Semantic versus phonological aphasia treatments for anomia

MISSING IMAGE

Material Information

Title:
Semantic versus phonological aphasia treatments for anomia a within-subject experimental design
Physical Description:
x, 108 leaves : ill. ; 29 cm.
Language:
English
Creator:
Ennis, Methlee Richardson, 1950-
Publication Date:

Subjects

Subjects / Keywords:
Communication Sciences and Disorders thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Communication Sciences and Disorders -- UF   ( lcsh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1999.
Bibliography:
Includes bibliographical references (leaves 99-107).
Statement of Responsibility:
by Methlee Richardson Ennis.
General Note:
Printout.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 021557076
oclc - 43688657
System ID:
AA00020401:00001

Full Text











SEMANTIC VERSUS PHONOLOGICAL APHASIA TREATMENTS
FOR ANOMIA:
A WITHIN-SUBJECT EXPERIMENTAL DESIGN

















By

METHLEE RICHARDSON ENNIS


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


1999














ACKNOWLEDGMENTS

Many people deserve recognition and my gratitude for their contributions of talent,

time, and energy to make my doctor of philosophy degree a reality, and I would like to

express my appreciation to those persons whose help was especially meaningful.

First and foremost, to my committee chairperson, Dr. Leslie Gonzalez Rothi, I would

like to express my appreciation for her genius in conceptual contributions on this as well

as on other research projects, for her guidance, and for the resources and opportunities

she made possible during my doctoral work. She is the source of inspiration for much

research and many dissertations, including this one. I am happy to call her a friend as

well as mentor. I will continue to enjoy and to learn from her in the future.

To Dr. Anastasia Raymer I would like to express my appreciation for her assistance

in the conceptualization of the complex research design as well as other areas of this

dissertation and for lending her extensive experience in treatment-efficacy research. In

addition, I am eternally grateful for the many, many hours of editing that she contributed,

for her academic expertise that she so generously shares with me to prepare for the future,

and for her support and encouragement during this lengthy project. She sets an example

to which many of us aspire professionally, and her friendship is equally dear to me.

To Dr. Lewis P. Goldstein, whose sense of humor lightened the load, I wish to give

thanks for his perspective as well as his generosity in time and in training during my

graduate program. His suggestions were invaluable, and he is a delight to know.








To Dr. Kenneth Heilman and Dr. Steven Nadeau and fellow researchers from

neurology and neuropsychology, I would like to extend thanks for your stimulating ideas

and contagious enthusiasm for research that I have enjoyed during my years as a graduate

student.

To Dr. Timothy Hackenberg, Dr. Howard Rothman, and Dr. Geralyn Schulz, as well

as Dr. Linda Lombardino, I extend my thanks for their time and contributions to this

research. Each one added a new and beneficial perspective.

To my friend and former fellow graduate student, Dr. Beverly Jacobs, I would like to

give thanks for her unfailing friendship and encouragement with my dissertation and with

all the other pieces of the picture necessary to complete a doctor of philosophy degree.

To Dr. Mitchell Carnell, Jr., long-time mentor and friend, I would like to give thanks

for being the source of it all. Without him, I would never have conceived the idea and left

the city of my dreams for a new one. Twenty years is a long time to inspire someone,

and he is still at it.

To my husband, John B. Ennis, who knows all the reasons why I love him so much, I

would like to thank him for his interest, for being available to me as a partner and friend

and more, and for helping me make all my dreams come true. His contributions are too

numerous to count!

To my parents, William Everette Richardson, Sr., and Ruth Cox Richardson, and all

of my family and friends both near and far who have provided many years of love and

friendship, interest, and support, I thank you all sincerely. You know who you are.

Finally, to the Department of Veterans Affairs, who funded a portion of this

research, and to the veterans and other people who suffered from communication








disorders from strokes and who participated as research subjects, I extend my sincere

appreciation. Ultimately, this research was for all of them.














TABLE OF CONTENTS
Rage

ACKNOWLEDGEMENTS ........................................... ii

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

CHAPTERS

1 INTRODUCTION .............................................. 1

Historical Perspective ............................................ 2
Issues Related to Aphasia Treatment Efficacy ......................... 5
Within-Subject Experimental Design ................................ 8
Anomia Treatment .............................................. 11
A Cognitive Neuropsychological Model of Naming .................... 15
A Study by Howard, Patterson, Franklin, Orchard-Lisle, and Morton....... 20
Statement of the Problem ......................................... 25

2 M ETHODS .................................................... 27

Subject Description and Selection .................................. 27
Inclusion Criteria ............................................ 27
Lesion Localization .......................................... 28
Subjects ................................................... 28
Evaluation of Subjects ........................................... 29
Preliminary Assessment ...................................... 30
Lexical Cognitive Tests ....................................... 32
Experimental Design and Procedures ............................... 36
Experimental Tasks and Stimuli................................ 37
Treatment Protocols ......................................... 43
Semantic Treatment ............................................. 44
Phonological Treatment .......................................... 48
Scoring and Analysis ............................................ 52

3 RESULTS .................................................... 55

Semantic Treatment Research Question 1-a .......................... 55
Semantic Treatment Research Question 1-b ......................... 60
Semantic Treatment Research Question 1-c .......................... 65









page
Phonological Treatment Research Question 2-a ....................... 67
Phonological Treatment Research Question 2-b ........................ 70
Phonological Treatment Research Question 2-c ........................ 75
Comparison Research Question 3-a ................................ 77
Comparison Research Question 3-b ................................ 77
Comparison Research Question 3-c ................................. 78

4 DISCUSSION .................................................. 80

Research Questions .............................................. 81
Research Questions 1-a and 2-a .................................. 81
Research Questions 1-b and 2-b................................. 85
Research Questions 1-c and 2-c .................................. 85
Research Questions 3-a, 3-b, and 3-c.............................. 86
Clinical Implications ............................................. 87
C conclusions .................................................... 88
Future Research ................................................. 89

APPENDICES

A STIMULI FOR SUBJECT 1 ...................................... 91

B STIMULI FOR SUBJECT 2 ...................................... 92

C STIMULI FOR SUBJECT 3 ...................................... 93

D STIMULI FOR SUBJECT 4 ...................................... 94

E SAMPLE OF SEMANTIC TREATMENT CUES ...................... 95

F SAMPLE OF PHONOLOGICAL TREATMENT CUES ................ 96

G ABBREVIATED SEMANTIC TREATMENT PROTOCOL ............. 97

H ABBREVIATED PHONOLOGICAL TREATMENT PROTOCOL ...... 98

REFERENCES ..................................................... 99

BIOGRAPHICAL SKETCH .......................................... 108














LIST OF TABLES


Table pagne

2-1. Subject characteristics ........................................... 30

2-2. Results of the Western Aphasia Battery ............................. 31

2-3. Results of the Boston Naming Test................................. 32

2-4. Percent correct performance by each subject on the
Florida Semantic Battery ....................................... 35

2-5. Results of the Battery of Adult Reading Function...................... 36

2-6. Counterbalanced order of subjects and treatments and number of
baseline sessions ............................................. 39

3-1. Oral-naming performance rating scale......................... ... 55

3-2. Oral-naming performance with semantic treatment.................... 59

3-3. Generalization during semantic treatment (untrained generalization
word set only) ............................................... 61

3-4. Nonword oral-reading/word-repetition control task during semantic
treatm ent .................................................... 63

3-5. Oral-naming performance with phonological treatment................. 68

3-6. Generalization during phonological treatment (untrained generalization
word set only) ............................................... 72

3-7. Nonword oral-reading/word-repetition control task during phonological
treatm ent ................................................... 74

3-8. Comparison of semantic and phonological treatments .................. 78














LIST OF FIGURES


Figure p..age

1-1. Cognitive neuropsychological information-processing model............ 16

1-2. Oral-naming route ............................................. 17

2-1. Results of the Florida Semantics Battery............................ 35

3-1. Composite of performance for oral naming, generalization, and
m maintenance ................................................. 56

3-2. Oral-naming performance with semantic treatment .................... 59

3-3. Generalization during semantic treatment (untrained generalization
word set only) ............................................... 61

3-4. Nonword oral-reading/word-repetition control tasks during
semantic treatment ........................................... 63

3-5. Oral-naming performance with phonological treatment................. 68

3-6. Generalization during phonological treatment (untrained generalization
word set only) ................................................ 72

3-7. Nonword oral-reading/word-repetition control tasks during
phonological treatment ........................................ 74














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

SEMANTIC VERSUS PHONOLOGICAL APHASIA TREATMENTS
FOR ANOMIA:
A WITHIN-SUBJECT EXPERIMENTAL DESIGN

By

Methlee Richardson Ennis

December 1999

Chairperson: Leslie J. Gonzalez Rothi
Major Department: Communication Sciences and Disorders

Word retrieval impairments anomiaa) may relate to dysfunction of either semantic or

phonological stages of lexical processing. Recent clinical work guided by

neuropsychological perspective has led researchers to apply semantic and phonological

treatments for word retrieval impairments to target the presumed stages of (processing)

dysfunction. Some evidence suggests that there is no one-to-one correspondence

between the most effective treatment and the type of word retrieval impairment.

However, few studies have contrasted different treatments in the same patients. We

describe the effects of word retrieval treatments in four aphasic subjects. Each subject

participated in two treatments, which incorporated a within-subject crossover

experimental design with multiple baselines across subjects and behaviors. The two

treatments utilized yes/no questioning that focused on semantic versus phonological








judgments about stimuli in hopes that making these decisions would enable the subject to

more accurately name the stimuli aloud. We predicted that these treatments would be

beneficial to the experimental subjects, that the subjects would differ in the relative

responsiveness to the two treatments, and that the direction of that differential might be

explained by differences in the nature of their lexical deficits. The conclusions of this

study were the following: a) The treatments were effective, and effects were maintained.

b) The treatments were differentially effective, and these differences appeared related to

the nature of each subject's lexical deficits) and spared processes.














CHAPTER 1
INTRODUCTION

Anomia, a failure in word retrieval in picture naming or conversation, is the most

pervasive symptom found in aphasia (Goodglass & Wingfield, 1997). Descriptions of the

treatment of anomia have spanned from ancient to modem times with reports of varying

degrees of success.

While investigations for aphasia treatment in general and for anomia in particular

have been numerous, most investigations report results that are inconclusive or, when

significant, are less than robust. Often treatment research is anecdotal rather than

experimental and, when experimental, has been hampered by limitations of group

experimental designs that obliterate individual differences which may be psychologically

relevant to word-retrieval processes. In addition, many prior experimental reports suffer

from lack of appreciation for a theory of how lexical information is organized.

Some studies have incorporated a phonological approach and others a semantic

approach, but fewer studies have contrasted approaches in the same patient. Howard,

Patterson, Franklin, Orchard-Lisle, and Morton (1985a, 1985b) conducted a series of

group studies that incorporated a theoretical framework of a word retrieval system that

incorporates semantic and phonological lexical processes that are functionally distinct.

They evaluated the benefits of semantic versus phonological cuing hierarchies for the

treatment of anomia and reported an advantage for the semantic treatment. However, this








conclusion may be questioned because of a number of problems with their study,

including issues related to experimental design and patient characteristics.

The present study was developed as a follow-up to Howard, Patterson, Franklin,

Orchard-Lisle, and Morton (1985b) and was based on a cognitive neuropsychological

model of lexical processing that provides a framework to explore the relationship

between the semantic and phonological cuing treatment in patients with anomia. The

experimental procedures incorporated a crossover treatment design with multiple

baselines to counterbalance for treatment effects. Because of the marked heterogeneity

of aphasic patients in their response to treatment, a within-subject experimental design

was utilized to better isolate individual differences in response to the two treatments. In

contrast to the conclusions of Howard and colleagues, our hypothesis was that some

anomic patients would respond better or at least equally well to the phonological

treatment when compared to the semantic treatment.

Historical Perspective

The history of aphasia treatment has been a long one. Earliest accounts of aphasia

date as far back at 1700 B.C. when Egyptian surgical records first mentioned the word

brain and described patients with skull fractures who had a disturbance of speech.

Egyptian physicians treated aphasia with grease from the fat of the gazelle, serpent,

crocodile, and hippopotamus (Howard & Hatfield, 1987). The eleventh-century Arab

Avicenna treated aphasia using a new neuropsychiatric adjunct: "cashew (Anacardium),

recommended for practically all psychiatric and neural affections, especially aphasia"

(Mettler, 1947, p. 538).






3

Since those early Egyptian times the treatment for aphasia has been varied in content

and in methodology. From 30 to 1800 A.D. two prevalent conceptions of the genesis of

aphasia account in part for the different types of therapy which followed. In the Middle

Ages, those who believed that aphasia was caused by paralysis of the tongue commonly

used cauteries and blisters (i.e., application of a hot instrument used to blister and irritate

parts of the body) applied to the neck in the hope of thereby stimulating the organ (e.g.,

tongue) believed to be "sluggish" (Critchley, 1970; Howard & Hatfield, 1987). In

contrast, physicians of that time period such as Pliney the Elder, Johann Schenck von

Grafenberg (Schenckius), and Trousseau observed that the aphasic patient's tongue was

not paralyzed and attributed the speech disturbance to the abolition of the facility of

memory (Howard & Hatfield, 1987).

Another major factor that contributed to the wide variety of treatments for aphasia

arose in part because aphasia remained poorly defined. Medical writers, in their

reference to disordered speech and language, confused a number of conditions which

were actually quite distinct (Critchley, 1970). Because the terminology regarding aphasia

was often used for describing speechlessness from any cause, they failed to distinguish

between aphasia and the following types of disorders: a) dysarthria, b) psychotic

disorders of speech (e.g., occurring in schizophrenia), c) aphonia, d) mutism associated

with hysterical aberrations, d) mental retardation, and f) dementia. Consequently, this

failure to distinguish aphasia from other speech and language disorders resulted in varied

treatments which were sometimes inappropriate and thus ineffectual (Critchley, 1970).

The focus of aphasia treatment changed once again with Johann Gesner's concept of

aphasia (as cited in Benton & Joynt, 1960). Gesner (as cited in Benton, 1965) attributed








language deficits following illness such as stroke to a specific impairment of verbal

memory and pointed out that these language deficits were not part of a general loss of

memory, nor due to paralysis of the tongue, but that the defect reflected a breakdown in

the ability to associate images or abstract ideas with their verbal symbols (Benton &

Joynt, 1960). His view anticipated the basic position of the connectionist neurologists of

the nineteenth century, such as Broca and Wernicke (Howard & Hatfield, 1987).

More systematic attempts at language treatment for aphasia followed the revelations

of Paul Broca's work (Howard & Hatfield, 1987). By the end of the nineteenth century, a

number of well-known physicians and psychiatrists were attempting rehabilitation of

spoken language, principally using direct speech retraining (Howard and Hatfield, 1987).

With the direct speech-training method, language is reconstituted by repetition of sound,

then words, and later phrases.

The early history of aphasia research focused little on the treatment of the language

disorder. Until World War I the majority of aphasia rehabilitation methods were largely

those elaborated for teaching children with retarded or defective speech or with hearing-

impairments (Howard & Hatfield, 1987). In general, there was less interest in the

practical problems of reeducation in aphasia because treatment was difficult to evaluate

in relation to the other factors which might also contribute to or limit improvement

(Weisenburg & McBride, 1964). In the few anecdotal reports and experimental group

studies of aphasia treatment, the positive benefits of treatment were extolled (Franz,

1906, 1924; Froeschels, 1914, 1916; Gutzmann, 1896, 1916; Mills, 1904).






5

Issues Related to Aphasia Treatment Efficacy

The years following Word War I witnessed an increase in aphasia treatment studies

primarily because of the large numbers of patients with penetrating cranial injuries

(Frazier & Ingham, 1920; Weisenburg & McBride, 1935). During the 1950's, the focus

of language treatment expanded from one predominately addressing the communication

problems of head injury patients to one addressing the communication problems of stroke

patients as well. These anecdotal treatments generally reported success. However, by

1970 aphasia treatments by speech-language pathologists which had continued for more

than three decades yielded little experimental data regarding efficacy of treatment. The

relatively small amount of data that did exist were equivocal with some studies showing a

positive change after aphasia treatment (Barton, Maruszewski, &Urrea, 1970; Marshall,

1976), some showing positive change if certain conditions were met (i.e., if treatment

was initiated within the first two months postonset or if treatment lasted at least six

months) (Vignolo, 1964), and with others showing no change after treatment (Sarno,

Silverman, & Sands, 1970).

In Darley's (1972) article "The Efficacy of Language Rehabilitation in Aphasia," he

reviewed the reasons for asking the field of speech-language pathology to address the

issue of aphasia treatment efficacy and challenged aphasiologists to respond to the

question with quantitative data. One of the difficulties in answering Darley's (1972)

efficacy question arose because until this time efficacy studies for aphasia treatment were

conducted with unselected, heterogeneous patient groups (Eisenson, 1949; Franz, 1906;

Frazier & Ingram, 1920; Mills, 1904; Sarno, Silverman, & Sands, 1970; Schuell, Jenkins,

Jimenez-Pabon, Shaw, and Sefer, 1975; Vignolo, 1964; Weisenburg & McBride, 1935,






6

Wepman, 1951). For example, some studies included aphasic patients who were

heterogeneous in terms of lesion location (Goodglass, Kaplan, Weintraub, & Ackerman,

1976; Li & Cantor, 1987; and Pease & Goodglass, 1978) or degree of aphasia severity

(Pease & Goodglass, 1978), or etiology of aphasia (Frazier & Ingram, 1920; Mills, 1904;

Weisenburg & McBride, 1935). Some patients with a diagnosis of apraxia of speech or

dysarthria were inappropriately treated as aphasic subjects (Singer & Low, 1933).

Finally, there was little recognition of the complexity of the language system with all

subjects treated equivalently; that is, studies grouped subjects with different language

deficits and administered them the same treatment (Barton, Maruszewski, & Urrea, 1970;

Weigel-Crump & Koenigsknoecht, 1973).

A second difficulty in responding to Darley's (1972) efficacy question arose

because most aphasia treatment research had been conducted in experimental group

designs (Eisenson, 1949; Franz, 1906, Frazier & Ingram, 1920; Marks, Taylor, & Rusk,

1957; Sands, Samo, and Shankweiler, 1969; Sarno, Silverman, & Sands, 1970; Schuell,

Jenkins, Jimenez-Pabon, Shaw, & Sefer, 1975; Vignolo, 1964). While group designs

have been effective in providing reliable data concerning the characteristics of aphasia

and related disorders, tracking the course of recovery, and constructing psychometrically-

sound measurement instruments, many modem researchers (Holland, 1975; Thompson,

& Keamrns, 1981; Howard & Hatfield, 1987; Kearns, 1991; Shewan, 1986; Hillis, 1991,

1993, 1998; Hillis & Caramazza, 1992; Raymer, Thompson, Jacobs & le Grand, 1993;

Greenwald, Raymer, Richardson, & Rothi, 1995; Jacobs & Thompson, in press; Drew

and Thompson, in press) have found group designs less successful as an experimental

approach to testing the effectiveness of aphasia treatment. Data regarding the efficacy of








aphasia treatment are difficult to obtain from group experimental designs for numerous

reasons. First, as stated previously, the heterogeneity of aphasic deficits is one of the

major problems in the constitution of subject groups for experimental designs (Howard,

1986) because aphasic subjects can differ considerably from one to another in etiology,

lesion location, severity, nature of the language problem, as well as influential variables

such as education, handedness, medical history, and chronicity. Each of these differences

may contribute to differences in responsiveness to a treatment under study, making

grouping of aphasic patients problematic. Second, all the variables which may influence

treatment responsiveness and effectiveness are not yet known and are thus not well

controlled in group designs (Le Dorze, Boulay, Gaudreau, & Brassard, 1994). Third,

although group designs allow for generalization of performance to the population as a

whole (Warren, cited in Chapey, 1986), group results are not always appropriately

generalized to the individual member of the group; some may benefit from the aphasia

treatment, and others may not (Howard & Hatfield, 1987). Fourth, group studies do not

usually provide sufficient information about the treatment which patients received. This

limitation renders replication of the experimental findings virtually impossible (Howard,

1986). Fifth, in group studies the pretest and posttest indicators of improvement have

typically utilized standardized aphasia tests which are not sensitive or relevant to the

language gains an aphasic patient may make over a period of time with a particular

treatment.

After more than 30 published group studies of aphasia treatment, little agreement can

be drawn from conclusions regarding the effectiveness of aphasia treatment (Howard &

Hatfield, 1987) because the group studies were based on the randomized clinical trial








which make methodological and empirical assumptions that are difficult to apply to

aphasia therapy (Howard, 1986). The group design is not designed to ask whether

particular forms of language disorders are amenable to treatment but instead ask whether

intervention in general works with aphasia. That is, group designs are not designed to

ask what forms of treatment are efficacious or to ask when they are efficacious, but

instead whether the activity of language treatment results in a change in language

behavior in aphasic patients in general. Thus, it is no surprise that group studies which

evaluate the effect of aphasia treatment yield inconsistent results and permit little

generalization to individuals with aphasia. A different type of experimental design is

needed to answer the questions regarding whether particular forms of language disorders

are amenable to treatment and what forms of treatment are efficacious (Holland, 1975;

Herson & Barlow, 19767; Kearns, 1991; Kratochwill & Levin, 1992; Franklin, Allison,

& Gorman, 1996; Rothi, 1995; Robey and Schultz, 1998; Robey, 1994).

Within-Subject Experimental Design

Since the 1970's, the within-subject experimental design (i.e., single-subject

experimental design) has become a frequently used research strategy for the investigation

of aphasia treatment (McReynolds and Keamrns, 1983; McReynolds and Thompson, 1986;

Keamrns, 1991). The advancement of clinical aphasiology through within-subject

experimental research is linked to the researcher's ability to relate experimental

questions, outcomes, and rationales to basic behavioral, cognitive, and linguistic theories

(Dietz, 1968; Hayes, Rincover, & Solnick, 1980; Johnston & Pennypacker, 1980; Keamrns,

1991). Perhaps more importantly, the cumulative impact of failing to relate experimental






9

questions to specific theories is the failure to challenge and to refine treatment principles

and to improve the ability to effectively treat aphasic individuals (Kearns, 1991).

Within-subject experimental research designs are experimentally controlled studies

which facilitate the exploration of differences in the response of each subject to treatment

as well as the exploration of similar responses to treatment among different subjects.

This ability to examine individual variations in response to treatment is one of the major

advantages of within-subject experimental designs over group experimental designs. A

common misconception, however, is that any design which uses one subject is a single-

subject experimental design (Shewan, 1986). To the contrary, well-developed within-

subject experimental designs require the investigator to replicate the experiment in a

number of subjects.

Within-subject research designs are experimental rather than descriptive or

correlative in nature (Kearns, 1991). Unlike uncontrolled case studies, single-subject

experimental designs incorporate a number of strategies to maintain experimental control

such as establishing a stable baseline, extending baselines across subjects, and examining

behaviors in withdrawal phases of the experiment.

The need to provide carefully detailed descriptions of patient characteristics, while

important in all treatment research, is heightened in within-subject experimental research

because the process of generalization from the results of such treatment is based on

logical rather than statistical means (Kearns, 1991). Consequently, the process of

inferring the relevance of individual subject data to other patients not involved in the

research effort is critically linked to having an adequate subject description (Keamrns,

1991.)








Researchers who choose within-subject experimental methodologies tend to forego

statistical means of controlling variability (Kearns, 1991). Instead, these designs use

experimental procedures which control or eliminate sources of variability, often without

the assistance of statistical inference testing. These investigators note that treatment

effects that are statistically robust can be unimpressive from a clinical perspective,

particularly when positive and negative outcomes balance one another (e.g., by averaging

extremely high or low scores) (Keamrns, 1991). Instead, investigators using within-subject

experimental designs more commonly use visual analysis of patterns of responses across

time as a primary means of examining within-subject variability (McReynolds & Kearns,

1983). The visual analysis of graphed data to observe changes in the level, slope, and

trend of the effects of treatment can be less objective than better known statistical

analysis procedures (DeProspero & Cohen, 1979). As a consequence investigators using

within-subject experimental designs emphasize the need to develop treatments that are

sufficiently powerful to make improvements in the graphed performance obvious through

visual inspections of the data, thus making the use of statistics superfluous (Kearns,

1991). Michael (1974) argued that treatment-effects which are readily apparent through

graphic analysis would routinely exceed tests of statistical significance.

One area of aphasia research which is inadequately examined is the direct

comparison of aphasia treatments (Keamrns, 1991). Although there have been several

recent treatment-treatment comparisons in the aphasia literature (Kearns & Yedor, 1990;

Loverso, Prescott, Selinger, & Riley, 1989), there remains insufficient data to assist

clinicians in making informed choices about which types of treatment would be most








efficacious for different aphasic patients (Kearns, 1991). One within-subject

experimental design that compares two treatments is the crossover design.

In the crossover design, each subject receives both treatments. The order of the two

treatments is counterbalanced among subjects. During the waiting period between the

two treatments, posttreatment measures of performance are taken to evaluate

maintenance of treatment effects. Baseline measures of performance prior to treatment

can be included to strengthen experimental control. In addition, multiple baselines can be

utilized to examine questions of generalization from treated to untreated stimuli.

Anomia Treatment

Anomia, an impairment of word retrieval, is a pervasive symptom in aphasia, and

much research has focused on understanding variables that influence whether or not an

aphasic person will be able to retrieve an intended word (Williams, 1983). For example,

studies have indicated that lower frequency words are more difficult for aphasic patients

to retrieve than high frequency words (Butterworth, Howard, & McLoughlin, 1984;

Howard, Patterson, Franklin, Morton, & Orchard-Lisle, 1984; Howes, 1964; Rochford &

Williams, 1965; Goodglass, Hyde, & Blumstein, 1969; Wepman, Bock, Jones, & Van

Pelt, 1956; Newcombe, Oldfield & Wingfield, 1965; Kay & Ellis, 1987; Zingeser &

Bemrndt, 1988). Brookshire (1972) also noted that the context of word retrieval might be

important. Aphasic patients were less likely to name words when they were embedded in

a list of difficult words in which they were previously unsuccessful in naming.

Many investigators have found that semantic factors such as imageability (Nickels &

Howard, 1995b), concreteness, prototypicality (Morrison, Ellis, & Quinlan, 1992),

semantic category (Morrison et al., 1992), operativity (i.e., manipulable, discrete,








availability to multiple senses, and firm objects) (Nickels & Howard, 1995a; Gardner,

1973, 1974), and experiential familiarity (Gernsbacher, 1984; Funnell & Sheridan, 1992)

have been shown to influence oral naming in either normal subjects, in patients with

aphasia, or in both groups (Nickels & Howard, 1995a).

Visual variables (Goodglass & Wingfield, 1997) also have been found to influence

the ease with which the external visual stimulus makes contact with the internal visual

representation of the object which affect the accuracy and speed of naming. The visual

variables that influence naming include visual area and visual angle (Snodgrass &

McCullough, 1986; Theios & Amrhein, 1989), visual complexity (Berman, Friedman,

Hamberger, & Snodgrass, 1989), and visual realism (with accuracy decreasing from the

real object to color pictures to black-and-white line drawings, depending upon

educational status) (Reis, Guerreiro, & Castro-Caldas, 1994).

In addition to semantic and visual factors that influence naming, characteristics of

the name itself may influence word retrieval, including the degree of agreement about the

name for the item (i.e., Kleenex versus tissue), word frequency, and familiarity

(Goodglass & Wingfield, 1997). Another variable that has shown relatively consistent

effects on aphasic naming performance is word length (i.e., number of syllables or

phonemes) with aphasic subjects less accurate with longer words (Caplan, 1987; Ellis,

Miller, & Sin, 1983; Goodglass, Kaplan, Weinbtraub, & Ackerman, 1976; Dubois,

Hecaen, Angelergues, Maufras de Chatelier, & Marcie, 1964).

Furthermore, both frequency and familiarity correlate highly with the age at which

children learn words (Feyereisen, van der Borght, & Seron, 1988), and some researchers

(Rochford & Williams, 1965; Hirsh & Ellis, 1994; Nickels & Howard, 1995b) suggest








that word-frequency effects in aphasic naming may in fact be attributed to rated age-of-

acquisition. Nickels & Howard noted that there are a variety of factors that may

determine the age at which children acquire words: names for basic-level objects are

easier to acquire (Rosch, Mervis, Gray, Johnson & Boyes-Braem, 1976); names for parts

of objects are harder to acquire than those for whole discrete objects (Markman &

Wachtel, 1988); young children avoid using names for objects that are difficult to

articulate (Schwartz & Leonard, 1982). These types of influential factors must be

considered in studies of anomia treatment.

Many other studies have reported efforts directed at improving word retrieval

abilities. For example, some earlier studies used general treatment programs which

incorporated a number of treatment techniques that reportedly improved naming in their

aphasic patients (Seron, Deloche, Bastard, Chassin, and Herman, 1979; Wiegel-Crump &

Koenigsknecht, 1973).

More common are studies, which evaluated the effects of various prompts, or cues

that assist aphasic patients to retrieve an intended word (Hillis, 1989; Linebaugh, 1990).

Effective cues have included word repetition, initial syllable prompts (phonemic cues),

open-ended sentences, and written and spelled words ( Li & Canter, 1987; Love & Webb,

1977; Pease & Goodglass, 1978; Podraza & Darley, 1977). However, Patterson, Purell,

and Morton (1983) demonstrated that the beneficial effects of cues such as repetition or

phonemic prompts might be negligible after only a thirty-minute delay.

A major limitation of these early anomia treatment studies is that researchers

(Barton, Maruszewski, & Urrea, 1970) compared different groups of aphasic patients-

categorized according to general classifications of aphasia (e.g., Broca's versus








Wernicke's aphasic patients). As a consequence, the results of these anomia studies have

been conflicting. Individual subjects from different aphasia classifications (e.g., Broca's

versus Wemicke's aphasic patients) as well as individual subjects within the same

aphasia classification (e.g., two Broca's aphasic patients) may respond differently to

treatment. For example, Pease and Goodglass (1978) found that anomic subjects

benefited significantly more from cuing than did Wemrnicke's and Broca's aphasic

patients, while Li and Canter (1987) found that responsiveness to cues was relatively

poor in their anomic aphasia group. Furthermore, Li and Canter (1987) found that their

conduction aphasic subjects tended to respond in a fashion similar to the Wemicke's

aphasic subjects, a finding which contradicted the results of the study by Pease and

Goodglass (1978).

An additional, important limitation of previous anomia treatment studies is that little

attention has been paid to the normal process of word retrieval when choosing cues or

treatment strategies for study. Current models of lexical processing describe a complex,

component system that lends itself to the use of therapeutic approaches in which

processing distinctions are emphasized. The value of these lexical models is the gleaning

of information which may be useful in predicting some of the factors which may

influence the success or failure of an aphasic patient's response to treatment. Although a

number of studies over the past 30-50 years investigated variables affecting anomia and

general strategies to overcome an instance of anomia, it is only recently that aphasic

research has emphasized treatment for anomia that is theoretically-motivated (Kearns,

1991; Lesser, 1989; Rothi, Raymer, Maher, Greenwald, & Morris, 1991; Jacobs &

Thompson, in press; Drew & Thompson, in press; Ellsworth & Raymer, 1998;






15

Greenwald, Raymer, Richardson, & Rothi, 1995; Hillis, 1991, 1993, 1998; Hillis &

Caramazza, 1992; Hillis, Rapp, Romani, & Caramazza, 1990; Kay & Ellis, 1987; Lowell,

Beeson, & Holland, 1995).

A Cognitive Neuropsychological Model of Naming

In recent years researchers have recognized the complexity of the processes involved

in word retrieval (Ellis & Young, 1988). Figure 1-1 displays a modification of a

cognitive neuropsychological information-processing model (Ellis & Young, 1988).

When individuals attempt to orally-name pictures, a task frequently used to assess word-

retrieval abilities in aphasic patients, a complex set of cognitive processes are activated.

These processes are thought to occur in a series of steps in which the individual steps

represent independent, modular processes that activate successive steps in the system in

cascade fashion (Humphreys, Riddoch, & Quinlan, 1988).

A simplified model of oral naming in normal subjects displayed in Figure 1-2 is

thought to involve the following components or psychological processes (Ellis & Young,

1988): 1) visual analysis: The subject sees the picture and identifies basic physical

features of the pictured object, such as size, shape, and contour. 2) object recognition

units: The subject recognizes the pictured object as familiar or unfamiliar (i.e., at a

"sensory specific" representation level which contains structural information about the

object). 3) semantic system: The subject applies meaning to the seen object, including

information such as object function, associated objects (e.g., car and hubcap), coordinate

objects (e.g., car and truck), location, and object category (e.g., food, transportation, or

clothing). Unlike the previous stage of object recognition units, the semantic stage is not

a sensory specific process, and thus does not contain structural information about the










HEARD WORD


SEEN OBJECT


VISUAL OBJECT


VISUAL ANALYSIS


OBJECT RECOGNITION UNITS


WRITTEN WORD


AUDITORY ANALYSIS


AUDITORY


INPUT LEXICON


( SEMANTIC SYSTEM





PHONOLOGICAL OUTPUT LEXICON GRAPHEME-
PHONEME
CONVERSION


PHONEME LEVEL


SPEECH


Figure 1-1: Cognitive Neuropsychological Information-Processing Model (a
modification of Ellis & Young, 1988). Routes for comprehension of yes-no questions,
oral naming of pictures, and oral reading of nonwords.









SEEN OBJECT


VISUAL OBJECT


VISUAL ANALYSIS

OBJECT RECOGNITION UNITS


*
/^SEMANTIC SYSTEMI


/

PHONOLOGICAL OUTPUT LEXICON

/
PHONEME LEVEL


SPEECH


Figure 1-2: Oral-Naming Route (a modification of Ellis & Young, 1988). Route for
naming a picture aloud. The semantic system processes the meaning of a word. The
phonological output lexicon processes the phonological characteristics of a word.








object. 4) phonological output lexicon: The subject then retrieves an abstract

phonological representation of the object name. With this information the individual may

know how many sounds are in the word, what the first sound is, what word rhymes with

it, and how many syllables are in the word. Levelt (1989) has suggested that different

aspects of the lexical phonological representation of a word become available

sequentially. According to Levelt's model, the first stage would yield information

regarding the first sound of the word and the number of syllables. The second stage

would yield information about rhyming words. Levelt seems to think that these stages

are all part of the process of activating lexical phonological representations.

5) phoneme level: The subject retrieves or constructs the individual distinct speech

sounds in the order of occurrence in a word. Then, the subject activates the premotor and

motor processes necessary to articulate the word.

While the process of picture naming involves these multiple stages, two of the

mechanisms, the semantic system and the phonological output lexicon, are critical in the

process of word retrieval across input modalities. Presemantic visual processes are

specific to picture-naming tasks and presumably are not involved in word retrieval in

tasks such as conversational discourse. The phoneme level of the system involves

processes for pronunciation of retrieved lexical items after the processing for the earlier,

more critical lexical retrieval processes in the semantic system and phonological output

lexicon are completed and consequently will not be investigated in this study.

Because at least two neuropsychological systems (i.e., semantic system and

phonological output lexicon) are critical in the process of word retrieval, anomia

treatment approaches may need to focus specifically on activation of semantic processes








and phonological processes. Semantic treatments would emphasize the processing of the

meaning associated with the word. Phonological treatments would emphasize the

processing of information about the phonological (or spoken) form of the word.

Currently, the development of anomia treatments from a cognitive

neuropsychological perspective is still in its early stages. However, some studies using

phonological and semantic treatments have been reported. Hillis (1991) studied oral-

reading by using (written) phonetic spellings (e.g., cat as /kat/) as a phonological

treatment in a patient with an impairment of the phonological output system and found

improved oral-reading with generalization of phonemic accuracy to the untrained task of

oral naming. Raymer, Thompson, Jacobs, and le Grand (1993) also studied the

effectiveness of a phonological treatment when they applied a phonological cuing

hierarchy in patients with deficits affecting the phonological output lexicon and found

improvement in oral naming as well as some generalization to naming of untrained

stimuli in some subjects. The phonological treatment by Raymer et al. (1993) included

the following: a) rhyming cue, b) initial phoneme cue, c) auditory model for repetition,

d) rehearsal of the target word, and e) a non-cued spontaneous attempt to name the target

word.

In contrast to these phonological treatments, Hillis-Trupe (1991) studied the

effectiveness of a semantic treatment to improve written picture naming performance.

The patient contrasted semantic features by learning distinctions between related items

(e.g., apple and cherry). Hillis found specific effects of semantic remediation strategies

at the level of the semantic system for generalization to oral naming. Ochipa, Maher, and

Raymer (1998) also studied the effectiveness of a semantic treatment in a patient with a








semantic deficit when they contrasted a semantic coordinate picture (e.g., lion) with a

target picture (e.g., tiger) and found the semantic treatment to be effective.

A Study by Howard. Patterson. Franklin. Orchard-Lisle, and Morton

Only one study (Howard, Patterson, Franklin, Orchard-Lisle, and Morton, 1985b)

directly compared the effectiveness of semantic versus phonological treatments in the

same patients. For the semantic treatment, Howard et al. (1985b) utilized the following

three tasks which they considered semantic in nature: a) matching spoken words to

pictures, b) matching written words to pictures, and c) making semantic judgments (e.g.,

Is a cat an animal?). Visual cues (e.g., pictures and written words) were used with two of

the three semantic tasks. Both the auditory and written-word input routes were

stimulated during the semantic tasks. In the semantic treatment, performance in a

confrontation oral-naming task of the picture stimuli after the treatment was significantly

more successful than in pre-treatment measures. In addition, Howard et al. (1985b)

found this "semantic treatment" effectiveness to be enduring when measured up to 24

hours posttreatment.

In the phonological treatment, Howard, Patterson, Franklin, Orchard-Lisle, and

Morton (1985b) used the following three tasks which they considered phonological in

nature: a) initial phoneme cue, b) repetition, and c) rhyme judgment. With the exception

of the repetition task, only the auditory input route was stimulated in the phonological

tasks. These investigators found that the phonological treatment was also somewhat

effective in improving performance on confrontation oral naming immediately after the

treatment but that the effects did not endure after a matter of minutes. This finding

essentially corroborates the findings of Patterson, Purell, and Morton (1983) that






21

phonological treatment results in a brief beneficial effect on oral-naming performance but

that the effects rapidly dissipate. In a comparison of the phonological and semantic

treatments, Howard et al. (1985b) found an advantage for the semantic treatment.

Although Howard, Patterson, Franklin, Orchard-Lisle, and Morton (1985b) have

strengthened their group research by incorporating some features of within-subject

experimental designs, other features could be incorporated and thereby possibly shed

further light on the differences in the effect of semantic versus phonological aphasia

treatments for different types of lexical word-retrieval deficits. First, in the group design

all of the data for individual subjects was averaged across the group for each session.

In addition Howard, Patterson, Franklin, Orchard-Lisle, and Morton (1985b)

contrasted the effects of these two oral-naming treatment techniques in a crossover

treatment, a within-subject experimental design, a design which lends itself to the effects

of the first treatment influencing the second treatment. Additionally, because

improvement was seen in the untreated naming control-stimuli, as well as in the treated

stimuli from both treatments, Howard et al. lost experimental control. With the loss of

experimental control, there is no clear evidence that the improvement resulted from the

treatment alone or from extraneous events in the environment or both, making

conclusions about the relevance of change (or lack thereof) on these measures invalid.

Because of the inherent inconsistencies in oral-naming in anomic, aphasic subjects

and because of factors in the environment that may influence naming, many research

designs require that researchers demonstrate a stable baseline measure in within-subject

experimental designs before a treatment study begins. In its basic form the crossover

treatment design used by Howard, Patterson, Franklin, Orchard-Lisle, and Morton








(1985b) does not require that a baseline measurement be taken prior to the initiation of

treatment. They designed their experiment such that the naming-control items would

serve to demonstrate experimental control. If they had demonstrated improvement only

in their treated stimuli, without the naming-control stimuli improving also, then Howard

et al. would have maintained experimental control with their design. However, if there is

a possibility that the effects of treatment will generalize to untreated stimuli, then some

other measuress, such as a stable baseline prior to treatment and a decline in

performance once treatment is withdrawn, is required to demonstrate experimental

control. If Howard et al. (1985) had demonstrated a stable, extended baseline in their

study, then they would have been able to better conclude that the improvement in treated

and untreated stimuli were likely due to the effects of treatment and not due to

uncontrolled, random events in the environment. Unfortunately, Howard et al. pretested

oral naming for pictured objects only twice prior to treatment. No information was given

regarding the stability of each patient's performance on these pretreatment measures.

Third, Howard, Patterson, Franklin, Orchard-Lisle, and Morton (1985b) reduced

some of the problems of group experimental designs by utilizing a design in which the

different treatments are applied and compared within the same subject. Each subject

participated in both the semantic and phonological treatments with half of the subjects

beginning with the semantic treatment and half beginning with the phonological

treatment. Unfortunately, Howard at al. grouped the data from all subjects by treatment.

Thus, the answer to other important questions cannot be examined: a) What is the effect

of the first treatment upon the second treatment? For example, does the effect of the first

treatment (semantic treatment) boost the impact of the second treatment (phonological








treatment) and vice versa? b) Does the effect of one treatment applied first over the other

treatment hold consistent across all subjects? c) If there is a difference in which

treatment is applied first, does the difference relate to differential impairments in the

neuropsychological processes of naming among subjects? More information may have

been obtained if the results of each treatment had been visually displayed in a separate

graph for each subject and if the reader had more information regarding the

neuropsychological bases for each subject's word-retrieval difficulty or difficulties.

Fourth, little is known about the pictured stimuli used by Howard, Patterson,

Franklin, Orchard-Lisle, and Morton (1985b) except that they were selected according to

a single criterion-the name of the pictured stimuli must be able to rhyme with another

name. However, other factors concerning the stimuli which may influence the subjects'

responses to treatment were not controlled. Better experimental control would have been

attained if the researchers had matched the pictured stimuli by some of the following

variables: a) age-of-acquisition, b) word length, c) familiarity, and d) word frequency.

Fifth, in most instances (five out of six tasks) Howard, Patterson, Franklin, Orchard-

Lisle, and Morton (1985b) did not target or link oral naming as part of the treatment

process in a study in which oral naming of pictured objects is the dependent variable.

Instead, the treatment targeted other forms of processes (i.e., semantic or phonological).

However, in one treatment (phonological) the subjects did have a direct opportunity to

orally name the picture in the repetition task. In addition, upon failure to orally name the

pictured stimuli, much of the treatment utilized by these researchers was not designed

specifically to be a treatment strategy to assist the subject in oral naming of the pictured

object the next time he attempted the task. In other words, perhaps the treatments would






24

have been more informative if they were specifically designed to take the patient through

the steps used by nonbrain-damaged individuals to retrieve a word (Levelt, 1989).

Sixth, although their treatment period was longer than in many studies, the length of

treatment to evaluate long-term change in oral-naming ability by chronic aphasic patients

was still very short, ranging from four to eight sessions only. By six weeks'

posttreatment, no significant contrasts between the results of the two treatments existed.

While information from this short length of treatment might lend support to a small

treatment trend, it is possible that different information might be obtained with a more

lengthy treatment period.

Seventh, demographic and neuropsychological data on each of the subjects was

sparse; thus the ability to generalize the effects of the treatments to other aphasic patients

is limited. Demographic information for each subject matched to the subject's response

to treatment would be beneficial in selecting other patients who might benefit from these

treatments as well as explaining some of the variation in response to treatment. Results

of these data for all 12 subjects with their distinctive and varied individual lexical deficits

were lumped together. A description of the varied lexical deficits for each subject based

on the neuropsychological information-processing model for single words would have

been more informative. Valuable information was potentially lost with the group data;

specifically, no information was gained regarding the effectiveness of the two treatments

relative to deficits explained by the information-processing model. For example, a

subject with a primary deficit in the phonological output lexicon may have benefited

more or to an equal degree from the phonological treatment than from the semantic

treatment. It is also unclear from these group data whether some individual subjects may






25

have retained the effects of treatment longer from the phonological treatment than from

the semantic treatment.

Statement of the Problem

A review of the literature reveals a long and varied history in the study of the

treatment of aphasia. Anomia itself is a complex phenomenon, and the results of

treatment have been diverse, possibly related to the problems inherent in the use of group

experimental designs to study treatment efficacy and the intricate nature of the language

system. Recently, aphasia treatment research has begun to incorporate advances in

within-subject experimental designs. In addition, with the relatively recent advancement

of cognitively neuropsychological models which are designed to represent the complexity

of the language system within normal persons in general and lexical retrieval in

particular, two levels of processing have been highlighted. In the study of Howard,

Patterson, Franklin, Orchard-Lisle, and Morton (1985b), effectiveness of anomic

treatments targeting these two distinct levels of processing-semantic and

phonological-were compared in a primarily group experimental design. The focus of

the current study is to again compare the relative effectiveness of treatment methods

targeting these two theoretically motivated levels of processing of words in anomic

patients. This study used a within-subject experimental design to provide experimental

control sufficient to allow adequate analysis of treatment effects.

To accomplish this endeavor, the present research study differs in a number of ways

from the study by Howard, Patterson, Franklin, Orchard-Lisle, and Morton (1985b).

Individual variations within subjects and between subjects were investigated. A single-

subject experimental design was utilized to evaluate the responses of individual aphasic






26

subjects with an anomic deficit. Semantic and phonological treatment interventions were

applied. The purpose of the study was to compare the relative effectiveness of the

semantic and phonological treatments for oral-naming deficits associated with aphasia.

The experimental questions were as follows: 1-a) Will aphasic subjects demonstrate

a significant improvement in oral-naming performance as the result of a semantic

treatment? 1-b) If the semantic treatment is effective, do the changes in oral-naming

performance also generalize to untreated stimuli? 1-c) Is the change in oral-naming

performance enduring at two weeks posttreatment? 2-a) Will aphasic subjects

demonstrate a significant improvement in oral-naming performance as the result of a

phonological treatment? 2-b) If the phonological treatment is effective, do the changes in

oral-naming performance also generalize to untreated stimuli? 2-c) Is the change in oral-

naming performance enduring at two weeks posttreatment? How do the semantic and

phonological treatments compare on the following: 3-a) degree of change, 3-b)

generalization, and 3-c) maintenance?














CHAPTER 2
METHODS


The purpose of this study was a comparison of the efficacy of two treatments-

semantic and phonological-for anomia

Subject Description and Selection

Subjects were individuals who were six months or greater post onset of a left

hemisphere cerebrovascular accident (CVA) which resulted in aphasia and word retrieval

impairments anomiaa). The experiment required four subjects to allow for replication

and to control for order effects. Subjects were recruited from an outpatient VA clinic and

community stroke support group in Gainesville, Florida and an (outpatient) academic

speech and hearing clinic in Norfolk, Virginia. All subjects appeared to have normal

cognition for their age (by informal report) and were living at home with family or with a

caregiver. While undergoing the experimental treatments, subjects did not participate in

other forms of speech-language therapy. Table 2-1 displays the subject characteristics.

Inclusion Criteria

The subjects were right handed, monolingual English speakers with at least a sixth

grade education who were greater than 18 years of age. Subjects were excluded if they

had a history of the following: a) other neurological illnesses (e.g., Alzheimer's disease,

Parkinson's disease, etc.) b) chronic medical illnesses (e.g., cancer, renal failure, etc.), c)

inability to repeat single words, d) comprehension inadequate to understand and perform






28

experimental tasks as indicated by a score of<30 on the yes/no question subtest of the

Western Aphasia Battery (WAB) (Kertesz, 1982), and e) severe sensory deficits (e.g.,

vision or hearing).

Lesion Localization

CT scans were examined to document a unilateral left hemisphere lesion and to

document a right hemisphere that appeared within normal limits. For two subjects the

lesion was plotted for locus and size using the techniques of Damasio and Damasio

(1989) to identify major regions of abnormality.


Subjects

Subject 1 (AW)

Results of AW's CT scan, performed three weeks after onset, revealed a large

frontoparietal lesion. His CVA resulted in aphasia, right-sided hemiplegia, and neglect of

right space. He also demonstrated a short-term auditory-verbal memory deficit.

Subject 1 was a retired railroad engineer.

Subject 2 (AS)

Results of AS's CT scan, performed one day and another performed four years after

onset, revealed a left posterior parietal lesion. Her CVA resulted in aphasia, mild right-

sided hemiparesis, and apraxia of speech. Severe limb apraxia was also observed.

Subject 2 was employed as an executive secretary at the time of her stroke.

Subject 3 (PB)

A CT, performed immediately after onset, was negative. A second CT scan,

performed nine months postonset due to seizures (but prior to this research study),

revealed an old left frontal lesion involving gray matter (area 47) and white matter








immediately undercutting Broca's area and surrounding regions (Brodmann's areas 44,

45, and 6). The insula and underlying white matter were also affected. PB's CVA

resulted in global aphasia, which later evolved into a fluent neologistic aphasia (i.e.,

typically associated with a posterior lesion). Her aphasia classification of Wemrnicke's

aphasia was anomalous for a frontal lesion. She initially demonstrated a hemiparesis of

the right side of her face and arm which resolved by the time of this experiment. She was

fully right handed and all her siblings and children were right handed. Subject 3 was a

housewife.

Subject 4 (FS)

A CT scan was not obtained at onset because FS did not seek medical attention for

his CVA. Seven months postonset (but prior to this research study) a CT scan and

MRI/MRA were performed when the patient reported a worsening of his speech, vision,

and hand weakness. Results of the CT scan revealed a left MCA distribution infarct and

a left occipitoparietal watershed-distribution infarct. Several regions in the periphery of

the left MCA infarct may represent more acute infarction. His CVA resulted in aphasia

and right upper extremity hemiparesis (e.g., unable to grip a pencil). Subject 4 was

disabled and unemployed at the time of this study but had been previously employed in

several trades, such as fisherman, farmer, and construction worker.

Evaluation of Subjects

Two standardized preadmission tests, the Western Aphasia Battery (WAB) and the

Boston Naming Test (BNT) (Kaplan, Goodglass, & Weintraub, 1983), were administered

to identify the presence of aphasia and word retrieval difficulties anomiaa).








Table 2-1: Subject Characteristics

Subjects Sex Age Educ Hand Time Lesion site Date
Postonset Postonset of
___ ____of CVA _____ CT Scan
51 (AW) M 59 9"h grade R 6 mos. Frontoparietal 3 weeks
S2 (AS) F 71 HS R 4 years L parietal day&
S 1 4 years
53 (PB) F 76 HS R 10 mos. L frontal w/ 9 months
anomalous
fluent
neologistic
aphasia
S4 (FS) M 48 GED R 1 year L temporal & L 7 months
occipitoparietal
vs frontal


Subjects with impaired word retrieval (<48 correct/60 possible on the BNT) participated

in the study.

Subjects also completed lexical-semantic tasks to characterize the pattern of word

retrieval deficits with the following tests: a) the Florida Semantics Battery (Raymer,

Maher, Greenwald, Morris, Rothi, & Heilman, 1990), b) the Semantics Associate Test

(Raymer, Greenwald, Richardson, Rothi, & Heilman, 1992) and c) selected subtests from

the Battery of Adult Reading Function (BARF) (Rothi, Coslett, and Heilman, 1984).


Preliminary Assessment

Western Aphasia Battery

The first preadmission measure was the Western Aphasia Battery, an aphasia

examination with subtests that evaluate four language functions: spontaneous speech,

auditory comprehension, repetition, and oral naming. Scores for each subtest has been

reported on a 10-point and 20-point scale. Results of the WAB shown in Table 2-2

revealed that each of the subjects demonstrated aphasia. Subject 1 (AW) demonstrated






31

symptoms primarily associated with Broca's (i.e., nonfluent with poor word generativity

and impaired repetition). Subject 2 (AS) demonstrated conduction aphasia. Subject 3

(PB), who initially presented with a global aphasia (i.e., impaired auditory

comprehension for simple commands and sparse nonfluent speech), evolved to a

Wemicke's aphasia (grammatically correct, fluent speech with neologisms) by 10 months

postonset. d) Subject 4 (FS) demonstrated nonoptic anomic aphasia, characterized by

relatively spared auditory comprehension and repetition; however, he demonstrated poor

lexical retrieval and poor word generativity in spontaneous speech but relatively

preserved visual confrontation naming on standardized tests.


Table 2-2: Results of the Western Aphasia Battery

Subtests ___ Subjects
S1 (AW) S2 (AS) S3 (PB) S4 (FS)

Spontaneous +12/20 +13/20 +14/20 +13/20
Speech possible possible possible possible

Auditory +8.6/10 +6.6/10 +6.1/10 +8.7/10
Comprehension possible possible possible possible

Repetition +5/10 +1.8/10 +6.8/10 +9.2/10
possible possible possible possible

Oral Naming + 4.7/10* +7.2/20 +3.7/10* +17/20
possible possible possible possible

Aphasia Quotient +60.6/100 +49/100 +61.2/100 +82/100
__________ possible possible possible possible
Key: 20 points possible for Spontaneous Speech & Oral Naming,
10 points possible for Auditory Comprehension & Repetition, &
100 points possible for Aphasia Quotient
* Scores from Sentence Completion Subtest not available.








Boston Naming Test

The second preadmission measure, the BNT, evaluated the subjects for anomia and

types of paraphasic errors (semantic or phonological). The results of the BNT shown in

Table 2-3 revealed that all four subjects demonstrated anomia associated with aphasia.


Table 2-3: Results of the Boston Naming Test

Subjects
~~~____~~_____Sl (AW) S2 (AS) S3 (PB) S4 (FS)
Number Correct +8/60 +4/60 +4/60 +43/60
possible possible possible possible

Number Correct w/Phonemic Cue unavailable +7/56 +7/56 +8/17
_____________________possible possible possible


Lexical Cognitive Tests

Florida Semantic Battery (FSB)

A detailed analysis of patterns of performance across lexical tasks was conducted

with the FSB, which consists of 120 items in each of six subtests (i.e., 720 total items).

The six subtests assess skills in the following areas: a) oral picture naming, b) written

picture naming, c) oral naming to definition, d) oral word reading, e) write to dictation,

and f) auditory word/ and written words/picture matching (i.e., matching pictures to

spoken and written words). The analysis from the FSB was utilized to identify the loci

of lexical impairments in the cognitive neuropsychological model for each subject. The

results of performances on the FSB are shown in Table 2-4 and Figure 2-1 by percent

correct. Patterns of word retrieval deficits were analyzed and compared to the

neuropsychological model to determine whether the subjects had primarily a semantic

dysfunction and/or a phonological dysfunction. Presence of a "central" semantic








dysfunction was determined by poor performance on all six subtests of the FSB while

presence of a more "peripheral" dysfunction was designated when a relatively spared

performance on one or more of the three comprehension tasks (i.e., auditory word/picture

match, written word/picture match, and oral naming to definition) in the context of an

impaired performance on any input or output modality (i.e., oral picture naming and

written picture naming) was noted. These more "peripheral" types of deficits may

involve an impaired output mode such as the phonological output lexicon in oral naming

or may involve an input modality such as visual identification of pictures or written

words. In reviewing each subject's individual performances on the FSB, two of the

subjects (Subjects 2 and 4) demonstrated a similar pattern of deficits: The primary failure

explaining their oral word-retrieval deficit would be at a very late level of processing

involving the phonological output lexicon specifically, while demonstrating a relatively

intact semantic system. Sparing of their semantic system was indicated by relatively

strong performance on both auditory and written comprehension matching tasks as well

as relatively strong performance on matching semantically related pictures. Specific

failure of the phonological output lexicon was indicated by their difficulty in making

spoken word forms available for retrieval (oral lexical word-retrieval), such as oral-

picture naming. The phonological output deficit was noted to be worse in Subject 2

compared to Subject 4.

Subject 1 demonstrated a similar pattern to Subjects 2 and 4 in that his oral

naming failure also appeared to result from a deficit in the phonological output lexicon

(i.e., as demonstrated by deficient performances of the oral picture naming and oral

naming to definition subtests) with relative sparing of the semantic system as indicated








by his relatively strong performance on the comprehension task for matching spoken

words to pictures. However, Subject 1 displayed an additional difficulty specific to the

visual channel as indicated by his problems with tasks specifically involving visual

analysis (i.e., impaired performance on written word/picture matching tasks, a written

picture naming task, and a semantic-associate picture-matching task) while

relatively sparing performance on tasks involving less vision (i.e., auditory word/picture

matching task). Although Subject 1 scored the lowest of the four subjects on the

semantic associate picture-matching task, his relatively strong performance in the

auditory word/picture matching task indicated that his semantic system was still

operational as measured through this one input modality.

In contrast, Subject 3 was different from all other subjects in that her pattern of

performance (i.e., failure on all FSB subtests) was consistent with a "central" semantic

deficit. Whether or not her "peripheral" lexical level systems were impaired is unknown

as the semantic failure inhibited assessment of all other systems. However, her reading

aloud on the BARF (as discussed in the next section) would indicate relative sparing of

some lexical level processes.

Battery of Adult Reading Function

Portions of the BARF, an adult reading test, were administered to assess the

integrity of the lexical and nonlexical reading routes for each subject. As shown in Table

2-5, each subject completed the following: a) oral reading of regularly spelled words, b)

oral reading of words with exceptional spellings, and c) oral reading of phonologically

plausible nonwords (i.e., nonwords). Impairments in nonword reading suggest

impairment in sublexical reading processes. Impairment in reading exceptional words









Table 2-4: Percent Correct Performance by Each Subject on the Florida Semantics
Battery

Subtests _______Subjects
S1 (AW) S2 (AS) S3 (PB) S4 (FS)
Oral Picture 43% 19% 14% 64%
Naming**
Written Picture 0%* 0%* 10% 32%
Naming**
Oral Naming to 35% 17%* 13% 65%
Definition
Auditory Word/ 90% 96% 47% 95%
Picture Match
Written Word/ 20% 94% 47% 93%
Picture Match
Semantic 57% 91% 69% 84%
Associate Picture
Match_______________
Average of Percent 40.8% 52.8% 33.3% 72.2%
Correct
* Test not completed due to fatigue of the subject.


Florida Semantic Battery


Average %
Across Subtests


Performance Across Individual Subte


100
? 80
5 60
40
S20
0.


Oral Px Written Px Oral Aud Wd Px Written Wd Semantic Aver %
Naming Naming Naming to Match Px Match Asso Px Correct
Definition Match
Subjects


-0-S4(FS) I


Figure 2-1: Results of the Florida Semantic Battery.


I -4-S1 (AW) -l--S2 (AS) -A- S3 (PB)






36

(i.e., irregular) suggests impairment in lexical reading processes. Impairment in all

subtests suggests impairment in both sublexical and lexical reading processes. Results of

testing revealed two subjects who were able to use lexical reading processes-Subject 3

(PB) and Subject 4 (FS). One subject was able to partially use lexical reading processes

for regular words (Subject 2-AS), and one subject was unable to use lexical reading

processes at all (Subject 1-AW).

An assessment of the nonlexical (i.e., nonword) reading route was necessary to

determine if a nonword reading task could be used as the control task. A failure in

nonword oral reading would indicate that a task utilizing this nonlexical reading route

would be appropriate as a control task. Results of the reading tests indicated that all four

subjects were significantly impaired in their ability to use sublexical reading processes

(i.e., orally read nonwords). Thus, this task was chosen to be used as an experimental

control for each subject except Subject 1.


Table 2-5: Results of the Battery of Adult Reading Function


* Test not administered due to degree of difficulty for subject


Experimental Design and Procedures

This single-subject experimental design incorporated features of both a crossover

design and multiple baselines across subjects and behaviors (McReynolds & Keamrns,

1983). With one exception the treatments took place in the subject's home in a quiet


Subtests Subjects
Sl (AW) S2 (AS) S3 (PB) S4 (FS)
Nonwords 0* +0/30 possible +3/30 possible +1/30 possible
Regular Words 0* +8/30 possible +23/30 possible +23/30 possible
Rule-governed 0* +0/30 possible +22/30 possible +20/30 possible
Words_________






37

room. Subject 4 (FS) was seen in two locations throughout both treatments-the

outpatient VA speech clinic and at his family's business office. The independent

variables were the two treatments, and the dependent variable was the accuracy of oral

picture naming.

Experimental Tasks and Stimuli

During the experiments, the performance in probe tasks was assessed to examine

immediate treatment effects and to maintain control in the single-subject experimental

design. The treatment task was oral picture naming, and the control task was oral reading

for all but one subject.

Pretreatment tasks and experimental stimuli

To establish a set of treatment stimuli, each subject completed an experimental

picture-naming task for nouns that was previously standardized on five normal adult

subjects. Experimental subjects named 260 black and white line drawings (some of

which were taken from Snodgrass and Vanderwart, 1980, and the BNT) of objects coded

for age-of-acquisition (Gilhooly & Logie, 1980; Carroll & White, 1973), word frequency

(Francis & Kucera, 1982), and syllable-length, factors that influence performance in

picture naming (Nickels and Howard, 1995). Pictures that the subjects were unable to

name on at least two of three occasions during baseline testing were placed into a pool of

potential stimuli for the treatment experiments. Subjects had to have at least 60 potential

picture stimuli to be included in the treatment experiments.

Training probe stimuli

From the pool of potential noun stimuli, three sets of 20 pictures (60 pictures total)

were selected for use as training stimuli and untrained generalization stimuli. The pool of








potential experimental pictures varied to some degree among subjects. (Appendices A-D

displays the 60 stimuli and 20 nonword stimuli for Subjects 1-4.) Therefore, the sets were

matched on psycholinguistic variables that can affect naming performance so that there

was greater assurance that treatment differences observed among subjects were likely to

be due to treatment effects and not to differences in picture stimuli. The selection of the

60 picture stimuli from the pool of 260 was based on the following criteria with the first

item in the series given the most importance and the last item given the least importance:

a) the number of times the subject failed to orally name the picture in baseline trials, b)

age-of-acquisition (AOA), c) syllable length, and d) ecological validity (i.e., stimuli

thought likely to be useful in the subject's everyday conversation). When possible, the

60 stimuli consisted of the pictures for which the subject demonstrated the most difficulty

in oral naming to contribute to a more stable baseline prior to treatment. The sets were

matched as nearly as possible on all four variables for three of the subjects. The sets

were matched for the number of times the subject failed to orally name the picture in

baseline testing and for ecological validity for Subject 1 (AW).

Control probe task

It was possible that each subject could improve in naming untrained pictures during

training. Therefore, a control measure was incorporated that was anticipated, based upon

knowledge of the lexical system, that would not improve in conjunction with treatment.

This task was similar in nature to the expressive oral naming task, but one that should not

be affected by the word retrieval treatments because reading relies on other

orthographic/phonologic mechanisms for accurate performance. This control task,








therefore, should only change in response to external factors such as repeated exposures

or spontaneous recovery.

Subjects 2 (AS), 3 (PB), and 4 (FS) used nonword oral reading as a control task.

Subjects read aloud from a list of 60 nonwords (Raymer & Bemrndt, 1996). Items that a

subject incorrectly read aloud during baseline testing were identified for a pool of

potential control stimuli. Twenty nonword stimuli were selected for the control task.

However, Subject 1 found oral reading extremely difficult and was reluctant to complete

the task. Therefore, word repetition was added as an appropriate control task for Subject

1 (AW) (Ellsworth and Raymer, 1998). Appendixes A-D display the control stimuli for

Subjects 1-4.

Counterbalancing order of treatments

As depicted in Table 2-6, four subjects received the two treatments in

counterbalanced order to control for order effects among the subjects. Two subjects

received Treatment 1 first, and two subjects received Treatment 2 first.


Table 2-6: Counterbalanced Order of Subjects and Treatments and
Number of Baseline Sessions

Subjects Treatment 1 Treatment 2 # of Baseline Sessions
1 (AW) Phonological Semantic 3
2 (AS) Semantic Phonological 4
3 (PB) Semantic Phonological 5
4 (FS) Phonological -Semantic 6


Pretreatment phase

During the pretreatment baseline phase, the two experimental probe tasks, oral

picture naming and control reading/repetition task, were observed for at least three

consecutive sessions. As a means of additional experimental control for repeated






40

exposure to stimuli, baselines were systematically extended up to six baseline

observations across the subjects who entered the study. See Table 2-6. If repeated

exposure (and not the effect of treatment) accounts for the changes observed,

improvements in individuals during prolonged baseline phases should be observed.

The purpose of the pre-treatment baseline measures was to establish a stable baseline

performance (little change in performance across sessions) on experimental tasks prior to

the initiation of treatment. A stable baseline was established for all four subjects by

visual inspection of the data (McReynolds & Keamrns, 1983). If a stable baseline could

not be established within 10 sessions, then the subject was not eligible for the study.

Treatment phase

The two experimental probe tasks (oral picture naming and nonword oral

reading/word repetition) were divided into four sets whose behaviors were observed

across experimental phases. The four sets consisted of 80 stimuli-60 for picture naming

and 20 for nonword reading/repetition. Four behaviors were measured across phases

(i.e., pretreatment, treatment, and maintenance) for each subject as follows: a) trained

picture naming set, b) picture naming set held in baseline and later trained in Treatment

2, or picture naming set previously trained in Treatment 1, c) untrained generalization

picture naming set, and d) nonword reading or word-repetition control task. Three of

these behaviors were oral naming behaviors, and a fourth behavior was an oral reading or

repetition measure. Treatment was provided for only two of the four behaviors and

performance was monitored in the other two behaviors (i.e., untrained generalization

picture-naming and untrained nonword oral-reading/word repetition).






41

During treatment, each session was initiated with probe tasks to assess acquisition

for the trained picture naming set(s), generalization of improvement to the untrained

picture naming set(s), and the level of performance in the control reading/repetition task.

The order of stimuli and task presentation was randomized across sessions.

To prevent the daily treatment probe measures on the 80 stimuli from becoming too

laborious, time-consuming, and frustrating for the subject prior to the treatment period for

each session, the stimuli which were not currently being treated (a total of 60 stimuli)

were divided into two sets for presentation in alternating sessions. Thus, half of the 60

untreated stimuli (i.e., oral naming of pictures and oral reading of nonwords) in addition

to the entire 20 trained picture stimuli were probed prior to each treatment session with

two exceptions: Subject 1 (AW) and Subject 3 (PB) were probed on half of the trained

picture stimuli in each probe session for a total of five observations across the 10

treatment sessions.

While the subject was in Treatment 1 phase, the 60 untreated stimuli measures

continued to provide baseline information on untrained picture stimuli and untrained

nonwords oral reading/repetition stimuli. During the Treatment 2 phase, the measures on

the 60 stimuli not currently being treated provided maintenance information on the 20

previously treated stimuli from Treatment 1 as well as continued baseline information for

untrained generalization picture-naming stimuli and untrained oral nonword reading/word

repetition.

During the treatment phase, the treatments were administered consecutively (i.e., a

crossover design), not simultaneously (i.e., alternating treatment design). After the

presession test probes were completed, the designated set of 20 pictures was trained for






42

oral naming. These 20 pictures were trained once or twice each session as time permitted

with a random order of presentation. Treatment sessions ranged from 60 to 90 minutes in

length. All treatment sessions were conducted in a quiet setting and 30% of sessions

were either videotaped or audiotaped for reliability scoring. Treatments for each subject

followed a schedule of two to three sessions per week, although occasional interruptions

to this schedule occurred (i.e., subject was hospitalized, holidays, etc.). Appendices E-F

display a sample of the question-cues for the semantic and phonological treatment.

Each subject received 10 treatment sessions for each of the two treatments for a total

of 20 treatment sessions. Rather than extending the training to criterion of 80-90% of

stimuli as is done in some single-treatment aphasia studies, an equal number of sessions

for each of the two treatments was used, which provided an equivalent comparison for

the two treatments and reduced the potentially positive effects of the first treatment on the

stimuli in the second treatment. Thus, the primary goal of this study was to show

whether or not the treatment was beneficial and not to demonstrate potential maximum

benefit from the treatments. Therefore, each subject was advanced to Treatment 2 after

10 treatment sessions plus four maintenance sessions over a two-week break (i.e., 14 total

sessions) regardless of degree of progress in treatment one. Criterion to demonstrate the

efficacy of treatment was a 21% or greater increase in the number of stimuli orally named

correctly from the first to the tenth session of treatment. While no standards exist

regarding the degree of improvement necessary to be interpreted as an effect of treatment,

an improvement of 21% after only ten sessions of treatment (plus four maintenance

sessions) was thought to represent a clinically meaningfutil change in behavior over a short

period of time.








Treatment 2 for each subject began after maintenance measures for Treatment 1 were

completed. Because the stimuli for the second treatment were monitored in an extended

baseline prior to the second treatment, no additional pretreatment measures were

necessary before Treatment 2 began. Maintenance measures were identical following

completion of Treatment 2 as for Treatment 1, although the maintenance period was

extended up to three months for some subjects, depending upon subject availability.

Finally, the name of the word set (i.e., semantic training set, phonological training

set, and generalization training set) was derived from the behavior (i.e., semantic

treatment, etc.) assigned to the word set. For example, the semantic training set

designated that these stimuli underwent the semantic treatment, not stimuli that were

purposefully related to each other with semantic information. The purpose of linking the

treatment behavior to the name of the word set was to facilitate the reader's ease in

tracking changes over time over the various conditions of treatment.

Treatment Protocols

In each treatment the examiner presented a picture and asked the subject to name it

aloud. The examiner provided the subject with immediate accuracy feedback. The

primary purpose of the treatments was to teach the strategies for word retrieval and not to

simply learn the names of the specific pictures being trained. Consequently, the subject

proceeded through the three steps of treatment regardless of his success or failure in

orally naming the picture. The examiner cued the subject to retrieve the name through

the use of phonological questions (i.e., making a phonological judgment) in one treatment

and through the use of semantic questions (i.e., making a semantic judgment) in the other

treatment. The six questions were designed to obtain a yes or no response regarding each






44

of the 20 trained pictures. To increase the likelihood that the patient continued to answer

the questions (i.e., make a judgment) throughout the length of the study, and to reduce the

possibility that the patient learned a response of "no" to a particular foil-question, three

different foil-questions were asked across rotating sessions to obtain a response of "no."

For example, for the target picture cat in session one the examiner asked, "Is this similar

to a lion?" In session two the examiner asked, "Is this similar to a desk?" In session

three the examiner asked, "Is this similar to a lion?" In session four, the examiner

displayed a picture of a cat and asked, "Is this similar to a chair?" and so forth. The same

procedure for rotating three different foil-questions to elicit a response of "no" was

followed throughout the study with each of the six different questions in both the

semantic and phonological treatments. Appendices G-H display abbreviated treatment

protocols for the semantic and phonological treatment.

For each picture stimulus, the order of the yes-no questions was counterbalanced

across trials. For example, in trial one if the correct response for the stimulus cat is

"yes," then in trial two the correct response for cat would be "no." In addition, in each

session the 20 yes-no responses were counterbalanced across sessions. For example, a

"yes" response was the designated correct response across 50% of the trials in each

session, and a response of "no" was the designated correct response in 50% of the trials

in each session.

Semantic Treatment

Introduction

The sequence of steps involved in the semantic treatment is shown in Appendix E.

The examiner introduced the purpose of the semantic cues by saying, "I'm going to show








you some pictures and ask you to name them. Then, we'll practice some ways to help

you remember the word-using categories. Maybe these cues will help you remember

this word or another word the next time you want to say it."

Step 1: oral naming

The examiner displayed a picture and asked, "What is this?" The examiner allowed

up to 10 seconds for the patient to name the picture. If the response were correct, the

examiner said, Yes, __(e.g., "cat"). Let's practice." When the subject was unable to

respond correctly, he or she was supplied with the correct response. If the response was

incorrect or No Response, the examiner said, "No, let's practice it."

Step 2: question subordinate category

The examiner then asked a question regarding the subordinate category of the

picture. For example, for the target picture barrel for a correct Yes response, the

examiner asked, "Is this in the category of containers?" For a correct No response, the

examiner asked, "Is this in the category of plants?" (or health items or biological items?).

If the response was correct, the examiner nodded Yes and said," It's a container." If the

response was incorrect or No Response, the examiner nodded no and informed the

subject of the correct subordinate category by saying, "It's a container." The questions

were designed to force the subject to make a semantic judgment regarding the

subordinate category of the picture stimulus.

Step 3: question coordinate words

The examiner then asked a question regarding a coordinate word in the same

category as the stimulus picture. For example, for the target picture barrel for a correct

Yes response, the examiner asked, "Is this in the same group of things as a bucket?" For






46

a correct No response, the examiner asked, "Is this in the same group of things as a

coach?" (or clerk or repairman). If the response was correct, the examiner nodded yes and

said, "It's in the same group of things as a bucket." If the response was incorrect or No

Response, the examiner nodded no and said, "It's in the same group of things as a

bucket." The questions were designed to force the subject to make a semantic judgment

regarding the coordinate relationship of words in a category.

Step 4: question associate words

The examiner then asked a question regarding an associated word in the same

category as the stimulus. For example, for the target picture barrel for a correct Yes

response, the examiner asked, "Can this hold beer?" For a correct No response, the

examiner asked, "Can this hold a backhoe?" (or steamroller or bulldozer). If the response

was correct, the examiner nodded yes and said, "Yes, it can store beer." If the response

was incorrect or No Response, then the examiner nodded no and informed the subject of

the correct response by saying, "It can be used to store beer." The questions were

designed to force the subject to make a semantic judgment regarding the associative

relationship of words in the same category.

Note that many of the correct No questions were distant (i.e., farfetched) from the

target to make the semantic decision easier for the subject to determine and to elicit

consistent yes-no responses across all subjects. When more closely related semantic

questions were pilot tested with normal subjects, there was much variability in yes-no

responses depending on the individual subject's background.








Step 5: retest 1 oral naming

After the three questions, then the examiner asked once again, "What is this?" The

examiner allowed 10 seconds for the subject to orally name the picture. The purpose of

the retest was to determine whether or not the cues were effective in eliciting the stimulus

when the stimulus could not be retrieved in Step 1 before the cues were given. If the

response was correct, the examiner said, "Yes, it's a barrel. Now let's practice saying it

in unison." If the response was incorrect or No Response, the examiner nodded no and

said, "It's a barrel. Now, let's practice it in unison."

Step 6: rehearsal repetition in unison

Then, the subject began the rehearsal (i.e., repetition) phase of the treatment to

attempt to consolidate the stimulus into long-term memory. First, the subject said the

word in unison with the examiner three times. If the subject had difficulty with word

production in unison, and if the examiner thought that it was appropriate (i.e., due to

apraxia of speech), then the examiner directed the subject to "Watch my mouth." If the

subject was still unable to repeat the stimulus correctly three consecutive times in unison,

then the examiner stopped for a brief period (i.e., three seconds) and then re-attempted

the task. If the subject was still unable to repeat the word correctly three consecutive

times in unison, then the examiner proceeded to the next step.

Step 7: rehearsal repetition after a model

Then the subject proceeded to repeat the stimulus after the examiner three times. In

this second phase of rehearsal, the examiner said, Repeat this word after me. Try to

think of the cues as you say it." If the repetition was correct, the examiner continued to

model the stimulus. If the response was incorrect on any of the three repetitions, the






48

examiner allowed the subject additional practice until he could name the picture

correctly-up to a maximum of five attempts after any one repetition trial.

Step 8: rehearsal solo verbal production

Then, the examiner said, "Say it three times by yourself." If the response was

correct, the examiner nodded yes. If the response was incorrect or No Response on any

of the three attempts, the examiner modeled the stimulus aloud and said, "Now say it

three more times by yourself."

Step 9: rehearsal silent rehearsal.

The examiner removed the picture and said, "Now keep saying it silently." (i.e.,

internal rehearsal). After five to ten seconds, the examiner said, "Think of the name

again."

Step 10: retest 2- oral naming

Then, the examiner displayed the picture and asked, "What is this?" If the response

was correct, the examiner nodded or said, "Yes, ("barrel," etc.). If the response was

incorrect or No Response, the examiner gave the correct response and proceeded to the

next picture.

Phonological Treatment

Introduction

The examiner introduced the purpose of these phonological cues by saying, "I'm

going to show you some pictures and ask you to name them. Then we'll practice some

ways to help you remember the word-using first sounds, syllables, and rhymes. Maybe

these cues will help you remember this word or another word like it the next time you

want to say it."








Step 1: oral naming

The examiner displayed a picture and asked, "What is this?" The examiner allowed

up to 10 seconds for the patient to name the picture. If the response was correct, the

examiner said, "Yes, (cup)". Let's practice. When the subject was unable to respond

correctly, he or she was supplied with the correct response. If the response was incorrect

or No Response, the examiner said, "No, let's practice it."

Step 2: question initial phoneme

The examiner then asked, "Is the first sound __?" For example, for the target

picture cu for a correct Yes response, the examiner asked," Is the first sound 'k'?" For a

correct No response, the examiner asked, "Is the first sound 't'?" (or 'g' or 's,' etc.) The

questions were designed to force the subject to make a phonological judgment regarding

the initial sound of the word. If the response was correct, the examiner nodded yes and

said, "It begins with 'k'." If the response was incorrect or No Response, the examiner

nodded no and informed the subject of the correct initial phoneme by saying, "It begins

with 'k.'

Step 3: question rhyme

The examiner then asked, "Does this rhyme with __?" For example, for the target

picture cu for a correct Yes response, the examiner asked, "Does this rhyme with pup?"

For a correct No response, the examiner asked, "Does this rhyme with seed?" (or time or

squawk). The questions were designed to force the subject to make a phonological

rhyming judgment. If the response was correct, the examiner nodded yes and said, "It

rhymes with pup." If the response was incorrect or No Response, the examiner nodded

no and said, "It rhymes with pup."








Step 4: question syllable length

The examiner then asked, "Does this word have (one, two, three, or four)

syllables?" For example, for the target picture cup for a correct Yes response, the

examiner asked, "Does this have one syllable?" For a correct No response, the examiner

asked, "Does this have two (three or four, etc) syllables?" The questions were designed

to force the subject to make a phonological decision regarding whether the word was long

or short by judging syllable length. If the response was correct, the examiner nodded yes

and said, "It has one syllable." If the response was incorrect or No Response, the

examiner nodded no and said, "It has one syllable."

Step 5: retest 1 oral naming

After the three questions, then the examiner again asked, "What is this?" The

examiner allowed up to 10 seconds for the subject to orally name the picture. The

purpose of the retest was to determine whether or not the cues were effective in eliciting

the stimulus when the stimulus could not be retrieved in Step 1 before the cues were

given. If the response was correct, the examiner nodded yes and said, "Yes, it's a cup.

Now let's practice saying it in unison." If the response was incorrect or No Response, the

examiner nodded no and said, "It's a cup. Now, let's practice in unison."

Step 6: rehearsal repetition in unison

Then, the subject began the rehearsal phase of the treatment to attempt to consolidate

the stimulus into long-term memory. In the first step of rehearsal, the subject said the

word in unison with the examiner three times. Specifically, the examiner said, "Say this

word with me three times." If the subject had difficulty with word production in unison,

and if the examiner thought that it was appropriate (i.e., due to apraxia of speech), then








the examiner directed the subject to "Watch my mouth." If the subject was still unable to

repeat the stimulus correctly three consecutive times in unison, then the examiner stopped

for a brief period (i.e., three seconds) and then re-attempted the task. If the subject was

still unable to repeat the word correctly three consecutive times in unison, then the

examiner proceeded to the next step.

Step 7: rehearsal repetition after a model

Then the subject proceeded to repeat the stimulus after the examiner three times. In

this second phase of rehearsal, the examiner said, "Repeat this word after me. Try to

think of the cues as you say it." If the repetition was correct, the examiner continued to

model the stimulus. If the repetition was incorrect on any of the three repetitions, the

examiner allowed the subject additional practice until he could name the picture

correctly-up to a maximum of five attempts after any one repetition trial.

Step 8: rehearsal solo verbal production

Then the examiner said, "Say it three times by yourself." If the response was

correct, the examiner nodded yes. If the response was incorrect on any of the three

attempts, the examiner modeled the stimulus aloud and said, "Now say it three more

times by yourself."

Step 9: rehearsal solo silent production

The examiner removed the picture and said, "Now keep saying it silently." After

five to ten seconds, the examiner said, "Think of the name again."

Step 10: retest 2 oral naming

Then, the examiner displayed the picture and said, "What is this?" If the response

was correct, the examiner nodded and said, "Yes, __ ("cup," etc.) (The examiner








repeated the word to reinforce the correct word production.) If the response was

incorrect or No Response, the examiner gave the correct response and proceeded to the

next picture.

Maintenance phase

To assess posttreatment maintenance of oral naming performance, the examiner

readministered all stimuli for oral naming and oral reading. The two treatments were

separated by a period of at least two weeks, during which time follow-up measures were

taken, which were identical to follow-up measures taken at the conclusion of the first

treatment. In some instances when the subject was available, posttreatment follow-up

measures for periods longer than two weeks were obtained (i.e., two-three months). The

follow-up schedule used for maintenance measures included the following: a) one day

posttreatment, b) three days posttreatment, c) seven days posttreatment, and d) 14 days

posttreatment which also was the pretreatment probe measure for the first session of

Treatment 2.

Generalization probes

To determine if generalization of the treatments) occurred from the trained picture

stimuli to the untrained picture stimuli, a set of 20 generalization picture stimuli was

probed continuously from the beginning to the end of the study.

Scoring and Analysis

Verbal responses were coded as correct or incorrect throughout the experiment.

Responses were scored on the patient's first attempt to orally name the picture (or orally

read the nonword). However, if the initial response for oral naming was incorrect by

only one phoneme, then the response was scored correct because the treatment was








designed to examine the patient's ability to retrieve lexical items rather than to articulate

the word. In addition, if the oral response was self-corrected midstream (i.e., before the

subject finished articulating the word), then the response was scored correct. The

dependent variable was the number of correct responses in each task for each set of

stimuli (three picture naming sets and one control task set).

One examiner scored all results for each subject. A second examiner scored the

subject's responses from videotapes or audiotapes for 30% of the sessions. Interobserver

scoring reliability for correct and incorrect responses was sampled. Point-to-point

scoring was 80% or better.

The number of correct responses over a period of time have been presented in graph

form and examined to determine if the dependent variable (i.e., oral naming) changed as

a function of the independent variable (i.e., treatment). The graphed data have been

displayed across time, behaviors, and subjects. The following three parameters were

evaluated in these data: a) the trend of the data, b) the level at which the behavior was

occurring according to the data, and c) the slope of the data pattern (McReynolds and

Keamrns, 1983).

The effectiveness of the independent variable (i.e., treatment) was measured by

comparing the direction or trend of the behavior before and after treatment and was

indicated by an increase, decrease, or no change in the occurrence of the behavior

(McReynolds and Keamrns, 1983). Second, the level at which the behavior occurred

before treatment (i.e., pretreatment baseline) was examined from the graphed data.

Although no standards exist regarding how high or low the rates should be in baseline,

the best criterion is that it must be low enough that the data in the treatment phase can be






54

defended (McReynolds and Keamrns, 1983). Third, the degree of the slope in the trend

was evaluated to indicate how strong the trend was. McReynolds and Keamrns (1983)

stated that if there is a pronounced slope in the trend, then it is evidence that the trend is

stronger than if the slope is a gentle one. The ultimate decision regarding the effects of

treatment was based on a comparison between the pretreatment and treatment phases

(McReynolds and Kearns, 1983).














CHAPTER 3
RESULTS

The purpose of this study was a comparison of the efficacy of two treatments-

semantic and phonological-for anomia. Three experimental questions were asked, and

the results relevant to each question will be discussed separately below.

Semantic Treatment Research Question 1-a

Will aphasic subjects demonstrate significant improvement in oral-naming

performance as the result of a semantic treatment?

Oral naming of trained stimuli

Table 3-1 displays the rating scale that served as our operational definition of degree

of improvement in oral-naming performance during the two treatments and reflects the

limited number of sessions (10) in this study. For example, if a subject showed a 10%

gain after only ten sessions of treatment, that gain was considered a mild improvement.


Table 3-1: Oral-Naming Performance Rating Scale

Percent Correct (100 percent possible) Degree of Improvement
0% No improvement
1-10% Mild improvement
11-25% Moderate improvement
26-100% Significant improvement


Figure 3-1 displays a composite of the results of the individual performances of the

four subjects across four behaviors observed for the duration of the study. The four

behaviors include two oral-naming behaviors (on different picture sets for two different












Subject I (AW)
Baseline Phono Tx Semanti Txi
20

10 --II-Tx 2-Sem

Z 0 I Control
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of Sessions


Subject 2 (AS)


15 3-Tx I -Ser
N e10. -STx 2-Phon
S-0-X Control
2 0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of Sessions

Subject 3 (PB)
Baseline Semantic Tx Phon TxMaint
.20
Is -4- Tx l-Sere
o10 -U Tx 2-Phon
E -A-- Gen
-40---Control
z 0)

Number of Sessions

[~~fe [~~x Subject 4 (FSti x [f
20
Is4-Tx 1-Phon
U
10 -U-Tx 2-Sem
EI5 AGen
z 0.iiii --X(--Control


u- mb e (M Se ss io ns
Number of Sessions


Figure 3-1: Composite of performance for oral naming, generalization, and maintenance.


qr V,9










treatments), generalization, and the nonword reading/word repetition control task.

Subject 2 (AS) demonstrated significant improvement in oral-naming performance

for trained pictures. Prior to the administration of the semantic treatment, her baseline

performance was established as stable across four sessions at 0% correct in each session.

At the beginning of the semantic treatment, Subject 2 correctly named zero of 20 trained

picture stimuli (0% correct). After the semantic treatment she orally named 12 of 20

(60% correct) trained picture stimuli, a gain of 60% over 10 sessions of treatment.

Subject 2 received the semantic treatment first, and thus the results of her oral-naming

performance were due to the semantic treatment itself and were not influenced by a prior

treatment.

Subject 4 (FS) demonstrated significant improvement in oral-naming performance as

well. Prior to the administration of the second treatment, his baseline performance was

monitored across 19 sessions. Baselines scores of these untrained stimuli ranged from

one to 10 of 20 correct responses while the first (phonological) treatment was being

administered and then stabilized at 10 of 20 (50% correct) correct responses. At the

beginning of the semantic treatment, Subject 4 correctly orally named 10 of 20 (50%

correct) of these (currently) trained picture stimuli. After the semantic treatment, he

correctly orally named 20 of 20 (100% correct) trained picture stimuli, a gain of 50%

over 10 sessions of treatment. However, Subject 4 received the semantic treatment

second, and thus the results of his oral-naming performance may have been influenced to

some degree by the preceding phonological treatment as well as by the semantic

treatment. However, his baseline performance was stable prior to initiation of the second






58

treatment, suggesting that the majority of his improvement in the second treatment was

due to the semantic treatment.

Subject 1 (AW) demonstrated a mild-to-moderate improvement in oral-naming

performance as a result of the semantic treatment. Prior to the administration of the

second treatment, his performance was monitored across 16 sessions. Baseline scores

ranged from three to nine of 20 correct responses while the first (phonological) treatment

was administered and then stabilized at eight of 20 (40% correct) correct responses. At

the beginning of the semantic treatment, Subject 1 correctly orally named five of 20 (25%

correct) trained picture-stimuli. After the semantic treatment, he correctly orally named

11 of 20 (55% correct) trained picture stimuli, a gain of 30% over 10 sessions of semantic

treatment. However, because Subject 1 's baseline performance had stabilized at 40%

correct (eight of 20) prior to the administration of the second treatment, the net gain from

the second treatment was 15% (55-40% accuracy) instead of 30%. Subject 1 also

received the semantic treatment second.

Subject 3 (PB) did not show an effect of treatment with the semantic treatment (or

the phonological treatment). Prior to the administration of the first treatment, her

baseline performance was established as stable across five sessions with 10% correct in

each session (two of 20 correct). At the beginning of the semantic treatment, Subject 3

correctly orally named two of 20 (10% correct) trained picture stimuli correctly. After

the semantic treatment, she correctly orally named one of 20 (5% correct), a loss of 5%

over 10 sessions of treatment. Subject 3 received the semantic treatment first.










First Treatment Semantic


Baseline Semantic Treatment Maintenance


-S2


1 3 5 17 9 11 13 15 17 19 21 23 25 27 29 31 33 35
SNumber of Sessions 3 months post treatment
I


0~









E .=
0


~II
I)












mE
z


Second Treatment + Semantic
I Semantic Treatment Maintenance
I Semantic Treatment Maintenance


--*-S41


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 3
Number of Sessions 2 months post treatment


Figure 3-2: Oral-Naming Performance with Semantic Treatment.


Table 3-2: Oral-Naming Performance with Semantic Treatment


Semantic Treatment Administered First
Subjects # Correct, # Correct, Net Gain # Correct,
Beginning of End of Or Loss at End End of
Treatment/ Treatment/ of Treatment/ Maintenance/
# possible # possible 100% possible # possible
S2 (AS) 0/20 (0%) +12/20 (60%) 60% +9/20 (45%)*
S3 (PB) +2/20 (10%) + 1/20 (5%) -5% +1/20 (5%)*
Semantic Treatment Administered Second
S4 (FS) +10/20 (50%) +20/20 (100%) 50% +20/20 (100%)**
S1 (AW) +5/20 (25%) +11/20 (55%) 15%*** +11/20 (55%)**
Key: 20 points possible & 100 percent possible
Maintenance after two weeks
** Maintenance after two months
*** S I concluded the first treatment with 40% correct and concluded the second
treatment with 55% correct, for a gain of 15% during the semantic treatment.


Baseline








Oral-naming summary

Figure 3-2 and Table 3-2 display the results of the semantic treatment for oral

naming of trained stimuli for the four subjects. Two of the four subjects demonstrated

significant improvement (50-60%), one subject demonstrated a moderate improvement

(15%), and one subject demonstrated a loss (-5%) in oral picture-naming performance

with the semantic treatment. Thus, response to oral-naming performance with the

semantic treatment was variable with three aphasic patients demonstrating improvement,

whereas one did not respond.

Semantic Treatment Research Question 1-b

If the semantic treatment is effective, do the changes in oral-naming performance

also generalize to untrained stimuli?

Generalization to untrained stimuli

Figure 3-3 and Table 3-3 display the results of generalization of 20 untrained picture

stimuli (generalization word sets only). Generalization in the amount of 15% and 5% to

the untrained word set was observed in two subjects (Subjects 4 and 1, respectively)

when the semantic treatment was administered second. Prior to the administration of the

second (semantic) treatment, Subject 4's baseline performance was monitored across 19

sessions. Baseline scores ranged from one to 15 of 20 (75% correct) correct responses on

untrained stimuli while the first (phonological) treatment was administered and then

stabilized at 13 of 20 (65% correct) correct. At the beginning of the second (semantic)

treatment, Subject 4 correctly orally named 10 of 20 (50% correct) untrained stimuli and

at the end of treatment correctly orally named 17 of 20 untrained stimuli (85% correct), a












Baseline
S 20

10

0 -
E o -


First Treatment Semantic
Semantic Treatment LAintan .nt-


1 3 5 17 9 11 13 15 17 19
I Number of Sessions
I


Baseline


------1
Second Treatment Semantic
I Semantic Treatme


3 months posttreatment


nt Maintenance


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33


Number of Sessions


2 months posttreatment


Figure 3-3: Generalization During Semantic Treatment (Untrained Generalization Word
Set Only).

Table 3-3: Generalization During Semantic Treatment (Untrained Generalization Word
Set Only)

First Treatment Semantic
Subjects # Correct, # Correct, Net Gain # Correct,
Beginning of End of Or Loss/ End of
Treatment/ Treatment/ 100% Maintenance/
# possible # possible possible # possible
S2 (AS) +1/20 (5%) +1/20 (5%) 0% +2/20 (10%)
S3 (PB) +3/20 (15%) +1/20 (5%) -10% +3/20 (15%)
Second Treatment Semantic
S1 (AW) +2/20 (10%) +6/20 (30%) 5%* +2/20 (10%)
S4 (FS) +13/20 (65%) +17/20 (85%) 15%** +18/20 (90%)
Key: 20 points possible & 100 percent possible
S1 achieved generalization rates as high as 45% and 50% accuracy in the first and
second treatments, respectively, for a net gain of 5%.
** S4 achieved generalization rates as high as 75% and 90% accuracy in the first and
second treatments, respectively, for a net gain of 15%.






62

gain of 35%. However, because Subject 4 achieved 75% accuracy in oral naming of

untrained stimuli at one point during the first treatment and achieved 90% accuracy in

oral naming of untrained stimuli at one point during the second treatment, the net

gain in generalization during the second (semantic) treatment was 15% (90-75%), not

20%.

Prior to the administration of the second (semantic) treatment, Subject 1's baseline

performance was monitored across 14 sessions. Baseline scores ranged from two to nine

of 20 (45% correct) correct responses while the first (phonological) treatment was

administered and then stabilized at seven of 20 (35% correct) correct. At the beginning

of the second (semantic) treatment, Subject 1 correctly orally named two of 20 (10%

correct) untrained stimuli and at the end of treatment correctly orally named six of 20

untrained stimuli (30% correct), a gain of 20%. However, because Subject 1 achieved

45% accuracy in oral naming of untrained stimuli at one point during the first treatment

and achieved 50% accuracy in oral naming of untrained stimuli at one point during the

second treatment, his net gain in generalization during the semantic treatment to

untrained stimuli was 5% (50-45%), not 20%.

No generalization was observed on the untrained word set in Subject 2 (AS) when

the semantic treatment was administered first, and, in this condition one subject (Subject

3) demonstrated a loss in oral-naming performance on the generalization word set.

Control measure

Because oral reading of nonwords is a task similar to oral naming of nouns but which

is thought to use different processes in the lexical system than used by oral naming, we

chose to follow performance on nonword oral reading as a control measure in three of the












Baseline


First Treatment Semantic
Semantic Treatment Maintenance


-4-S2


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33
I Number of Sessions


-4
---------------------I
Second Treatment Semantic
SSemantic Treatment Maintenance


-"-S4


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33
Number of Sessions


Figure 3-4: Nonword Oral-Reading/Word-Repetition Control Tasks During
Semantic Treatment.


Table 3-4: Nonword Oral-Reading/Word-Repetition Control Task During
Semantic Treatment

Semantic Treatment Administered First
Subjects # Correct, # Correct, Net Gain # Correct,
Beginning of End of or Loss at End End of
Treatment/ Treatment/ of Treatment/ Maintenance/
# possible # possible 100% possible # possible
S2 (AS) 0/20 (0%) 0/20 (0%) 0% (0%) 0/20 (0%)
S3 (PB) +4/20 (20%) +1/20 (5%) -3 (-15%) +1/20 (5%)
Semantic Treatment Administered Second
S4 (FS) +2/20 (10%) +1/20 (5%) -1 (-5%) +1/20 (5%)
Sl (AW) +9/20 (45%) +7/20 (35%) -2 (-10%) +7/20 (35%)
Key: 20 points possible & 100 percent possible








four subjects. However, because of his extraordinary difficulty with reading (i.e.,

premorbid dyslexia), Subject 1 refused to participate in oral-reading tasks. Therefore,

repetition of words was substituted for nonword oral reading as a control measure for

him. If an effect of treatment was observed with the trained (and untrained) stimuli but

not with the nonword oral reading/word-repetition task, then the treatment itself was

thought to be responsible for improvement in oral naming rather than other factors in the

environment.

No improvement was observed in nonword oral reading during the semantic

treatment for any of the subjects. Subject 2 (AS) correctly orally read 0 of 20 (0%

correct) untrained nonwords from beginning to end of the semantic treatment. The other

three subjects (2, 3, and 4) dropped in nonword oral-reading/word repetition performance

from beginning to end of the semantic treatment. Subject 2 (PB) correctly orally read

four of 20 (20% correct) nonwords at the beginning of treatment and one of 20 (5%

correct) at the end of treatment, a loss of 15%. Subject 4 (FS) correctly orally read two of

20 (10% correct) nonwords at the beginning of treatment and one of 20 (5% correct) at

the end of treatment, a loss of 5%. Subject 1 (AW), correctly repeated nine of 20 (45%

correct) words at the beginning of treatment and seven of 20 (35% correct) at the end of

treatment, a loss of 10%.

Thus, with no accompanying inclines in performance in nonword oral-reading/word

repetition across the semantic treatment, experimental control was maintained. Figure

3-4 and Table 3-4 display the results of the nonword oral-reading/word-repetition task.








Generalization/control summary

The effects of the semantic treatment were minimal on the 20 untrained stimuli from

the generalization word lists with the greatest gain occurring when the semantic treatment

was administered second (with Subject 4). No generalization to this untrained word set

was observed when the semantic treatment was administered first.

Baseline performances in the control tasks for all four subjects remained stable (i.e.,

demonstrated no upward trends) throughout the semantic treatment, and thus

experimental control was maintained.

Semantic Treatment Research Question 1-c

Is the change in oral-naming performance enduring two weeks after the semantic

treatment?

Maintenance

Figure 3-2 and Table 3-2 also display the results of maintenance after the semantic

treatment. Maintenance of trained stimuli in the semantic word sets was evaluated over

two different time periods, depending upon whether the semantic treatment was

administered first or second. When the semantic treatment was administered first,

maintenance was observed over a two-week period, before the second treatment was

initiated. When the semantic treatment was administered second, maintenance was

observed over a two-month period after treatment ended.

As seen in Figure 3-2 and Table 3-2, maintenance of oral-naming performance for

the trained pictures endured for two months at the same levels as that seen immediately

after completion of treatment in two subjects (Subjects 1 and 4) who received the

semantic treatment second.






66

Subject 4 (FS), who received the semantic treatment second, correctly orally named

20 of 20 (100% correct) trained pictures at the end of the semantic treatment and

maintained his level of performance at 100% accuracy for two months after treatment

ended. Subject 1 (AW), who also received the semantic treatment second, correctly

orally named 11 of 20 (55% correct) at the end of the semantic treatment and maintained

his level of performance at 55% accuracy for two months after treatment ended.

Subject 2 (AS), who received the semantic treatment first, completed oral naming of

trained pictures at 60% accuracy after 10 sessions of the semantic treatment, but her

performance dropped to 45% accuracy two weeks after treatment ended, a loss of 15%.

Subject 3 (PB), who also received the semantic treatment first, completed oral

naming of trained pictures at 5% accuracy after 10 sessions of semantic treatment and

maintained her level of performance two weeks after treatment ended. Note that Subject 3

did not respond to treatment and no gains were made to maintain afterwards.

Maintenance summary

Two subjects (Subjects 1 and 4), who received the semantic treatment second,

demonstrated enduring oral-naming performance (i.e., showed no loss in oral-naming

performance) two months after the semantic treatment at the same level as they had

demonstrated at the end of the semantic treatment. One subject (Subject 2), who received

the semantic treatment first, showed some loss two weeks after treatment ended. One

subject (Subject 3), who also received the semantic treatment second, did not respond to

treatment, ending both treatment and two weeks of maintenance with only one of 20

correct responses.








Thus, the effects of the semantic treatment in oral-naming performance were

enduring at two months after the semantic treatment ended for the two subjects who

received the semantic treatment second but were slightly less enduring after two weeks in

the subject who received the semantic treatment first.

Phonological Treatment Research Question 2-a

Will aphasic subjects demonstrate a significant improvement in oral-naming

performance as the result of a phonological treatment?

Oral naming of trained stimuli during phonological treatment

Figure 3-5 and Table 3-5 display the results of the phonological treatment for the

four subjects. Subject 1 (AW) and Subject 4 (FS), who received the phonological

treatment first, both showed a significant improvement in oral naming of trained picture

stimuli. Prior to the administration of the phonological treatment, Subject 1 's (AW)

baseline performance for oral picture-naming was established as stable across three

sessions at two of 20 correct (10% correct) with a range of two to three of 20 correct. At

the beginning of the phonological treatment, Subject 1 correctly named seven of 20 (35%

correct) trained picture stimuli. After the phonological treatment, Subject 1 correctly

orally named 15 of 20 (75% correct) trained picture stimuli, an improvement of 40% over

10 sessions of treatment.

Likewise, at the beginning of the phonological treatment, Subject 4's (FS) baseline

performance for oral picture-naming was established as stable across six sessions at two

of 20 correct (10% correct) with a range of two to four of 20 correct. At the beginning of

the phonological treatment, Subject 4 correctly named 11 of 20 trained picture stimuli














First Treatment Phonological
Baseline Phonological Treatment Maint


F~]


1 3 5 17 9 11 13 15 17 19 21 23 25 27 29 31 33
SNumber of Sessions 2 months posttreatment


Baseline


Second Treatment A Phonological
SPhonoloaical Treatment Maintenance


o0
=0 E
0
o











IL
eaz
00






oE
5


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35


Number of Sessions


3 months posttreatment


Figure 3-5: Oral-Naming Performance with Phonological Treatment.

Table 3-5: Oral-Naming Performance with Phonological Treatment


Subjects # Correct, # Correct, Net Gain # Correct,
Beginning of End of or Loss at End of
Treatment/ Treatment/ End of Maintenance/
# possible # possible Treatment/ # possible
100%
I_ possible
S4 (FS) +11/20 (55%) +19/20 (95%) 40% +18/20 (90%)*
SI (AW) +7/20 (35%) +15/20 (75%) 40% +14/20 (70%)*
Phonological Treatment Administered Second
S2 (AS) +6/20 (30%) +14/20 (70%) 40% +13/20 (65%)**
S3 (PB) 0/20 (0%) 0/20 (0%) 0% 0/20 (0%)*
Key: 20 points possible & 100 percent possible
After two weeks
** After three months


- S-2
IIJ


Phonological Treatment Administered First








(55% correct). After the phonological treatment, Subject 4 correctly named 19 of 20

trained picture stimuli (95% correct), an improvement of 40% over 10 sessions of

treatment. Because Subjects 1 and 4 received the phonological treatment first, the results

of their oral-naming performance were due to the phonological treatment itself and were

not influenced by a prior treatment.

Subject 2 (AS) also showed a 40% improvement in naming trained pictures. Her

baseline performance was established as stable across 17 sessions at seven of 20 correct

(35% correct). At the beginning of the phonological treatment, Subject 2 correctly orally

named six of 20 (30% correct) trained picture stimuli. After the phonological treatment,

she correctly orally named 14 of 20 trained picture stimuli (70% correct), an

improvement of 40% over 10 sessions of the phonological treatment. However,

Subject 2 received the phonological treatment second, and thus the results of her oral

naming performance may have been influenced to some degree by the preceding

semantic treatment as well as by the current phonological treatment.

Subject 3 (PB) did not show an effect of treatment with the phonological treatment

(or the semantic treatment). Prior to the administration of the phonological treatment, her

baseline performance in oral naming was established as stable across 16 sessions at one

of 20 correct (5% correct) with a range of zero to three of 20 correct. At the beginning

and end of the phonological treatment, Subject 3 correctly named zero of 20 pictures (0%

correct).

Oral-naming summary

Three of the four subjects (Subjects 1, 2, and 4) demonstrated a clinically significant

improvement (40% each) in oral-naming performance after the phonological treatment.






70

One subject (Subject 3) did not respond to the phonological treatment (i.e., no gains).

Two of the subjects (Subjects 1 and 4) received the phonological treatment first, and thus

their gains in treatment were thought to be due to the phonological treatment itself.

Thus, aphasic subjects who responded to treatment demonstrated a clinically significant

improvement in oral-naming performance as the result of a phonological treatment.

Phonological Treatment Research Question 2-b

If the phonological treatment is effective, do the changes in oral-naming

performance also generalize to untrained stimuli?

Generalization to untrained stimuli during the phonological treatment

Figure 3-6 and Table 3-6 display the results of generalization of 20 untrained picture

stimuli (generalization words sets only). All four subjects demonstrated levels of

improvement ranging from 5% to 25% in oral-naming performance for untrained picture

stimuli.

Subject 4 (FS) demonstrated a significant gain in generalization to 20 untrained

stimuli with the phonological treatment. Prior to the administration of the phonological

treatment, his baseline oral-naming performance was established as stable with a down-

sloping trend across six sessions at two of 20 correct (10% correct) with a range of one to

seven of 20 correct. At the beginning of treatment, he correctly orally named 10 of 20

(50% correct) untrained stimuli and at the end of treatment correctly orally named 15 of

20 (75% correct). Subject 4 (FS), who received the phonological treatment first,

demonstrated a gain of 25% in oral-naming performance of untrained stimuli after 10

sessions of treatment.






71

Subject 1 (AW) demonstrated a mild gain in generalization to 20 untrained stimuli

with the phonological treatment. Prior to the administration of the phonological

treatment, his baseline performance was established as stable across three sessions at

three of 20 correct (15% correct) with a range of two to three of 20 correct. At the

beginning of treatment, he correctly orally named five of 20 (25% correct) untrained

stimuli and at the end of treatment correctly orally named seven of 20 (35% correct).

Subject 1 (AW), who also received the phonological treatment first, demonstrated a 10%

gain in oral naming of untrained stimuli after 10 sessions of treatment.

Subject 2 (AS) also demonstrated a mild gain in generalization to untrained stimuli

with the phonological treatment. Subject 2, who received the phonological treatment

second, showed a 10% improvement in orally naming the untrained generalization

pictures. Prior to the administration of the phonological treatment, her baseline

performance was established as stable across 17 sessions at two of 20 correct (10%

correct) with a range of zero to four of 20 correct. Subject 2 correctly orally named one

of 20 (5% correct) at the beginning of the phonological treatment and three of 20 (15%

correct) at the end of treatment. However, Subject 2 correctly orally named as many as

four of 20 (20% correct) during the first treatment, and correctly orally named as many as

five of 20 (25% correct) during the second treatment, resulting in a net gain of 5% (25-

20%) in generalization to untrained stimuli during the phonological treatment.

Likewise, Subject 3 (PB) demonstrated a mild gain in generalization to untrained

stimuli with the phonological treatment. Subject 3 (PB), who also received the










First Treatment Phonological


Baseline Phonological Treatment Maintenance


Si
L--o-SJ




I mS31


1 3 5 j7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of Sessions


C !5
o 0

o
0z







S U
*



o 02
1C
E1


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of Sessions


Figure 3-6: Generalization During Phonological Treatment (Untrained Generalization
Word Set Only)


Table 3-6:


Generalization During Phonological Treatment (Untrained Generalization
Word Set Only)


First Treatment Phonological
Subjects # Correct, # Correct, Net Gain at End # Correct,
Beginning of End of Treatment/ of Treatment/ End of
Treatment/ # possible 100% possible Maintenance/
# possible # possible
S4(FS) +10/20 (50%) +15/20 (75%) +5/20 (25%) +15/20 (75%)
SI (AW) +5/20 (25%) +7/20 (35%) +2/20 (10%) +7/20 (35%)
Second Treatment Phonological
S2(AS) +1/20 (5%) +3/20 (15%) +2/20 (5%)** +5/20 (25%)
S3 (PB) +1/20 (5%) +2/20 (10%) +1/20 (5%) *
Key: 20 points possible & 100 percent possible
Not measured due to lack of progress.
* S2 orally named as many as 4 of 20 correctly during the first treatment and as many
as five of 20 correctly during the second treatment, for a net gain of one of 20 (5%).


Second Treatmenl- Phonological
I Phonological Treatment Maintenance
I Phonological Treatment Maintenance


Baseline






73

phonological treatment second, showed a 5% improvement in naming the untrained

generalization pictures. Prior to the administration of the phonological treatment, her

baseline was established as stable across 16 sessions at three of 20 correct (15% correct)

with a range of one to three correct. Subject 3 correctly orally named one of 20 (5%

correct) at the beginning of treatment and correctly orally named two of 20 (10% correct)

at the end of treatment, a gain of 5%. Subject 3 correctly orally named as many as three

of 20 (15% correct) correct in the first treatment and named as many as four of 20 correct

(20% correct) in the second treatment, a gain of 5%.

Control measure

Figure 3-7 and Table 3-7 display the results of the nonword oral-reading/word-

repetition control task. No significant improvement (i.e., no upward-sloping trend) was

observed in nonword oral-reading during the phonological treatment for any of the

subjects. Subject 4 (FS) demonstrated a stable baseline with one-to-two of 20 correct

responses in nonword oral reading during the phonological treatment. Subject 3 (PB),

who did not respond to treatment, demonstrated a stable baseline with a range of two-to-

three of 20 correct responses throughout the phonological treatment. Subject 2 (AS)

correctly orally read 0 of 20 (0% correct) untrained nonwords from beginning to end of

the phonological treatment. Subject 1 (AW), who received the phonological treatment

first, demonstrated a loss in word-repetition from eight to six of 20 correct responses

during the phonological treatment. Thus, with no accompanying inclines in performance

in nonword oral-reading/word repetition across the phonological treatment, experimental

control was maintained.










First Treatment Phonological


1 3 5 17 9 11 13 15 17 19 21 23 25 27 29 31 33
I Number of Sessions


Baseline


Figure 3-7:


---------------1I
Second Treatment Phonological
I Phonological Treatment Maintenance


-4--S41




S2]






1-l--S


1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Number of Sessions


Nonword Oral-Reading/Word-Repetition Control Tasks During
Phonological Treatment


Table 3-7: Nonword Oral-Reading/Word-Repetition Control Task During
Phonological Treatment

Phonological Treatment Administered First
Subjects # Correct, # Correct, Net Gain # Correct,
Beginning of End of or Loss at End End of
Treatment/ Treatment/ of Treatment/ Maintenance/
# possible # possible # possible # possible
S1 (AW) +8/20 (40%) +6/20 (30%) -2 (-10%) +7/20 (35%)
S4 (FS) 0/20 (0%) +2/20 (10%) +2 (10%) +2/20 (10%)
Phonological Treatment Administered Second
S2 (AS) 0/20 (0%) 0/20 0%) 0 (0%) 0/20 (0%)
S4 (PB) +2/20 (10%) +3/20 (15%) +1 (5%) *
Key: 20 points possible & 100 percent possible
Not tested due to lack of progress








Generalization/control summary

All four subjects demonstrated generalization during the phonological treatment

ranging from 5% to 25% on 20 untrained stimuli in the generalization word sets. Two of

these four subjects (2 and 3) received the phonological treatment second, and their

generalization to untrained stimuli was 5% each. Subjects 4 and 1 who received the

phonological treatment first demonstrated generalization to 20 untrained pictures with

gains of 25% and 10%, respectively.

No upward-sloping trends) in baseline control tasks for nonword oral reading and

word-repetition indicated strong experimental control during the phonological treatment.

Thus, the effects of the phonological treatment in oral-naming performance

generalized to untrained stimuli with mild-to-moderate gains in all three subjects who

responded to treatment.

Phonological Treatment Research Question 2-c

Is the change in oral-naming performance enduring two weeks after the phonological

treatment ended?

Maintenance

Maintenance of trained stimuli in the phonological word sets was evaluated over two

different time periods, depending upon whether the phonological treatment was

administered first or second. When the phonological treatment was administered first,

maintenance was observed over a two-week period, before the second treatment was

initiated. When the phonological treatment was administered second, maintenance was

observed over a two-month period after treatment ended. Table 3-5 also displays the








results of maintenance of oral-naming accuracy for trained pictures during the

phonological treatment.

Subject 4 (FS), who received the phonological treatment first, completed oral naming

of trained pictures at 95% accuracy after 10 sessions of phonological treatment and

maintained his performance two weeks after the treatment ended at 90% accuracy, a loss

of only 5%. Likewise, Subject 1 (AW), who also received the phonological treatment

first, completed oral naming of trained pictures at 75% accuracy after 10 sessions of

phonological treatment and maintained his oral-naming performance two weeks after the

treatment ended at 70%, a loss of only 5%.

Subject 2 (AS), who received the phonological treatment second, completed oral

naming of trained pictures at 70% accuracy after 10 sessions of phonological treatment

and maintained her performance over three months at 65%, a loss of 5%.

Subject 3 (PB), who also received the phonological treatment second, did not make

any gains during phonological treatment (0% correct) and was not observed for

maintenance due to lack of progress.

Maintenance summary

Maintenance of oral-naming accuracy for trained pictures with the phonological

treatment endured at a high level in three of the subjects. Only a 5% loss was observed

after two weeks in the two subjects (Subjects 1 and 4) who received the phonological

treatment first and in one (Subject 2) of the subjects after three months who received the

phonological treatment second. One subject (Subject 3) did not respond to treatment and

had no gains to maintain.






77

Thus, the effects of the phonological treatment in oral-naming performance endured

at nearly the same level two weeks after the phonological treatment ended as they had at

the end of treatment in the two subjects who received the phonological treatment first and

were only slightly less enduring three months after treatment ended in the subject who

received the phonological treatment second.

Comparison Research Question 3-a

How do the semantic and phonological treatments compare on degree of change?

Comparison on degree of change

Comparisons of results of the semantic and phonological treatments are displayed in

composite Figure 3-1 and Table 3-8. These scores reflect performance for subjects only

when the designated treatment was administered first to avoid the influence of a prior

treatment. Improvement in oral naming of trained pictures was observed in both

treatments. The largest improvement in oral naming of trained pictures was seen in the

semantic treatment with Subject 2 (AS) who gained 60%. However, gains of 40% in oral

naming of trained pictures with the phonological treatment were demonstrated with both

Subjects 4 (FS) and 1 (AW). Thus, when a subject responded to the treatment, the

greater degree of improvement in oral-naming performance was observed with the

semantic than the phonological treatment.

Comparison Research Question 3-b

How do the semantic and phonological treatments compare on generalization? (See

Figure 3-1.)








Table 3-8: Comparison of Semantic and Phonological Treatments

Behavioral Measure Semantic Treatment Phonological Treatment
S2 (AS) S3 (PB)* S4 (FS) SI (AW)
Oral Naming 60% -5% 40% 40%

Generalization 0%** -10% 25% 10%
(20 Untrained)______ ___________________________
Maintenance 75% 0% 95% 93%
Key: 100 percent possible
S3 did not respond to either treatment.
** This score does not reflect the 27.5% level of generalization AS achieved with one
of the two untrained word sets during the semantic treatment.


Comparison of generalization

Generalization was greater to the untrained generalization word sets with the

phonological treatment (range of 5-25%) than the semantic treatment (range of-10 to

15%). More subjects demonstrated generalization with the phonological treatment (four

subjects) than with the semantic treatment (two subjects).

Comparison Research Question 3-c

How do the semantic and phonological treatments compare on maintenance? (See

Figure 3-1.)

Comparison of maintenance

Maintenance was measured on 20 trained pictures for two weeks after the first

treatments) ended. The gains for both treatments were essentially unchanged at follow-

up except in Subject 2 who lost some from the semantic treatment.

Treatment comparison summary

Three of the four subjects responded to treatment. An advantage in oral-naming

performance in trained nouns for the semantic treatment was observed in two subjects

and an advantage for the phonological treatment was observed in one subject. An






79

advantage for the phonological treatment was observed for generalization to the untrained

generalization word sets and maintenance of the trained word sets.














CHAPTER 4
DISCUSSION


This study investigated the efficacy of two treatments-semantic and phonological-

for anomia secondary to aphasia. The two treatments consisted of asking a patient yes-no

questions that require semantic or phonological judgments regarding a stimulus (i.e., a

word associated with a pictured object) immediately prior to requiring them to name the

stimulus aloud. These questions (semantic or phonological) were followed by a rehearsal

phase (i.e., saying the word aloud, then silently) to facilitate subsequent word production

attempts. A review of the literature suggested that the semantic treatment would be more

powerful than the phonological treatment in improving an anomic patient's attempts at

lexical retrieval. We compared performance across several aphasic subjects with anomia,

and these results have enabled us to reach several conclusions about the relative efficacy

of these two treatments. In this chapter we will discuss our conclusions as they relate to

previous research and as they relate to the cognitive neuropsychological mechanisms

underlying the oral-naming process.

Four subjects with a clinical diagnosis of aphasia were treated with both the semantic

and phonological treatments in counterbalanced order. A crossover single-subject

experimental design with multiple baselines across subjects and behaviors was used. The

experimental procedure consisted of oral naming of (pictured) nouns. We have compared

performance on this task within and across subjects and have interpreted these

comparisons as they relate to the research questions posed in the Introduction.






81

Research Questions

Research Questions 1-a and 2-a:

Will aphasic subjects demonstrate significant improvement in oral-naming

performance as the result of semantic and/or phonological treatments?

The results of study of the subjects in this project indicate that the answer to this

question is yes; three of four subjects responded to the semantic treatment, and the same

three of four subjects responded to the phonological treatment. As would be expected

with heterogeneous aphasic subjects, the degree of response to the two treatments varied

across subjects with some subjects responding better to one treatment than the other.

Specifically, we found that, as in prior studies, two (Subjects 2 and 4) of our three

subjects (who responded to the treatment) responded better to the semantic treatment than

to the phonological treatment. However, another (Subject 1) of the three subjects

responded to the phonological treatment more dramatically than the semantic treatment.

Finally one subject (Subject 3) did not respond at all.

An important interest of treatment efficacy research should be the identification of

factors that would make a subject a candidate for a particular treatment. Regarding the

question of who may or may not benefit from a treatment, one might look to the nature of

their lexical deficits to identify potential reasons. As described on page 33, Subjects 1, 2,

and 4 displayed similar deficits of the phonological output lexicon. Subject 1

demonstrated an additional impairment specific to visual input processing, and Subjects

1, 2, and 4 all appeared to have relatively spared semantic processing. In contrast,

Subject 3 suffered an impaired semantic system with some sparing of lexical level

processing. In reviewing their relative responses to the treatments, it is interesting to note






82

that both Subjects 2 and 4, the subjects with similar deficits, responded similarly to the

treatments, both in degree of response as well as direction. In comparison, the subject

(Subject 3) whose defect was quite dissimilar in nature (i.e., a semantic deficit) responded

quite differently to the interventions when compared to Subjects 2 and 4 in that she did

not respond at all. Thus, our experience with Subjects 2,4, and 3 would suggest that

patients whose deficits of the word retrieval system are similar in nature (Subjects 2 and

4) may respond similarly to treatments while those with qualitatively different deficits

(Subject 3) may be predicted to respond to treatment differently.

Within the cognitive neuropsychological disciplines of word retrieval there are

multiple contributory processing components. While two patients may share deficient

component deficits, they may also differ in the integrity of the remaining components,

and these may contribute additional differences to response to treatment. For example,

while Subjects 1, 2, and 4 all have relatively spared semantic systems and impaired

phonological output lexicons, Subject 1 has an additional problem at the input lexicon

level specific to visual information. This visual deficit is not noted in Subjects 2 and 4.

Additionally, we found that while Subject 1 responded to phonological treatment (and to

a similar degree-40% gain) like Subjects 2 and 4, he did not respond dramatically (15%

gain) to the semantic treatment as did Subjects 2 and 4 (60% and 50% gains,

respectively). Thus, this case, Subject 1, underscores the notion that our task in

recommending treatment is not as simple as pairing one deficit with one treatment.

Instead, a constellation of factors contributing to the subjects' ability to respond to a

treatment must be considered.






83

Subject 1 responded poorer than Subjects 2 and 4 to the semantic treatment. His

poorer response might suggest a link between the nature of the vision-specific lexical

level processing and how one responds to semantic information Some researchers

(Warrington & Shallice, 1984) have proposed that some stimuli (i.e., for living things)

are in large part crucially defined semanticallyy) by their sensory (e.g., visual) features.

For example, a zebra has stripes. Therefore, one might speculate that visual sensory

information is important for processing the meaning of some types of stimuli. In this

study the semantic cues purposefully avoided visual information as much as possible and

were delivered auditorily rather than visually (i.e., other than looking at the picture).

Subjects, such as Subject 1, who had an impaired visual-semantic route, might have some

degree of difficulty generating an internal image of the object that would diminish their

ability to retrieve the target word. A subject with this type ofunderspecified visually

dependent semantic coding might find the semantic cues used in this study that were

largely devoid of visually-coded information insufficient to boost the threshold necessary

for word retrieval. Conversely, subjects with an intact visual-semantic route should find

no disadvantage when responding to the semantic cues in this study (i.e., which avoided

visual information), and such was the case with Subjects 2 and 4. Related to this

explanation is an additional explanation that Subject l's semantic system may have been

partially underdeveloped prior to this stroke from a lifelong reading impairment, which is

also in part a visually coded system. It is unclear at this time whether or not a preexisting

reading disability (i.e., dyslexia) would impair the semantic system to the extent that all

visually coded incoming information would require an even greater threshold to be

reached for successful word retrieval after a stroke.








Subject 3 (PB), who had a deficit in the central point of convergence in lexical

processing, the semantic system, did not respond to either treatment. Some might

interpret this lack of response such that a semantic deficit yields a poor prognosis for

treatment. At this point this conclusion would be premature. Subject 3 is only a single

subject who had a semantic impairment and who did not respond to either treatment.

Additional research with such patients is needed to determine if a deficit in the semantic

system should be an exclusion criterion for this form of treatment.

In summary, differences in the nature of the deficits and spared processes in the

lexical semantic systems of the four experimental subjects in this study yielded different

responses to the treatments provided. The subjects with a deficit primarily in the

phonological output lexicon demonstrated a significant response to both types of

treatment. The subject (Subject 1) with a combination of deficits in the phonological

output lexicon and the visual/written input systems responded equally well to the

phonological treatment, when compared to others with the same phonological output

deficit. However, Subject 1 responded less dramatically to the semantic treatment when

compared again to the same subjects. Finally, the subject who had a deficit in the

semantic system failed to respond to either treatment.

Finally, additional factors should be considered that would require one to tailor

treatment protocols to individual needs in order to boost the potential yield of a treatment

program. One such variable that may negatively impact response to treatment is the

fatigue described by some of our subjects. Although both Subjects 2 and 4 benefited

from both treatments, Subject 2 (AS) reported more often that the two treatments were

"exhausting" than did Subject 4 (FS). Initially, Subject 2 described exhaustion after 30






85

minutes, even for confrontation naming of pictures during baseline testing as well as

during presession probe testing. However, at the time Subject 2 began the second

treatment, fatigue appeared to have substantially lessened. This increase in endurance

during the second treatment may have arisen in part because by the end of the first

treatment she was successfully naming many more of the 60 pictures correctly than at the

beginning of the first treatment when she initially struggled to name all 60 picture

stimuli. Thus, her gradual success in oral-pictures naming over the course of the first

treatment may have reduced her experience of fatigue for the second treatment. In

addition, individual sessions were shortened (beginning with session 6) by reducing the

length of time spent in the session for presession baseline probes, and she began to make

noticeable progress in the first (semantic) treatment. Shortening the amount of time at

the first of each session for baseline probes appeared to facilitate better concentration and

absorption during the treatment portion of each session.

Research Questions 1-b and 2-b:

If the semantic and/or the phonological treatments are effective, do the changes in

oral-naming performance also generalize to untrained stimuli?

Generalization

A minimal amount of generalization was observed with these two treatments.

Further research is needed to determine whether or not better generalization would be

observed if the treatments had been extended.

Research Questions 1-c and 2-c:

Is the change in oral-naming performance enduring two weeks after the semantic

and/or phonological treatment?








Maintenance

A high level of maintenance was observed for two weeks after the end of both the

phonological and semantic treatments with the effects from the semantic treatment only

slightly less well maintained. This relatively high level of maintenance after both

treatments was attributed to the fact that the phonological and semantic treatments were

sufficiently effective such that once the subjects acquired a word, he or she was usually

able to retrieve it in a generally consistent manner.

Research Questions 3-a, 3-b. and 3-c:

How do the semantic and phonological treatments compare on degree of change,

generalization, and maintenance?

Comparisons of oral naming, generalization, and maintenance

While both treatments produced an improvement in oral-naming performance, the

highest degree of change as defined by highest percent improvement by an individual

subject was more robust in the semantic treatment (60%) than the phonological treatment

(40%). We and other researchers (Howard, Patterson, Franklin, Orchard-Lisle, and

Morton, 1985b) attributed the greater degree of improvement with the semantic treatment

to facilitation of lexical processes thought to occur at the point of convergence in lexical

processing in the semantic system. However, although an advantage in oral-naming

performance (i.e., higher percent change) for trained nouns was found with the semantic

treatment in two of the three subjects, an advantage in oral-naming performance was

found in one subject with the phonological treatment.

Generalization to the untrained generalization word set with the phonological

treatment showed some improvement but was less than dramatic. In contrast,








generalization to the untrained generalization word set with the semantic treatment was

not observed. It is unclear why some degree of generalization in these word sets was

observed with the phonological and not with the semantic treatment.

A high level of maintenance was observed from both the phonological and semantic

treatments with a slightly diminished degree of maintenance observed with the semantic

treatment. However, the period of observation for maintenance should be longer than

two weeks. Further research is indicated to determine if there is a functional impact

beyond the treatment measures.

Clinical Implications

The most significant clinical outcome from this study focuses on treatment efficacy

and outcome. Although there are selected exceptions, this study provides evidence that

language treatment is efficacious for aphasic patients, including those who may be

several years postonset of CVA. Because of the strong response of most of these subjects

to treatment, it is possible that additional treatment may have furthered their

improvement. As it were, reports from the subjects and their families indicated that

lifestyle changes were made as a result of treatment. One subject expanded her social

circle by participating in her community stroke support group, joining a Bible study from

her church, and going to lunch with friends. Her physician also commented on her

improved communication skills as she began to verbally participate more in her office

visits rather than relying on her daughter to report medical information for her. All of

these new experiences served to improve the patient's self-esteem, reduce the feelings of

isolation and dependency that she had been experiencing, and decrease demands on her

family members. Another subject, who has been disabled and unemployed, stated that he








was going to renew his commercial fisherman's license and try to start working again,

now that he was talking better. He also began to handle more of his business and medical

affairs independently of his family, all of which required him to talk to strangers to

explain what he needed. He stated that his feelings of accomplishment and self-esteem

had increased as a result of his improved communication. Finally, he began to socialize

more by visiting family members who lived out-of-state and by exchanging football

rivalry talk with the examiner during treatment.

In this study the researchers chose a limited number of treatment sessions to more

closely parallel current practice in speech-language pathology, which have only a limited

number of sessions to show results. This study provides evidence that language therapy

for aphasia is efficacious. In fact, this study provides evidence that many patients may

benefit from long-term aphasia treatment. In addition, this study supports the view that

language intervention can be beneficial in some patients years after their stroke.

Conclusions

The findings of this study have major implications for treatment intervention. First,

elements of both semantic and phonological cues should generally be administered to

most aphasic subjects for the best therapeutic benefit. In some subjects the semantic

treatment yielded a better performance in oral naming of picturable nouns. However, the

phonological treatment did provide measurable and clinically substantive yield as well.

The semantic treatment yielded better results in two subjects with a deficit primarily in

the phonological output lexicon but who had a relatively preserved semantic system. One

subject had a relatively spared semantic system and phonological input lexicon, and he

responded significantly better to the phonological treatment. Finally, one subject with an






89

impaired semantic system failed to respond to either treatment. Thus, better results of

treatment might be obtained by examining not only impaired but also spared functions in

each case and matching these to treatment.

Future Research

The question arises as to how subjects respond to a treatment that combined features

of both the semantic and phonological treatment. More research is needed to find the

most efficacious treatments that yield the best results in the shortest amount of time. In

addition more subjects with a variety of spared lexical processes is needed to learn which

of these factors may influence a successful outcome in treatment and to potentially

predict which type(s) of treatments may be most efficacious for which patients.

Several questions remain unanswered from these data. The first question is whether

or not additional subjects (such as Subject 1) with a relatively spared phonological input

lexicon and semantic system but impaired phonological output system would also

respond better to the phonological treatment than the semantic treatment. Subject 1

(AW) had a preexisting developmental learning disability-dyslexia. His auditory

system may have been more strongly developed and widely distributed (than a subject

without a developmental learning problem) prior to his stroke after a lifetime of

compensating for his developmental reading impairment, which may have made him

more receptive to processing phonological cues. It is beyond the scope of these data to

determine whether or not all subjects with a relatively spared auditory system would

respond to the phonological treatment.

Finally, new medical research is making headway with pharmacological agents that

appear to open new thresholds for plasticity in the adult brain, which has long been






90

thought to become hard-wired with puberty. With the advent of these drugs that

potentially make the nervous system more plastic and thus receptive to language

intervention, a combination of the two treatments may become the treatment of choice for

the twenty-first century. Refined fMRI capabilities may enable researchers to pretest and

posttest areas of the brain that may respond to this combination of treatments.