J Am Acad Audiol 12 : 254-260 (2001)
Performance on a Spanish
Picture-Identification Task
Using a Multimedia Format
June Antablin McCullough*
Richard H. Wilson'
Abstract
An option for estimating the word-recognition performance of patients who
do not speak
English as a first language involves using auditory materials, presented in the
patient's native
language, in conjunction with a closed-set response mode incorporating pictures
or written
words. The advantage of this auditory/visual paradigm is that the audiologist
is not required
to know the foreign language and is therefore not required to judge the
accuracy of an oral
response to speech stimuli in a foreign language . Spanish auditory/visual materials, known
as the Spanish Picture-Identification Task, were developed to be used
in a computerdriven multimedia administration and scoring format . Performance data,
both in open(word-recognition) and closed-set (word-identification) response modes, were established
for the Spanish Picture-Identification Task using subjects whose first language
was Spanish .
The results from the open-set paradigm indicate that the Spanish
Picture-Identification Task
word lists are essentially equivalent to conventional Spanish and English
materials used
for word recognition . Findings from the closed-set conditions indicate that the
Spanish PictureIdentification Task materials are appropriate for estimating the word-identification abilities
of Spanish-speaking adult listeners.
Key Words: Computer audiometry, multilingual audiometry, speech identification, speech
recognition
emographics suggest that audiologists
are likely to encounter the challenge of
D evaluating the speech-recognition abilities of patients who have a limited knowledge
of English (U .S . Bureau of the Census, 1990).
The increasingly multilingual patient population has produced a demand for speech audiometric materials that are appropriate for
non-native English speakers . One set of professionally recorded speech-recognition materials intended for Spanish speakers was
described by Weisleder and Hodgson (1989) .
The materials (known as Spanish Bisyllables
and available commercially from Auditec of St .
*Communication Disorders and Sciences Program,
San Jose State University, San Jose, California, 'Auditory
Research Laboratory, James H . Quillen VA Medical Center,
Mountain Home, Tennessee, and Departments of Surgery
and Communicative Disorders, East Tennessee State
University, Johnson City, Tennessee
Reprint requests : Richard H. Wilson, VA Medical
Center, Mountain Home, TN 37684
254
Louis) consist of four 50-word lists of bisyllabic
consonant-vowel-consonant-vowel Spanish
words in a conventional, oral, open-set response
paradigm . Bisyllabic words were used as stimuli because monosyllabic words in Spanish are
rare . Normative data using native Spanish
speakers as subjects with oral responses scored
by a Spanish-speaking audiologist revealed that
the Spanish Bisyllables produced psychometric
functions similar to the functions for English
monosyllabic words. The similarity between
functions must be viewed as coincidental as
both the language (word lists) and speakers of
the words were different. The conventional oral,
word-recognition (open-set) paradigm, however,
is limited because audiologists who do not speak
Spanish are uneasy or unwilling to score an
oral response in a foreign language . This problem is compounded when other foreign languages are encountered .
An option that is uniquely suited for estimating the word-recognition performance of
non-native English speakers involves auditory/visual materials in a word-identification
Performance on a Spanish Picture-Identification Task/McCullough and Wilson
(closed-set) paradigm . Auditory/visual materials consist of an auditory-stimulus portion (target words) in the patient's language and a visual
response foil containing written words or pictures, one of which corresponds to the target
word . The task of the patient is to listen to the
target word and to point to the word or picture
that represents it . The advantage of the pointing-response paradigm is that the audiologist
can score the test simply by noting whether the
appropriate word or picture was indicated instead
of judging the accuracy of an oral response in a
foreign language . Thus, no knowledge of a foreign
language is required by the audiologist . Spitzer
(1980) and Comstock and Martin (1984) have
demonstrated the utility of using Spanish auditory/visual speech materials, scored by Englishspeaking audiologists, in the clinic with patients
whose first language is Spanish .
The Picture-Identification Task, developed
for nonverbal adult patients (Wilson and
Antablin, 1980) and used with fourth-grade children (McCullough et al, 1992), is an English
example of an auditory/visual word-identification
test paradigm . In the project reported here, the
Picture-Identification Task paradigm was
extended to Spanish target words and corresponding pictures . The Spanish materials, known
as the Spanish Picture-Identification Task, are
presented in a computer-driven multimedia format that is a technological advance from the
once-used picture cards or notebooks (McCullough et al, 1995) . In the multimedia format,
auditory target words in Spanish are initiated
by the audiologist using a menu on a computer
monitor in the control room . The target word is
presented through earphones to the patient,
who is seated in front of a second monitor on
which four pictures corresponding to the target
word and three rhyming alternatives are displayed in a quadrant arrangement . Following the
auditory presentation of the stimulus word, the
patient responds to one quadrant of the visual
response foil by using a point-and-click strategy
or by touching the screen of a touchscreen monitor. In this way, the identification test is administered and scored automatically using a simple
language-to-place transformation on a computer
monitor. Recently, the auditory/visual paradigm
was extended to Russian materials, and normative performance data from adults with normal hearing were reported (Aleksandrovsky et
al, 1998). The current investigation describes two
experiments in which normative performance
data for the Spanish Picture-Identification Task
were obtained using adult listeners, both in
open-set (word-recognition) and pointing, closedset (word identification) response paradigms,
using listeners whose first language was Spanish . In experiment 1, psychometric functions for
the Spanish Picture-Identification Task word
lists (open set) were generated and compared
with functions from conventional English and
Spanish materials . In experiment 2, the psychometric characteristics of the Spanish PictureIdentification Task (closed set), administered in
the multimedia format, were studied .
METHOD
he development of the Spanish PictureT Identification Task has been detailed in a
previous report (McCullough et al, 1995) . Briefly,
the test vocabulary consists of common, familiar, bisyllabic Spanish nouns and verbs that are
easily pictured and that have three rhyming
alternatives of a similar description. A total of
100 vocabulary items, distributed into 25
4-word groups, were compiled . Two 50-word lists
(lists 1 and 2, Table 1) were constructed by
assigning 2 words from each 4-word group as tar-
get words for list 1 and the remaining 2 items
as target words for list 2 . The assignment of
target words to the test lists was done to ensure
a balanced distribution of initial phonemes
among the lists . The target words and carrier
phrase, spoken by a female talker (country of origin : Mexico), were recorded and edited digitally
to establish the auditory-stimulus portion of
the materials . The artist's renditions of the
vocabulary items, arranged in quadrants on 25
color pictured response foils, completed the
visual-response portion of the materials .
Experiment 1
Procedures
The goal of experiment 1 was to describe the
normative performance of adult listeners whose
first language was Spanish in open-set conditions
on the Spanish Picture-Identification Task . This
was the traditional auditory-stimulus, oral
response recognition paradigm . A second goal
was to compare the psychometric function of
the Spanish Picture-Identification Task materials with the psychometric function of the Spanish Bisyllables obtained by Weisleder and
Hodgson (1989) . To minimize talker differences
between the two sets of materials, lists 1 and 2
of the Spanish Bisyllables were recorded digitally
using the same female Spanish talker who spoke
255
Journal of the American Academy of Audiology/Volume 12, Number 5, May 2001
Table l 100 Target Words Comprising the
Spanish Picture-Identification Task
List 1
balon (balloon)
barba (beard)
barca (boat)
besa (kiss)
boca (mouth)
bola (ball)
bota (boot)
caja (box)
canta (sing)
capa (cape)
cara (face)
carne (meat)
carcel (jail)
coger (catch)
cono (cone)
correr (run)
foto (photo)
gorro (cap)
hueso(bone)
jota (J)
ladron (robber)
llama (knock)
manta (blanket)
masa (dough)
misa (mass)
mono (monkey)
nina (girl)
ojo (eye)
oso (bear)
pala (shovel)
papa (potato)
pico (sting)
pito (whistle)
prisa (hurry)
queso(cheese)
rama (twig)
raton (rat)
reza (pray)
roja (red)
ropa (rope)
rota (broken f .)
sala (living room)
salto (jump)
Santo (Saint m .)
talon (heel)
tasa (cup)
toca (knock)
toro (bull)
vina (vine)
voto (vote)
List 2
ala (wing)
balcon (balcony)
barra (bar)
baston (cane)
bata (robe)
beso (kiss)
bola (ball)
bolsa (purse)
cabra (goat)
cama (bed)
cana (cane)
carga (load)
carta (letter)
casa(house)
coca (Coke)
comer (eat)
coser (sew)
dama (lady)
fresa (strawberry)
halcon (hawk)
jamon (ham)
Ilanta (tire)
Iloro (cry)
mala (sick f .)
mapa (map)
mesa (table)
moto (motorcycle)
nota (note)
ocho (eight)
oro (gold)
peso (money)
pino (pine tree)
pina (pineapple)
piso (floor)
plaza (plaza)
rina (fight)
risa (laugh)
roca (rock)
ronca (snore)
rosa (rose)
roto (broken m .)
saco (sack)
sapo (frog)
Santa (Saint m .)
tapa (lid)
tisa (chalk)
tono (note)
trono (throne)
vota (vote)
zorro (fox)
the Spanish Picture-Identification Task materials. Both recordings were made in a sound
booth with a microphone (AKG Acoustics, Model
C 460 B), a preamplifier (Symetrix, Model
SX202), and a 16-bit analog-to-digital converter
(Antex, Model SX-10) that sampled 20,000
points/sec . The carrier phrase, "Diga Usted
"
256
("Say") was recorded to precede the target
words. The digitized files of both word lists were
output by the computer and recorded onto digital audiotape (Sony, Model PCM 2500A,B) .
Subjects in experiment 1 were 16 native
Spanish speakers (mean age = 27 years) from a
variety of countries and Hispanic cultures
(Columbia, El Salvador, Ecuador, Guatemala,
Mexico, Peru, Puerto Rico, Spain, and the United
States). All subjects exhibited normal auditory
sensitivity (515 dB HL re : ANSI, 1989) at octave
intervals from 250 to 8000 Hz and had negative
histories of middle-ear pathology.
Two 50-word lists (Spanish Picture-Identification Task and Spanish Bisyllables, in alternating order of presentation) were presented
monaurally to each listener at each of five levels (24, 30, 36, 42, and 48 dB SPL) . To minimize
learning effects, the materials were presented
in an ascending order. Thus, each subject listened
to 10 lists within a 1-hour session. The listeners were instructed to write down the word that
was heard and were encouraged to guess (Nelson and Chaiklin, 1970) . List order (1 or 2) and
test ear were counterbalanced among subjects .
The speech materials were reproduced on a digital audiotape player (Sony, Model DTC-670)
and routed through an audiometer (GSI, Model
16) into the test earphone (TDH-50P). All testing was done in a sound-treated room .
Results and Discussion
Mean percent correct performance and standard deviations for the Spanish Picture-Identification Task and the Spanish Bisyllables in the
open-set response mode at each of five presentation levels are listed in Table 2. Mean performance on lists 1 and 2 within each set of
materials was similar (within 6%). Thus, the
mean data from both lists were combined to
produce an overall group mean for the Spanish
Picture-Identification Task and an overall group
mean for the Spanish Bisyllables at each level.
Group mean performance ranged from 11 .6 percent (24 dB SPL) to 85 .0 percent (48 dB SPL) for
the Spanish Picture-Identification Task and
from 10 .6 percent (24 dB SPL) to 88 .6 percent
(48 dB SPL) for the Spanish Bisyllables . Standard deviations ranged from approximately 6
percent to 12 percent at each level.
Figure 1 illustrates the psychometric functions for the Spanish Picture-Identification Task
(open squares) and for the Spanish Bisyllables
(open triangles) that were developed in experiment 1. For comparison, the data for the Spanish
Performance on a Spanish Picture-Identification Task/McCullough and Wilson
Table 2 Mean and SD Percent Correct Recognition (N = 16) for Lists 1 and 2 of the
Spanish Picture-Identification Task and for Lists 1 and 2 of the Spanish Bisyllables
at each of Five Presentation Levels (dB SPL)
30
36
42
48
35 .2
60 .0
78 .5
85 .2
List 2
Mean
8 .5
6 .5
14 .7
33 .2
62 .0
77 .2
84 .7
Mean
11 .6
34 .2
61 .0
77 .6
85 .0
List 1
Mean
10 .7
41 .5
65 .6
78 .2
87 .0
Mean
10 .5
38 .5
72 .5
79 .0
90 .2
10 .6
6 .8
40 .0
69 .0
78 .6
88 .6
24
Spanish Picture-Identification Task (open set)
List 1
Mean
SD
SD
Lists 1 and 2 Combined
SD
8 .8
8 .2
10 .9
10 .3
7.9
11 .8
11 .5
10 .8
4 .8
5 .8
5 .0
8 .1
9 .7
4 .7
Spanish Bisyllables (open set)
SD
List 2
SD
Lists 1 and 2 Combined
Mean
SD
6 .8
7 .3
6 .1
12 .9
8 .8
13 .4
Bisyllables (lists 1 and 2) reported by Weisleder
and Hodgson (1989) are shown as filled triangles .
To describe the performances, the data were fit
with third-degree polynomials from which the
slopes of the functions were calculated between
the 20 percent and 80 percent correct points . The
slopes of the functions for the Spanish PictureIdentification Task and for the Spanish Bisyllable materials, spoken by the same female
speaker, were 3 .7 percent/dB and 3 .8 percent/dB,
respectively. The slope of the function calculated for lists 1 and 2 of the Spanish Bisyllables
reported by Weisleder and Hodgson (1989) was
only slightly steeper, 4 .2 percent/dB . The differences between the functions were slight, and
all functions approximated good word-recognition performance (above 80 percent) at the highest level . At the 50 percent correct point, the
three functions differed by 2 .3 dB with slopes at
the 50 percent correct point that ranged from 4 .5
(Weisleder and Hodgson, 1989) to 4 .0 percent/dB
(Spanish Picture-Identification Task) . At the 50
percent correct point, then, the Spanish Picture-Identification Task materials are minimally more difficult than the Spanish Bisyllables,
33 .8 and 31 .8 dB SPL, respectively. This relationship between the two sets of materials spoken by the same female speaker may be related
to the designed inclusion of rhyming cognates on
the Spanish Picture-Identification Task, which
increased the pool of potential responses in the
5 .8
7 .5
8 .1
12 .8
5 .8
6 .4
5 .8
6 .7
mental lexicon activated by an acoustic/phonetic sequence . Thus, the probability of confusions increased. This line of reasoning is
consistent with the neighborhood frequency and
neighborhood density components of the Neighborhood Activation Model of spoken language
(Luce and Pisoni, 1998).
100
60
0
0 60
° 40
r
20
0
20
25
30
35
40
45
50
55
60
Presentation Level (d8 SPL)
Figure 1
Mean percent correct recognition performance (open set, N = 16) as a function of presentation level
for the Spanish Picture-Identification Task (squares, y =
59 .58 - 10 .4972x + 0.4778x2 - 0.0052x3, r2= .99), the
Spanish Bisyllables (open triangles, y = -213 .7829 +
12 .7198x - 0.1488x2 + 0.0003x', r2 = .00), and the
Weisleder and Hodgson (1989) data from the Spanish
Bisyllables, lists 1 and 2 (filled triangles, y = -58.6765 1 .46616x + 0.2354x2 - 0.0029x3, r2 = .99) .
257
Journal of the American Academy of Audiology/Volume 12, Number 5, May 2001
The results of experiment 1 indicate that the
Spanish Picture-Identification Task, which consists of common, bisyllabic, phonetically diverse,
rhyming target words, are essentially equivalent to conventional Spanish and English wordrecognition materials when presented in an
open-set paradigm . The psychometric characteristics of the Spanish Picture-Identification
Task in the picture-pointing, closed-set response
mode were the topic of experiment 2 .
Experiment 2
Procedures
The goal of experiment 2 was to obtain normative data for the Spanish Picture-Identification
Task in the picture-pointing, closed-set paradigm . Ten listeners (mean age = 25 years) whose
first language was Spanish served as subjects . Of
the 10 listeners, 8 had participated in experiment 1 and 2 were new to the protocol . All subjects met the criteria for normal hearing
sensitivity defined in experiment 1.
Color pictures representing the visualresponse portion were scanned and the auditory
stimuli were transferred from digital audiotape to
the Macintosh format. The test administration software using a computer-driven format for multimedia presentation is described elsewhere
(McCullough et al, 1994). In experiment 2, a Macintosh II series computer communicated with two
color monitors, one in the control room and one in
the test room. The sound output from the computer
was delivered through an audiometer (GSI, Model
16) into the test earphone (TDH-50P).
The subjects were tested in a double-walled,
sound-treated room . In the control room, the
experimenter selected the target word from a
menu on the control monitor. Once the target
word was initiated, a series of computercontrolled events occurred . First, the response
foil appeared on the monitor located in the test
room with the listener. Concurrently, a window
containing the response foil appeared in the
upper left-hand corner of the control monitor for
verification and scoring purposes . Second, the
carrier phrase ("Ensename
" ["Show me
"]) and target word were presented to the
test ear. Third, the subject responded to the
presentation by pointing to the picture in the
quadrant that represented the target word . The
subjects were instructed to respond following
each target word presentation . The experimenter
observed the response through the window of the
test chamber and scored the response by point258
ing and clicking with a computer mouse on the
control monitor. (Our preference was to use a
touchscreen monitor with the subjects, but this
technology was not available to us .) All subjects
listened to one 50-word list (list 1 or 2, in alternating order among subjects) presented at seven
ascending increments (14, 20, 26, 32, 38, 44,
and 50 dB SPL) .
Results and Discussion
The mean percent correct performance data
(and standard deviations) for the Spanish
Picture-Identification Task in picture-pointing,
closed-set paradigm are given in Table 3. Mean
performance on lists 1 and 2 was similar at all
levels (within 6%) with the exception of the 20
dB SPL condition. At this level, mean performance on list 1 was 17 percent poorer than for
list 2. Standard deviations, ranging from 3 to 10
percent, also were large at this level. The differences between the means and the wider range
of performance on lists 1 and 2 at 20 dB SPL
reflected both the larger variability associated
with word-identification performance near
threshold and the small number of datum points
per list. Since performance on lists 1 and 2 at the
remaining six levels was similar, the mean data
from both lists were combined to produce an
overall group mean for the Spanish PictureIdentification Task in the closed-set response
mode . Group mean performance was 32 .4 percent at 14 dB SPL, which is close to chance performance on a four-interval forced-choice task,
and increased to 95 .8 percent at 50 dB SPL.
The group mean data were used to generate the psychometric function for the Spanish
Picture-Identification Task in the picture-pointing, closed-set response mode, shown in Figure
2 as filled squares. The psychometric function
was fit with a third-degree polynomial to calculate the slope of the function . On the portion
of the function representing 20 to 80 percent
correct performance, the slope was gradual
(1 .8%/dB). In fact, performance improved by
approximately 15 percent between each presentation level from 14 to 32 dB SPL. Beyond
this level, performance reached a gradual
plateau, with best performance of 95 .8 percent
occurring at the highest level (50 dB SPL) . As
can be seen in Figure 2, performance on the picture-pointing task was considerably better
through the performance range of presentation
levels than performance on the open-set materials (open squares), which were replotted from
Figure 1. At the 30, 50, and 70 percent correct
Performance on a Spanish Picture-Identification Task/McCullough and Wilson
Table 3
Mean and SD Percent Correct Identification (N - 10) for Lists 1 and 2 of the
Spanish Picture-Identification Task, Picture-Pointing Response Mode
Presentation Level (dB SPL)
List 1
Mean
SD
List 2
Mean
SD
Lists 1 and 2 Combined
Mean
SD
14
20
26
32
38
44
50
34 .8
10 .2
39 .6
7.7
67 .2
5.2
74 .8
10 .0
87 .6
6.2
92 .8
2.7
94 .8
3.0
30 .0
8 .5
57 .2
8 .7
61 .6
8 .0
80 .8
85 .2
91 .6
96 .8
32 .4
48 .4
64 .4
77 .8
86 .4
92 .2
95 .8
9 .2
12 .1
points, the functions for the two response modes
were separated by 6 .8, 13 .5, and 15 .7 dB, respectively. As is obvious from the data in Figure 2,
the slopes of the functions differed substantially. At the 50 percent correct points, slopes of
2 .7 and 4 .0 percent/dB were observed for the
closed and open sets, respectively. The relationship between functions generated from
closed- and open-set versions of the Spanish
Picture-Identification Task is consistent with
data from the English Picture-Identification
Task, in which the closed- and open-set functions
were similarly displaced by 9 to 12 dB (Wilson
and Antablin, 1982) .
Two factors account for the differences
between the functions for the open- and closed-set
materials . First, throughout the range of presentation levels, uncertainty is higher for an open- (rel-
atively infinite response choices) than for a
closed-set paradigm (in the present study, only
four response choices) . In its purest form, the
uncertainty factor is reflected in the minimal
differences between the open- and closed-set
functions at the highest presentation levels . Second, with the closed-set paradigm, there is a 25
percent chance that simply by guessing, the listener will respond correctly throughout the range
of presentation levels . That is, if the listener did
not hear any stimuli and responded to the closedset materials, performance owing to chance would
be 25 percent correct . In all probability, this
"guessing factor" is minimal at the highest performance (and presentation) levels . At the high
levels, guessing may be involved only in instances
in which there is "noise" in the data, as when the
listener momentarily "tunes out" to the task at
hand . At the lowest presentation levels, however, guessing is the dominant factor in the
closed-set paradigm as represented by the -30
4 .8
7 .0
6 .9
8 .1
6 .3
4 .6
3 .0
3 .6
3 .0
percent correct performance at the lowest presentation level.
An error analysis was performed to determine if certain target words or groups of words
were more difficult to identify than others . The
analysis was conducted by examining individual
subject data from the 32, 38, 44, and 50 dB SPL
presentation levels, at which mean performance
ranged from 78 to 96 percent correct . The analysis revealed that a majority of the target words
were identified correctly at 32 dB SPL and above
since only 19 of the 100 target words were missed
more than 10 percent of the time . Of the 19
words, 9 words were missed between 10 and 20
percent of the time, and 10 words were missed
between 20 and 30 percent of the time . The 10
most commonly missed words were "zorro" (fox),
100
80
60
40
20
0
15
20
25
30
35
40
45
Presentation Level (dB SPL)
50
55
Figure 2 Mean percent correct identification/recognition performance as a function of presentation level
for the Spanish Picture-Identification Task (N = 10),
closed-set response mode (filled squares, y = -14.1002
+ 3.5089x - 0.019x2 - 0.0003x', r2 = .99), and the Spanish Picture-Identification Task, open-set response mode
(open squares), from experiment 1
Journal of the American Academy of Audiology/Volume 12, Number 5, May 2001
"gorro" (cap), "prisa" (hurry), "peso" (dollar),
"beso" (kiss), "besa" (kiss), "rama" (twig), "tono"
(note), "rina" (fight), and "manta" (blanket).
Four of the 10 most commonly missed target
words included two pairs of rhyming words that
appear on the same response foil, suggesting that
these words were confused with each other. One
of the pairs of target words, "peso" and "beso,"
differs in the voiced/voiceless feature in the initial consonant position . This finding suggests
that voicing may be difficult to distinguish in an
identification task involving isolated rhyming
words, without benefit of contextual cues .
One pair of commonly missed target words,
"zorro" and "gorro," did not share similar phonetic features such as voicing or place of articulation. Confusion between these words appears
to be related to differences in familiarity of certain vocabulary items among subjects of different Hispanic cultures . Informal questioning at
the close of the test sessions revealed that some
of the subjects whose country of origin was Mexico did not commonly use the words "rina" (fight),
"manta" (blanket), or "zorro" (fox). Moreover,
some of the subjects not of Mexican origin did
not commonly use the words "gorro" (cap) or
"rama" (twig) . Interestingly, many of the subjects
commented that, although these target words
were not the most common words associated
with the pictured response in their dialect or culture, they were able to discern that the target
word corresponded to the appropriate picture
when the alternative pictures were surveyed .
This finding indicates that adults, unlike children, appear to be fairly resilient to differences
in receptive vocabulary.
Of the 10 commonly missed target words, 5
appeared on list 1 and 5 appeared on list 2. The
even distribution of difficult target words among
the test lists and the excellent word-identification
performance (96%) at 50 dB SPL suggest that
overall performance is not significantly affected by
the inclusion of a few difficult items on each list .
The results from experiment 2 indicate that
the Spanish Picture-Identification Task materials are appropriate for administering to adult listeners who speak Spanish as their first language
for the purposes of establishing word-identification performance. The test may be administered and scored by audiologists who are
unfamiliar with Spanish. The computer-driven
multimedia platform continues to be refined to
include automatic test administration and scoring and touchscreen technology. Moreover, ageappropriate materials for Spanish-speaking
children currently are being developed. Finally,
260
the development of test materials in other common languages is ongoing to ensure that increasing numbers of multilingual patients may benefit
from the auditory/visual word-identification paradigm used as a component of their hearing
evaluation.
Acknowledgment . The authors gratefully acknowledge
Mr. Jon Birck of Virtual Corporation for designing the
computer software used in this investigation. This study
was funded in part by the California State University
Research Funds Committee. Laboratory facilities for the
recordings were provided by the Medical Research Service
and the Rehabilitation, Research and Development
Service, Department of Veterans Affairs, at the VA Medical
Center, Long Beach, CA . The two lists of 50 words from
the Spanish Picture-Identification Task materials are
available in a recognition format on compact disc using
the carrier phrase "Diga Usted -" ("Say -'). For information, contact [email protected].
REFERENCES
Aleksandrovsky I, McCullough JA, Wilson RH . (1998) .
Development of suprathreshold word recognition test for
Russian-speaking patients . JAmAcad Audiol 9:417-425 .
American National Standards Institute . (1989) .
Specifications for Audiometers. (ANSI S3 .6-1989) . New
York : ANSI.
Comstock CL, Martin FN . (1984) . A children's Spanish
word discrimination test for non-Spanish-speaking clinicians. Ear Hear 5:166-170 .
Luce PA, Pisoni DB . (1998) . Recognizing spoken words:
the neighborhood activation model. Ear Hear 19 :1-36.
McCullough JA, Cunningham LA, Wilson RH . (1992) .
Auditory-visual word-identification test materials : computer application with children . J Am Acad Audiol
3 :208-214 .
McCullough JA, Wilson RH, Birck JD, Anderson LG .
(1995) . A multimedia approach for estimating speech
recognition of multilingual clients . Am JAudiol 3:19-22 .
Nelson DA, Chaiklin JB . (1970) . Writedown versus talkback scoring and scoring bias in speech discrimination
testing. J Speech Hear Res 13 :645-654 .
Spitzer JB . (1980). The development of a picture speech
reception threshold test in Spanish for use with urban
U.S . residents of Hispanic background . J Commun Dis
13 :147-151 .
U.S . Bureau of the Census. (1990) . Statistical Abstract
of the United States : 1990. Washington, DC : U.S .
Government Printing Office .
Weisleder P, Hodgson WR. (1989) . Evaluation of four
Spanish word-recognition-ability lists . Ear Hear
10 :387-393 .
Wilson RH, Antablin JK. (1980) . APicture-Identification
Task as an estimate of the word-recognition performance
of nonverbal adults . J Speech Hear Disord 45 :223-248 .
Wilson RH, Antablin JK . (1982) . Picture-Identification
Task: a reply to Dillon . J Speech Hear Disord 47 :109-112 .