: Aspects of language in a social-semiotic perspective (70-96) Geelong, Victoria: Deakin University Press. Hier, D.B., Hagenlocker, K., & Shindler, A.G. (1985). Language disintegration in dementia. Brain and Language, 25, 117-133. Kaplan, E., Goodglass, H., & Weintraub, S. (2001). Boston naming test Philadelphia, PA: Lea and Febiger. Kertsz, A. (1982). Western aphasia battery. New York: Grune & Stratton. Marshall, J., Pring, T., & Chiat, S. (1998). Verb retrieval and sentence production in aphasia. Brain and Language, 63, 159-183. Meyers, P.S. (1979). Profiles of communication deficits in patients with right cerebral hemisphere damage. In R.H. Brookshire (Ed.) Clinical aphasiology conference proceedings (38-46). Minneapolis: BRK Miceli, G., Silveri, M.C., Romani, C., & Caram azza, A. (1989). Variation in the patterns of omissions and substitutions of gra mmatical morphemes in the spontaneous speech of so-called agrammatic patients. Brain and Language, 36, 447-492. Miller, J. F., & Chapman, R. S. (1991). SALT: A computer program for the systematic analysis of language transcripts Madison, WI: University of Wisconsin. Nicholas, L.E., & Brookshire, R.H. ( 1993a). A system for quantifying the informativeness and efficiency of the conn ected speech of adults with aphasia. Journal of Speech and Hearing Research, 36, 338-350. Nicholas, L.E., & Brookshire, R.H. (1993b). A system for scoring main concepts in the discourse of non-brain-dama ged and aphasic speakers. Clinical Aphasiology, 21, 87-99.

PAGE 53

43 Nicholas, L.E., & Brookshire, R.H. (1995). Th e presence, completeness, and accuracy of main concepts in the connected speech of non-brain-damaged and aphasic adults. Journal of Speech and Hearing Research, 38, 145-156. Nicholas, M., Obler L.K., Albert, M.L., & He lm-Estabrooks, N. (1985). Empty speech in AlzheimerÂ’s disease and fluent aphasia. Journal of Speech and Hearing Research, 28, 405-410. Obler, L.K., & Al bert, M. (1986). Action naming test. Unpublished test. Olness, G.S. (2005). Discourse genre and verb performance in picture-elicited discourse of adults with aphasia. Presented at the Clinical A phasiology Conference, Sanibel Island, FL. Pashek, G.V. (1997). A case study of gestur ally cued naming in aphasia: Dominant versus nondominant hand training. Journal of Communication Disorders, 30, 349366. Prins, R., & Bastiaanse, R. (2004). Revi ew: Analysing the spontaneous speech of aphasic speakers. Aphasiology, 18 (12), 1075-1091. Rauscher, F.H., Krauss, R.M., & Chen, Y. (1996). Gesture, speech, and lexical access: The role of lexical movements in speech production. Psychological Science, 7 (4), 226-231. Raymer, A.M., Ciampitti, M., Holliway, B., Singletary, F., Blonder, L.X., Ketterson, T., Heilman, K.M., & Gonzalez-Rothi, L.J. Semantic-phonologic treatment for noun and verb retrieval impairments in aphasia. Manuscript submitted for publication. Raymer, A.M., Singletary, F., Rodriguez, A ., Ciampitti, M., Heilman, K.M., & GonzalezRothi, L.J. Gestural+verbal training effects for noun and ve rb retrieval in aphasia. Manuscript submitted for publication. Roberts, J, & Wertz, R.T. (1989). Comp arison of spontaneous and elicited oralexpressive language in aphasia. Clinical Aphasiology, 18, 479-488. Rochon, E.R., Saffran, E.M., Sloan-Berndt, R.S., & Schwartz, M. (2000). Quantitative analysis of aphasic sentence production: Further development and new data. Brain and Language, 72, 193-218. Rose, M., & Douglas, J. (2001). The differential facilitato ry effects of gesture and visualization processes on object naming in aphasia. Aphasiology, 15 (10/11), 977990. Rose, M., Douglas, J., & Matyas, T. (2002). Th e comparative effectiveness of gesture and verbal treatments for a speci fic phonologic naming impairment. Aphasiology, 16 (10/11), 1001-1030.

PAGE 54

44 Rothi, L.J.G., Raymer, A.M., & Heilman, K. M. (1997). Limb praxis assessment. In L.J.G. Rothi & K.M. Heilman (Eds.), Apraxia: The neuropsychology of action (6173). Hove, East Sussex, UK: Psychology Press. Saffran, E.M., Sloan-Berndt, R., & Schwartz, M. (1989). Th e quantitative analysis of agrammatic production: Procedure and data. Brain and Language, 37, 440-479 Schwartz, M.F., Saffran, E.M., Bloch, D.D., & Dell, G.S. (1994). Disordered speech production in aphasic and normal speakers. Brain and Language, 47, 52-88. Shadden, B. (1998). Information analyses. In L.R. Cherney, B.B. Shadden, & C.A. Coelho (Eds.) Analyzing discourse in co mmunicatively impaired adults (85-114). Gaithersburg, MD: Aspen. Terrell, B.Y., & Ripich, D.N. (1989). Di scourse competence as a variable in intervention. Seminars in Speech and Language, 10, 282-297. Thompson, C.K., Lange, K.L., Schneider, S. L., & Shapiro, P. (1997). Agrammatic and non-brain damaged subjectsÂ’ verb and verb argument structure production. Aphasiology, 11, 473-490. Tompkins, C.A., Boada, R., McGarry, K., J ones, J., Rahn, A.E., & Ranier, S. (1993). Connected speech characteristics of ri ght-hemisphere-damaged adults: A reexamination. Clinical Aphasiology, 21, 113-122. Trupe, E.H., & Hillis, A. (1985). Paucity vs. verbosity: Another analysis of right hemisphere communication defici ts. In R.H. Brookshire (Ed.), Clinical aphasiology conference proceedings (83-96). Minneapolis: BRK. Ulatowska, H.K., Allard, L., & Bond Chap man, S. (1990). Narrative and procedural discourse in aphasia. In Y. Jo anette, & H.H. Brownell (Eds.) Discourse ability and brain damage: Theoretical and empirical perspectives (180-198). New York: Springer-Verlag. Ulatowska, H.K., Freedman-Stern, R., Doye l, A.W., & Macaluso-Haynes, S. (1983). Production of narrative di scourse in aphasia. Brain and Language, 19, 317-334. Ulatowska, H.K., North, A.J., & MacalusoHaynes, S. (1981). Production of narrative and procedural discourse in aphasia. Brain and Language, 13, 345-371. Williams, S.E., Li, E.C., Volpe, A.D., & Ritter man, S.I. (1994). The influence of topic and listener familiarity on aphasic discourse. Journal of Communication Disorders, 27, 207-222. Yorkston, K.M., & Beukelman, D.R. (1980). An an alysis of connected speech Samples of aphasic and normal speakers. Journal of Speech and Hearing Disorders, 45, 2736.

PAGE 55

45 BIOGRAPHICAL SKETCH Christina del Toro is a graduating masterÂ’s student in the University of Florida department of Communication Sciences and Diso rders. During her masterÂ’s program she completed a masterÂ’s thesis on aphasia under th e mentorship of Lori Altmann, Ph.D. Ms. del Toro received her B.A. in Communi cation Sciences and Disorders from the University of Florida in May 2004. In her senior year she completed a senior honors thesis on aphasia with Diane Kendall, Ph.D., which was accepted as a poster presentation at the 14th NIDCD-sponsored Research Symposium. Over her four years of college she was honored with membership into Phi Eta Sigma honor society, Golden Key honor society, Phi Sigma Theta honor society, and Tau Sigma transfer student honor society. She has also been on the DeanÂ’s and Presiden tÂ’s List for her GPA. While attending UF full-time, she has worked as research assist ant at the VA Brain Rehabilitation Research Center in Gainesville, Florida. Her dutie s have included collec ting reliability data, developing screening forms using the Autoda ta software program, and most recently study coordinator for a mild aphasia assessm ent protocol developed by Anna Moore, Ph.D. In August 2006, Ms. del Toro will be gin her doctoral degree in Communication Sciences and Disorders at the University of Florida under the mentorship of Diane Kendall, Ph.D.