1137 1465Raymer Anastasia 072052 Nov08 2007 Time 023536PM

MossTalk Training for Word Retrieval Across Semantic Categories
Tiffany Johnson, Erin Todd, & Anastasia Raymer*
Old Dominion University, Norfolk VA; *Brain Rehabilitation Research Center, Gainesville, FL
ABSTRACT
Few studies have examined the ability of patients to work
independently with computer training programs to improve
word retrieval in aphasia. In a single-participant design, we
investigated effects of independent computerized training
with MossTalk multi-mode matching exercises in two
individuals with severe word retrieval impairments. We
examined effects of treatment for trained words and
untrained words from the trained semantic categories and
untrained categories. Both participants improved in picture
naming for trained words. One showed limited
generalization to untrained words and gains in standardized
testing. Improvements were smaller than those reported for
patients in a prior clinician-assisted MossTalk study.

INTRODUCTION
•Large word retrieval treatment literature reports largely
training specific effects with little generalization to untrained

words (Nickels, 2002)
•Some generalization observed when training within
semantic categories; i.e. training of some category members
generalizes to improvement in naming other members of the
same semantic category (Spencer et al., 2000; Kiran &
Thompson, 2004)
•Training techniques are needed that patients can implement
independently to increase opportunity to practice functional
vocabulary
•One means of supporting independent practice is through
the use of computers
•Much training software available for purchase; only limited
data on usefulness
•MossTalk Words (Fink et al., 2001)
•experimental computer program for word retrieval training
•Fink et al (2002) showed effectiveness of MossTalk Cueing
Hierarchy Training module with partial clinician assistance
•Raymer et al (2006) showed effectiveness of MossTalk
Multimodality Matching Exercises with clinician assistance
- patterned after studies of comprehension training for word

retrieval in aphasia (Pring et al., 1990; Nickels & Best, 1996)
-Effects were greatest when training took place 3-4 times per
week as compared to 1-2 times per week

PURPOSE
•To explore usefulness of MossTalk Words Multimode
Matching exercises administered independent of a
clinician
•To examine generalization of treatment effects within
and across semantic categories
Acknowledgments:
Appreciation to Ruth Fink, of the Albert Einstein
Healthcare Network, who generously provided a copy of
MossTalk Words. This study was supported in part by the
Department of Veterns Affairs Office of Rehabilitation
Research and Development.

METHODS

RESULTS


Participants: 2 right handed individuals with stroke
leading to right hemiparesis, aphasia, apraxia
P1
P2
Age:

62 yrs

52 yrs

Gender:

M

F

Education:

9 yrs


17 yrs

Time post CVA:

3 yrs

4 1/2 yrs

Aphasia Classification Broca

Broca

Lexical Testing (Zingeser & Berndt, 1990)
Pic Naming Nouns 13/60
Wd/Pic Verif Nouns 36/60
Breakdown:
semantic

39/60

50/60
mild semantic

Single Subject Treatment Design:
Daily Probe Task/Stimuli:
Name to picture confrontation
Stimuli: 50 nouns 4 x 6 colored pictures
from 4 semantic categories: clothing, vegetables,
toiletries, kitchen items
Tx Set 1: n=16 - 4 items from 4 categories
Tx Set 2: n=16 - 4 items from 4 categories
Generalization set: n=8 – 2 items from 4 categories
Control set: 10 items from unrelated semantic categories
(tools, furniture) matched for frequency
Correct responses: recognizable spoken word allowing for
articulatory distortions.
Dependent variable: # correct
Procedure: Single participant treatment design
3-4 Baseline Sessions
Training: 5 days per week for 2 weeks

Probe sessions: 1-2 times per week
Treatment competed after 2 weeks or 2 sessions at 90%
accuracy
Treatment Protocol: MossTalk Words
Multimode Matching Exercises (Fink et al., 2001)
Independent practice daily
1) Spoken + Written word/picture matching
- computer says word and presents written word;
- pt. touches correct picture of 4 related choices

Picture Naming Effect Sizes
Phase 1
P1
Trained 1
4.0
Trained 2 (untrained) -.57
Generalization
.14
Control
.25

Phase 2
Trained 1 (maint) 1.07
Trained 2
1.01
Generalization
.65
Control
.48

P2
3.0*
1.19
1.29
.86

Western Aphasia Battery:
P1 Pre
Fluency (max 10)
2
Comprehension (max 10) 4.5 5.2

Repetition (max 10)
2.2
Naming (max 10)
2.3
Total AQ (max 100)
35.9
Boston Naming Test (max 60)

2

Post
3

P2 Pre
3
5.7
7.4
1.2
3.4
2.2

5.1
33.2
36.4

Post
0

5

12

8

1.2
4.8
48.8

.46
2.97
1.36

2.16

DISCUSSION
Naming Training (Figures 1 & 2):
• Both participants demonstrated modest, but
significant naming improvements, greater for P2
than for P1; P2 replicated the effect in a second
set of training items
• Very little generalized improvements in untrained
items from same semantic categories and from
different semantic categories (control set)
• Improvements in this independent study not as
strong as earlier study (Raymer et al., 2006) using
MossTalk Words with clinician assistance
Standardized Testing Results
• Both participants improved slightly on BNT, and
P2 improved also on WAB – primarily in
comprehension tasks

Why no generalization to untrained items in same

semantic category?
-Methodologic differences in our study versus others
-Although Spencer et al (2000) reported generalization of
naming training within a semantic category, they trained
only one category at a time for many, many sessions
-Kiran & Thompson (2002) reported generalization to
untrained category members when training atypical
category items, but not when training typical category
members
-Within constraints of MossTalk Words, we trained items
that were fairly typical of the given categories, and we
trained 4 categories at one time
-In order to promote generalized training effects, suggests
need to train one category at a time for many sessions;
possibly need to train atypical items

- computer says word

• Improvement in comprehension compatible with
training protocol in which auditory
comprehension critical component

- pt. touches correct picture of 4 related choices

Alternative Explanations for Results

Fink, Brecher, Montgomery, & Schwartz (2001). Moss Talk Words. Philadelphia: Albert
Einstein Healthcare Network.

• Spontaneous Recovery or Repeated Exposure?

Kiran, S., & Thompson, C.K. (2002). Journal of Speech-Language-Hearing Research,
46, 608-622.

- computer presents written word

• Both subjects many years post onset of aphasia

Nickels & Best (1996).Aphasiology, 10, 109-136.

- pt. touches correct picture of 4 related choices

• Effects for training words surpassed effects for
untrained words, suggesting repeated exposure
may have contributed to small generalized
improvements, but training effects far surpassed
simple practice.

2) Spoken word/picture matching

3) Written word/picture matching

Participant then attempted to say name of target picture.

References
Fink, Brecher, & Schwartz. (2002). Aphasiology, 16, 1061- 1086.

Nickels, L. (2002). Therapy for naming disorders: Revisiting, revising, and reviewing.
Aphasiology, 16, 935-979.
Pring, White-Thomson, Pound, Marshall, & Davis (1990). Aphasiology, 4, 479-483.
Raymer, A.M., Kohen, F., & Saffell, D. (2006). Aphasiology, 20, 257-268.
Spencer, K.A., Doyle, P.J., McNeil, M.R., Wambaugh, J.L., Park, G., & Carroll, B.
(2000). Aphasiology, 14, 567-584.
Zingeser, L.B., & Berndt, R.S. (1990). Brain and Language, 39, 14-32.