Computer-Assisted Vocabulary Acquisition: The CSLU Vocabulary Tutor in Oral-Deaf Education Lecia J. Barker University of Colorado

Deficits in vocabulary have a negative impact on literacy and interpersonal interaction for deaf children. As part of an evaluation, an outcomes assessment was conducted to determine the effectiveness of a computer-based vocabulary tutor in an elementary auditory/oral program. Participants were 19 children, 16 profoundly deaf and 3 hearing. The vocabulary tutor displays line drawings or photographs of the words to be learned while a computer-generated avatar of a “talking head” provides synthesized audiovisual speech driven from text. The computer system also generates printed words corresponding to the imaged items. Through audiovisual reception, children memorized up to 218 new words for everyday household items. After 4 weeks, their receptive vocabulary was tested, using the avatar to speak the name of each item. Most of the students retained more than half of the new words. The freely available vocabulary tutor, whose characteristics can be tailored to individual need, can provide a language-intensive, independent learning environment to supplement classroom teaching in content areas.

Half of all deaf seventeen- and eighteen-year-olds who took the 1996 Stanford Achievement Test read at the fourth grade level (Holt, Traxler, & Allen, 1997). Although research shows that deaf students’ potential for learning is the same as that of their hearing peers (Quigley & Paul, l989; Bradley-Johnson & Evans, 1991), they are This material is based on work supported by the National Science Foundation (9996246 and 9996075). Special thanks to Pamela Connors, who played a significant role in planning the study and capturing the data. This study would not have been possible without the interest and contributions of Dr. Patrick Stone, Christine Soland, Alice Davis, George Fortier, Kerry Gilley, Linda Goodwin, and, especially, the children. Correspondence should be sent to Lecia J. Barker, Alliance for Technology, Learning, and Society Evaluation and Research Group, University of Colorado, UCB 320, Boulder, CO 80309–0320 (e-mail:[email protected]). © 2003 Oxford University Press

significantly less likely to graduate from high school than hearing children and much more likely to work as manual laborers (Easterbrooks, 1999). According to educators, literacy is the strongest predictor of success in the education of deaf children (Carver, 1990; Carney & Moeller, 1998), yet increasing the literacy of deaf children is difficult for many reasons, one of which is inadequate vocabulary. Vocabulary is critical to reading comprehension (Bradley-Johnson & Evans, 1991; Garrison, Long, & Dowaliby, 1997; Paul, 1996; Quigley & Paul, l989; Smith, 1997) and, by extension, to the development of “higher level thinking skills and to [understanding] advanced abstract concepts in later years” (Roffe, 1998, p. 27). In fact, “learning, as a languagebased activity, is fundamentally and profoundly dependent on vocabulary knowledge” (Baker, Simmons, & Kameenui, 1995). Yet vocabulary development is especially difficult for deaf children, because they “have no undistorted access to the flow of language and information in the environment” (Vess & Douglas, 1995, p. 1127). Vocabulary and language deficiencies have effects beyond education in the interpersonal and private realms. It has long been argued that deaf children’s development of personal identities and cognitive function are negatively affected by their relatively impaired language ability (Breslaw, Griffiths, Wood, & Howarth, 1981; Carney & Moeller, 1998). Deaf children tend to be passive in relation to their hearing communication partners (Johnson, 1997), in part because carrying on ordinary conversation requires a good foundation in the vocabulary for the situation as well as the ability to spon-

188

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

taneously produce it. Vocabulary is also important for solitary activities, such as watching television and reading (Jensema, McCann, & Ramsey, 1996). The vocabulary deficit is not specific to educational content areas but extends to everyday words that hearing persons take for granted. One solution, according to the Deaf Children’s Society of British Columbia (2000), is for parents to “label everything by its name in the house (e.g., ‘stereo’, ‘lamp’, ‘mirror’).” However, families often develop routines and conventions that make it easy to communicate without using language (K. Gilley, telephone interview, 2000). And when they do use language, “hearing people, especially parents, who live with deaf children often ‘dumb down’ their language to make themselves more easily understood, [using] . . . fewer idioms, adjectives, and synonyms” (Roffe, 1998, p. 24). Further, most parents know their children have holes in their vocabularies, but older children often resist this type of active education delivered by their parents. For these reasons, vocabulary acquisition is a typical goal of treatment by speech pathologists and teachers of students in manual, total, and auditory/oral programs for both content-related and “everyday” subject matters (Carney & Moeller, 1998). Research on vocabulary instruction with hearing children shows that intentional vocabulary instruction is more effective for developing vocabulary knowledge than incidental vocabulary instruction, that is, through ordinary experience with language (Baker, Simmons, & Kameenui, 1995). However, no single best method of intentional vocabulary instruction has been identified (Beck & McKeown, 1991). The explicit teaching required for intentional vocabulary instruction can be quite time-consuming for teachers. As a consequence, teachers often seek out less time-consuming ways of teaching vocabulary. One choice is computer software that teaches vocabulary, but relatively few software packages are available for developing vocabulary for children with hearing loss, and even fewer are available for children in auditory/oral programs, where word discrimination and production are considered critical to the development of oral language (Ling & Ling, 1978). This study investigated the acquisition of “everyday” vocabulary of deaf children in an auditory/oral program using specially designed computer-based vocabulary applications.

Background of the Study Development of the Toolkit for Use by Deaf Children In September 1997, the Center for Spoken Language Understanding (CSLU) at the Oregon Graduate Institute received a National Science Foundation Challenge Grant for a 3-year project to develop interactive technologies and applications for learning and language training with profoundly deaf children. These tools were developed and tested in collaboration with the educators and deaf students at an oral school in Portland, Oregon. “Oral School”1 is a day school that provides auditory/oral elementary education for 16 profoundly deaf children integrated with a smaller number of hearing children. The CSLU Toolkit includes technology for text-to-speech synthesis and speech recognition, which can be used to drive written word-to-speech and speech-to-word coding and also to generate visual speech by a speaking avatar, an animated talking face, nicknamed “Baldi” (Massaro, 1998). The CSLU Toolkit also includes a number of important further features (for details, see http://cslu.cse.ogi.edu). At the time of the study, children at Oral School, both deaf and hearing, had been using Baldi (their way of referring to the toolkit) for 3 years, over which time CSLU researchers made many modifications to accommodate teacher and student needs. To aid the integration of the toolkit into teaching and learning, two CSLU staff members worked on site at the school during the 3-year grant period, supporting toolkit use, acting as liaisons between researchers and school staff, and interacting with teachers and students on a regular basis. With their support, the teachers and speech pathologist integrated the toolkit use into the curriculum at Oral School to enable independent practice with speech production and reception while students were learning vocabulary and content. During the third year of the project, independent evaluation of the project was sought to gather and analyze data on the teaching and learning outcomes resulting from use of the toolkit. This report is the first based on this evaluation. In addition to the results of the vocabulary study that form the focus here, data collection included interviews (two with each Oral School teacher and the speech pathologist, three with the school’s director, one or more with each toolkit researcher and pro-

Computer-Assisted Vocabulary Acquisition

grammer, and numerous, lengthy conversations with the on-site liaisons); observations (ordinary classroom and nonclassroom school-based situations, children’s use of various toolkit applications); and document review (weekly reports, articles written by CSLU researchers). I also attended a retreat in which all members of the project participated, including researchers, programmers, educators, and liaisons. In interviews, Oral School teachers described the following benefits from children’s toolkit use as a language learning tool. First, toolkit-based applications can be multimodal, permitting practice with reception, production, reading, and writing. Using headsets and microphones, children can produce language, listen to their own recordings, listen to Baldi saying the same utterance, and correct their own speech. Also, both discrimination and production can be built into the same task, which more closely mimics a natural language situation. Second, natural or synthetic speech is automatically synchronized with Baldi’s realistic lip, tongue, mouth, and facial movements, permitting practice with lipreading and increasing the amount of information available for reception (Massaro, 1998). Third, the orientation of Baldi’s face can be changed for viewing from different perspectives while he is speaking. When deaf students have difficulty making a sound, teachers often try to show them the place of articulation. This teaching function can be simulated by making Baldi’s face transparent, which reveals the movements of the teeth and tongue while producing speech. Fourth, the children see Baldi as ultimately patient, speaking slowly and carefully, and repeating lessons as many times as necessary for a child to learn. The visual and interactive realism of Baldi is good; I observed one child replying “hi” to Baldi when he greeted her. Finally, teachers say that one of their greatest teaching challenges is finding the time to teach children individually. Use of the toolkit allows children to work independently within the classroom on individually tailored learning programs, while the teacher can work with specific children, one at a time, for individual face-to-face instruction.

Vocabulary Tutor Instructional Design During the third year of the grant period, Oral School educators discussed using the toolkit to help children de-

189

velop their relatively deficient vocabularies of everyday items. Several examples were given that identified this need. Children were reported trying to have conversations, using phrases such as “the window in the front of the car,” “the big shelf where the sink is,” or “the step by the street,” rather than “windshield,” “counter,” or “curb.” Oral School educators believed that memorizing the names of things in their world would help children to take the next step to become “flexible with vocabulary” (Yoshinaga-Itano, 1997). The lead CSLU liaison worked with teachers to create a set of applications that would incorporate teachers’ vocabulary instruction methods, be easy to create, and would appeal to the children. Oral School teachers and the speech pathologist have master’s degrees with specific training for auditory/oral education, placing Oral School teachers into Brill’s (1997) “best” categories of competency and interest in teaching. Teachers’ methods of vocabulary instruction were elicited, then two meetings were held during which teachers agreed on what they thought a vocabulary acquisition application should contain. The resulting instructional method included both visual and aural reception and production. When Baldi speaks a word, the child is required to respond by mouse-clicking the corresponding image or written word on the screen. According to Johnson (1997), such stimulus-response applications can help deaf children to memorize words while at the same time shaping auditory reception and speech production. The vocabulary pedagogy was incorporated into applications teachers created using an embedded software utility (CSLU vocabulary “wizard”). This utility allows users to insert the desired image, select portions of the image for vocabulary instruction, enter the text to be associated with these selections, and individualize the instruction by child, if desired. The applications were put through a series of test-and-revise cycles to ensure technical usability and incorporation of the desired teaching method. One teacher, who described herself as “computer illiterate,” reported creating applications in about 10 minutes. The applications generated by using the wizard, called vocabulary tutors, include Baldi’s speaking face with both facial dynamics and acoustic speech, as well as written words and pictures with several clickable areas; these areas become highlighted when the mouse is passed over them. The lessons are

190

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

consistent with Easterbrook’s (1999) suggestions that software for deaf education be visual, presenting information in pictures or with graphics and with simple, minimal text. Consistent with the instructional procedure developed by teachers at Oral School, the vocabulary tutors required that each vocabulary item be presented at least four times (picture-speech/speaking head) and the written representation of the word be presented once; in addition, students were required to orally produce each item once. Baldi gave spoken feedback for correct and incorrect answers; visual feedback appeared as smiley faces or unhappy (though amusing) smileys. Although students saw the written word once, emphasis was on audiovisual reception, with all testing done using reception only. The vocabulary tutor sequence is presented in Table 1.

Table 1

Method Participants Ten girls and nine boys used the applications. Oral School is split into two groups, the “upper school” and the “lower school.” Student characteristics are summarized in Table 2. All children in the lower school class used the lessons. In the upper school class, all the deaf students and two of five hearing students used the lessons. All the deaf children have profound bilateral sensorineural hearing loss (90 dB HL or greater); nine were deaf at birth and seven developed deafness during infancy or early childhood due to meningitis or other illnesses. Some of the children use cochlear implants exclusively, some use hearing aids with FM systems exclusively, and some of the children have a cochlear implant in one ear and hearing aid in the other. According

Sequence of vocabulary tutor

Segment

Purpose

Actions

Pretest

Introduce lesson objectives to student; provide baseline data for determining how manywords a child actually learns.

Presentation

Associate the visual images of the words with receptive and written language.

Practice

Drill to ensure child has associated words with their images.

Posttest

Assess whether child should repeat the lesson or move on.

Production

Child practices expression and self-assesses the quality of the expression.

Baldi says, “click on the ” for each item defined as a learning goal by the teacher. The student moves the mouse around the picture. A line around defined items becomes highlighted and flashes upon mouse-over. This is repeated for each word. Baldi says, “Let me show you” and the word “demonstration” appears momentarily on the screen. Baldi says, “this is the ” for each item and the highlighted item flashes. The written form of the word is also shown in this stage. Baldi then asks the user to click on the item. Baldi repeats this sequence until the student clicks on the right image for each word. Baldi says, “Let’s practice” and the word “practice” appears on the screen. He says, “click on the ” or “where’s the ?” The student clicks on it. If correct, he or she gets positive aural and visual feedback (“good,” “great,” a smiley face); if wrong, the student gets negative feedback (“sorry,” “no,” and frowning smiley) and another word is randomly presented. When a child correctly identifies all words twice (or more, depending on teacher choice), he or she moves on to the posttest. Baldi says “click on the ” and the child gets only one try. If correct, Baldi tells the student “correct” or “good” and the student gets a smiley face. If incorrect, Baldi tells the student “sorry, that is not the ” and the student gets a frowning smiley face. Baldi says, “this is the ; say .” The child’s utterance is replayed for the child and recorded in a.wav file for teachers to listen later.

Computer-Assisted Vocabulary Acquisition Table 2

191

Profile of participants using vocabulary tutor

Group

Age range

Deaf

Hearing

Total

Incidence of additional problems

Upper school

11–14 years (mean 12.3) 8–10 years (mean 8.6) 8–14 years (mean 11.15)

11

2

13

5

1

6

Usher Syndrome (1), ossification of cochlea (2), ADHD (1), poor fine motor skill (2) Usher Syndrome (1), ossification of cochlea (1)

16

3

19

Lower school Total

to teachers’ subjective judgments, all of the children are average to good students relative to their language ability on a “poor–average–good” scale. As in most groups of deaf children, the additional difficulties of these children vary widely. These additional problems, which are indicated in Table 2, may account for some of the variance in the results.

Vocabulary Items and Lesson Development The teachers’ aim was to help the children develop their vocabulary of everyday items. The words, all nouns, fit into two categories: objects or actions (e.g., using the gerund form of a verb, such as “digging”) that children have seen in their homes, at school, in supermarkets, but for which they have no linguistic label. Students photographed many of the items to be learned in their own homes. According to Tarulli (1998), photographs have strong pedagogical value for deaf children, since they help them to learn language to describe what they know best, their home and family, and can allow them to build specific vocabulary in context. In addition, teachers chose drawings of common objects as learning goals. Teachers chose the items they believed students should learn as new vocabulary. The photographs and drawings were computer scanned by technical support staff who integrated them in the vocabulary tutor. Lessons were not personalized for each child, though a great deal of personalization is made possible by the capabilities of the toolkit and associated software. Six applications were made for exclusive use by lower-school children (these had fewer words). An additional 22 lessons were made for use by both groups, though the lower-schoolchildren used fewer of these. Most lessons contained 10 words, though some contained 15 and those made specifically for the lower school

Usher Syndrome (2 [siblings]), ossification of cochlea (3), poor fine motor skill (2)

had as few as 5. The usage details and words included in the lessons are shown in Table 3.

Student Use of Applications Oral School children had used the toolkit routinely during the 2 years before this assessment. By the time of the study, there were no reported difficulties in its use. While using the toolkit, some children looked at the animated talking face and others did not, depending on their ability to hear and see, or personal preference. All the children used headphones and microphones during the test period.2 Students used the segments of each lesson as described in Table 1. Once a student received a score of 100% on the posttest, he or she moved on to another lesson the next day. Lessons were repeated daily until the student scored 100%.

Data Acquisition Data from each use of a lesson were captured automatically, totaling 217 sessions. Data included item identity clicked on for each question of the pretest, practice, and the posttest; student name; time elapsed in seconds for the entire lesson and for each segment. Correct and incorrect items were not recorded during presentation. Retention was tested 4 weeks after each successfully completed lesson, using the posttest procedure only.

Dependent Variables and Experimental Hypotheses The number of words per lesson was recorded as mouse clicks for each item. While the applications required that students produce the words, production competence was not assessed in this study. Both immediate

192

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

Table 3

Lessona

Usage details, pretest results, and words of each lesson No. students

Times used

Lower school pretest results

Upper school pretest results

M (SD)

M (SD)

Items

1

15

21

2.75 (2.22)

2.82 (1.33)

6

2

15

28

3.75 (2.63)

3.27 (1.49)

6

3

15

28

1.75 (2.06)

3.45 (1.92)

7

4

15

26

4.75 (2.06)

5.36 (1.43)

10

5

14

23

2.67 (1.15)

4.00 (2.24)

10

6

14

30

4.33 (5.13)

3.27 (2.57)

10

7

10

21

Raw score 2b

2.33 (3.35)

10

8

10

18

Raw score 2b

3.67 (1.00)

10

9

10

16

Raw score 1b

3.89 (1.76)

11

10

9

15

Raw score 1b

3.13 (3.36)

13

11

9

17

—

3.11 (2.93)

10

12

9

11

Raw score 10c

5.38 (1.51)

10

13

10

16

Raw score 10c

3.40 (3.60)

11

14

8

10

Raw score 10c

5.86d (3.02)

10

15

7

11

—

5.00d (4.93)

15

16

6

11

—

7.29e (4.96)

14

17

6

13

—

4.00e (3.87)

13

Words included Bicycle pump, bicycle wheel, extension cord, tool box, water heater, broom Hiking backpack, gas lawn mower, push lawn mower, cooler, water bottle, ten speed bicycle Glasses, bowls, plates, mugs, bread box, tea kettle, oven Chair, cordless phone, microwave, coffee maker, dishwasher, mini blinds, floor, ceiling, cupboard, counter Dish soap, sink, dish drainer, faucet, handle, salt and pepper shakers, place mat, booster chair, tablecloth, napkins Cutting board, loaf pan, strainer, ice cream scooper, wooden spoon, rubber spatula, tablespoon, teaspoon, can opener, silverware Electric mixer, egg beaters, measuring cups, rolling pin, rack, grater, cheese slicer, serving spoon, ladle, serving spatula Kitchen knives, pot rack, toaster, stool, pot, burner, outlet, basket, frying pans, kitchen towels Couch, coffee table, stereo, table lamp, fireplace, TV stand, mantle, vase, TV antenna, remote control C-clamp, jigsaw, saw, grounding plug, circular saw, tape measure, wrench, scraper, hatchet, workbench, pliers, hatchet, mallet Shower head, curtain rod, sponge, stopper, soap dish, medicine cabinet, bath mat, shower curtain, hair dryer, towel rack Door lock, side mirror, glove compartment, bucket seat, gear shift, steering wheel, visor, seat belt, vent, armrest Door handle, windshield wiper, gas gauge, rearview mirror, speedometer, turn signal, horn, ignition, brake, accelerator, dashboard License plate, brake light, backup light, taillight, backseat, child’s seat, gas tank, headrest, hubcap, tire Stepladder, feather duster, circuit breaker, dustpan, plunger, vacuum cleaner, attachments, ironing board, iron, pipe, light bulbs, dryer, washing machine, laundry, laundry detergent Hammock, sprinkler, barbeque, lounge chair, wheelbarrow, garden hose, drainpipe, gutter, watering can, patio, power saw, lawn, hedge clippers Picking, mowing, trimming, watering, digging, planting, painting, carrying, salting, shoveling, chopping, raking, filling

Computer-Assisted Vocabulary Acquisition Table 3

Lessona

Usage details, pretest results, and words of each lesson (continued) No. students

Times used

Lower school pretest results

Upper school pretest results

M (SD)

M (SD)

Items

Words included Brake, seat, sprocket, pedal, reflector, fender, handle bars, hand brake, cable, valve, spokes, gear changer, chain X-ray, wheelchair, sling, band-aid, cast, examining table, crutch, attendant, stethoscope Chart, doctor, stretcher, instruments, oral hygienist, drill, basin, pedal, dentist Nurse, patient, stitches, alcohol, cotton balls, gauze bandage, gauze pads, needle, syringe Blinds, bed spread, wallpaper border, bookshelf, poster, toys Chest of drawers, banner, light switch, thermostat, picture, bathrobe, door knob Desk, lamp, keyboard, computer monitor, stereo, cord, wastepaper basket Bunk bed, ladder, guard rails, armchair, headboard Forehead, eyebrows, eyelashes, nostril, cheeks, chin Clock, door frame, desk drawers, pegs, cubby, bed frame

18

2

2

—

1.50 (2.12)

13

19

2

5

—

2.00 (2.83)

9

20

2

3

—

2.50 (0.71)

9

21

2

2

—

0.50 (0.71)

9

Tutor 1

6

11

2.00d (1.06)

—

6

Tutor 2

5

14

4.60d (2.07)

—

7

Tutor 3

4

9

3.60d (3.21)

—

7

Tutor 4 Tutor 5

4 5

7 10

2.50d (0.58) 2.80d (1.79)

— —

5 6

Tutor 6

3

5

2.00 (1.00)

—

6

217

383

Total

193

9.3

a

“Lessons” were developed for children in the upper school, though they were also used by lower school children. “Tutors” were developed for children in the lower school. b

Only one lower-school child used this lesson.

c

One Lower-School child used this lesson; this was a hearing child.

d e

Includes one hearing child.

Includes two hearing children.

acquisition and longer-term retention were analyzed. The study addressed two questions. With respect to immediate vocabulary acquisition, what are students’ immediate learning gains from using lessons and what differences emerged across groups (upper versus lower schools; deaf versus hearing children)? To what degree are these learning gains retained over time?

Results Time Spent on Task The mean number of minutes lower-school students spent using lessons was 147 minutes, or nearly 21⁄2 hours. Upper-school students spent about an hour more using lessons than lower-school students. Table 4 shows the

total time students in the upper and lower schools used the lessons, as well as the mean number of attempts.

Vocabulary: Basic Measures and Immediate Gain To calculate vocabulary gain, it was necessary to establish a reliable measure of the extant (pretraining) vocabulary of each child. Since children’s responses were not monitored for accuracy, some items clicked in response to a spoken name during the pretest period may have been arrived it by chance. However, these items were identified during the training phase, when they were incorrectly produced. This adjusted measure (“already known”) showed no difference between the school class groups, t(df = 17) = .901, p = ns). Table 5 shows the means and distributions of the corrected value (the al-

194

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

Table 4

Total time using lessons

Table 6

Total lessons attempted

Minutes using lessons

Baseline scores by school Baseline scores (already known)

School

n

M (SD)

M (SD)

Group

n

M (SD)

Lower Upper Total

6 13 19

17.3 (10.7) 21.4 (8.1) 20.1 (8.9)

146.6 (111.1) 215.7 (99.2) 193.9 (105.3)

Upper school Lower school

13 6

38.3 (24.7) 22.1 (17.8)

175

Immediate gains

Total words presented Deaf Hearing Combined Words already known Deaf Hearing Combined Words correct on posttest Deaf Hearing Combined

Min.

Max.

M (SD)

12 13 12

217 72 217

116.1 (60.7) 51.7 (33.5) 105.9 (61.4)

1 13 1

86 52 86

32.2 (24.4) 38.3 (22.0) 33.2 (23.6)

150

125

2 13 2

215 72 215

113.1* (33.2) 51.3 (62.0) 103.4* (62.2)

19 participants: 16 deaf, 3 hearing. *Significant learning gain over “already known,” p < .0001.

ready known scores), as well as total words presented to students in the lessons and the number correct on the posttest by the deaf and hearing groups and combined scores. Deaf children learned a mean of 81 new words (range: 0–158) over the study period. As expected, hearing children scored significantly better on words already known than deaf children. Also, the mean posttest score of the deaf children was a significant difference over their already known score, whereas the mean posttest score of hearing children was not a significant difference over their already known score. Table 6 shows the mean values for baseline scores (based on already known) for upper and lower school. The difference is not statistically significant (Levene’s test for equality of variance = .464, p = .172). Figure 1 shows the immediate vocabulary gains by individual students. The scatterplot indicates that all but two students experienced an overall gain in vocabulary. For most of the deaf children, represented by circles on the chart, gain increased significantly in proportion to their pretest scores. For hearing children, represented by solid triangles, gain was lower, showing a

Post-test, numbers correct

Table 5

100

75

50

25 Deaf Hearing

0 0

25

50

75

100

'Already known', numbers correct

Figure 1 known.

Correlation between posttest scores and already

larger original vocabulary. Hearing children scored 81% on pretests, whereas deaf children scored only 23% correct on pretest. This difference is statistically significant (Mann-Whitney U = .000). Overall, students could correctly identify significantly more words at the end of each lesson than they could identify at the beginning (baseline mean score 3.78 per lesson, posttest mean score 9.02 per lesson; p < .0001). Students correctly identified about five words per lesson that they had not previously known. Figure 2 shows students’ cumulative learning gains by lesson.

Retention It is reasonable to expect that some words will be forgotten over time, especially if they do not arise in conversation or other context. An average of 32 days after achieving 100% on the posttest of each lesson, students were given the same posttest with no preparation. The mean number of words retained per students was 38.5 out of the 70 they had learned, or 55% of their posttest scores. Table 7 shows the overall mean gains for the upper and

Computer-Assisted Vocabulary Acquisition

Figure 2. Table 7

Cumulative learning gains. Words retained after four weeks M (SD)

Upper school (13 students) Words already known No. correct on posttest No. correct on retest Lower school (6 students) Words already known No. correct on posttest No. correct on retest

38.3 (24.7) 119.9* (65.1) 74.4* (39.3) 22.2 (17.8) 67.7* (38.8) 41.2* (6.1)

*Significant gain over baseline score (already known) at p < .0001.

lower schools. Though retest scores were significantly lower than posttest scores, the gain was still significantly improved compared with baseline scores.

Discussion These results suggest that the vocabulary tutor is an effective means of explicitly teaching vocabulary, despite some loss of vocabulary in the period following immediate testing. On average, children learned 70 words for everyday objects from immediate use of the lessons and remembered nearly 39 of these words after more than 4 weeks. Had the students been taught in the context of some other learning goal, it is possible that they would have retained more of the items learned in isolation from other school-based activities. Oral School teachers believe that the tutor could be effectively incorporated into the curriculum, for instance, in the context of classroom instruction for subjects such as anatomy. Teachers suggested that it could be useful in reinforcing vocabulary knowledge within such subject areas. The kinds of vocabulary words acquired are only those that can be pre-

195

sented adequately in visual form. This covers many nouns, some actions, and some prepositions (over the mat, under the mat). As teachers point out, however, as presently configured, the application is not suitable for teaching abstract concepts. Children had mixed reactions to using the vocabulary tutor. Some liked it; others did not. Also, as in any classroom, children had differential success. For example, one child with other learning difficulties was unhappy because the task was too hard for him. For this child, just sitting still and manipulating the mouse while listening to Baldi was difficult. Some children worked hard at learning; they were attentive to what they were doing and cared to get the answers right. One child enjoyed using the tutor and performed well, but personal problems at the end of the year caused his work to decline. Teachers and the on-site liaisons reported that the children were happy when they got all happy faces (i.e., received 100% correct) and that they often compared their results with each other. Statistical variability in the results of this study reflects the individual variation in the student community, which is marked. These results must be interpreted in light of these differences. The lack of statistically significant difference between upper- and lower-school students’ baseline scores in number of words categorized as “already known” deserves further attention. One would expect that older children would have a larger vocabulary than younger children, yet all these children showed a similar level of deficit relative to their hearing peers in their verbal knowledge of the names of everyday household items. Oral School teachers suggest that this is probably due to the routines that families develop for getting things done. Once these routines are in place, parents may forget the need to explicitly teach children the names of things. Also, as children get older, they may resist this sort of explicit teaching. The disturbing failure to develop a basic vocabulary despite extensive schooling, as revealed by this study, suggests that more research needs to be conducted in this area, both for learning the reasons for the deficit and for developing solutions. Oral School’s director believes the vocabulary tutor can provide consistent exposure to content. Moreover, its use does not displace other important learning activities. Teachers questioned the amount of time the children spent using the tutor, however, and suggested that

196

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

the students use the toolkit at home for everyday vocabulary and at school to augment their vocabulary within content areas. The tutor has been criticized for being nothing more than an expensive, high-tech memorization tool; however, teachers and the speech pathologist believe that deaf children need particular help with memorizing words and that the tutor has the potential to provide essential vocabulary acquisition scaffolding in a multimodal environment. The tutor’s emphasis on audiovisual reception, so that which must be memorized is the sound of words rather than their written form, is another important feature of it. Learning to discriminate sounds and apply meaning to them is an important part of literacy instruction for oral-deaf education. Several families requested the toolkit and the vocabulary tutors for use over the summer so that they could teach their children the words they would need for particular summer activities (e.g., camping) or just to increase their children’s vocabularies. This is a particularly important outcome, in light of studies of spoken language development in children. In their 1996 study, Musselman and Kircaali-Iftar identified several factors contributing to high performance among children with severe or profound deafness. Two of these were structured teaching by parents and parents who are highly committed to their children’s language development. According to Oral School teachers, using Baldi has raised parents’ awareness of what their children do not know and what they need to work on. The ease of use of the vocabulary wizard for developing the lessons is an important contributor to parents’ ability to use the tools in their homes.

Limitations of the Study A key weakness of this study is the lack of a control group for comparison of results of teaching vocabulary using other methods or for tracking the children’s increase in vocabulary without explicit training. I believe that these children would not have made the gains they made without the vocabulary tutor, considering the similarities in vocabulary knowledge between the younger and older groups. Although no claim is made that toolkit use would result in better results than other vocabulary acquisition methods, it would be interesting to compare

teaching methods if the number of varying factors across teaching methods could be minimized. In addition, no attempt was made in this study to include any gains children may have made in their speechreading. Another study might use a broader definition of learning outcomes and compare the speechreading skills of students interacting with Baldi to students interacting with real persons. Deaf children have about three times as many additional disabilities as hearing children (Pollack, 1997), which adds confounding variables and makes generalizations quite difficult. Conducting a similar study for a larger audience that also factors in children’s etiologies, learning disabilities, problems with implants, and so on would improve our knowledge of how this tool works for aiding vocabulary acquisition for different segments of the oral-deaf education population. Further, the sample size in this study is quite small. Though generalizations cannot be made safely, researchers and educators may consider whether the groups of students and learning environments that they study or educate are analogous to those described in this article.

Implications for Practice This study contributes a small piece to the critical need for outcomes data that can inform decision making in schools (Amiot, 1998; Carney & Moeller, 1998; Easterbrooks, 1999); such studies are needed to address the effectiveness of different types of educational tools and for different subgroups of the deaf population. Most deaf children in the United States are educated in public schools (Easterbrooks, 1999), which is a serious problem because few public school teachers have the in-depth training or resources needed to effectively teach deaf children. Teachers can often comfortably assume that hearing children know certain words when they introduce new subjects. For example, hearing seventh-grade students usually already know, have heard, or can guess the meaning of the words “border” or “boundary.” A teacher can count on the presence of this knowledge and get on with the geography lesson. Unfortunately, mainstreamed, hearing-impaired children may not recognize the words, so the lesson may be lost on them or take longer to learn. The vocabulary tu-

Computer-Assisted Vocabulary Acquisition

tor could be used to bring these students and other students with language deficits up to speed for terminology needed within particular content areas. The vocabulary tutor, as one small tool in the acquisition of language, could be used inexpensively in schools, given that it can be downloaded and installed free via the internet.3 Its ease of use takes teachers little time to support these children. The developers have also begun work on facilitating language acquisition for Spanish-speaking deaf children, the number of whom is increasing rapidly in the United States. These children have special difficulties, comprehending printed material significantly worse than white, non-Hispanic American deaf children (Easterbrooks, 1999). Whereas no educational tool or technology should be considered a “magic bullet,” it is clear that the CSLU vocabulary tutor may be an effective way to increase vocabulary and literacy of children in oral-deaf education in an independent, aural language– intensive environment, freeing teachers’ time for more individualized instruction with other children.

Notes 1. This nomenclature protects the anonymity of the school. 2. One child used the written version of the words, rather than a visual image, for all segments of each lesson due to difficulties with her amplification system. 3. The CSLU toolkit and vocabulary tutor can be downloaded from http://cslu.cse.ogi.edu/toolkit.

References Amiot, A. (1998). Policy, politics, and the power of information: The critical need for outcomes and clinical trials data in policy-making in the schools. Language, Speech, and Hearing Services in Schools, 29, 245–249. Baker, S. K., Simmons, D. C., & Kameenui, E. J. (1995). Vocabulary acquisition: Synthesis of the research. Eugene, OR: National Center to Improve the Tools of Educators. Beck, I., & McKeown, M. (1991). Conditions of vocabulary acquisition. In R. Barr, M. Kamil, P. Mosenthal, & P. D. Pearson (Eds.), Handbook of reading research (vol. 2, pp. 789– 814). New York: Longman. Bradley-Johnson, S., & Evans, L. D. (1991). Psychoeducational assessment of hearing impaired students: Infancy through high school. Austin, TX: Pro-ED. Breslaw, P. I., Griffiths, A. J., Wood, D. J., & Howarth, C. I. (1981). The referential communication skills of deaf children from

197

different educational environments. Journal of Child Psychology, 22(3), 269–282. Brill, R. G. (1997). The educational preparation of oral teachers of the deaf. American Annals of the Deaf, 412(3), 65–70. Carney, A. E., & Moeller, M. P. (1998). Treatment efficacy: Hearing loss in children. Journal of Speech, Language, and Hearing Research, 41, S61–S84. Carver, R. J. (1990). The challenge: Access, literacy, and the deaf child. Plenary address, 12th Biennial Convention of American Society of Deaf Children, Vancouver, British Columbia. Deaf Children’s Society of British Columbia Literacy Guide. (n.d.). Useful tips on how to stimulate literacy in deaf children [Web Page]. Retrieved May 2000 from http://dww.deafworldweb. org/pub/l/lit-tips.html Easterbrooks, S. (1999). Improving practices for students with hearing impairments. Exceptional Children, 65(4), 537–554. Garrison, W., Long, G., & Dowaliby, F. (1997). Working memory capacity and comprehension processes in deaf readers. Journal of Deaf Studies and Deaf Education, 2(2), 78–94. Holt, J. A., Traxler, C. B., & Allen, T. E. (1997). Interpreting the scores: A user’s guide to the 9th Edition Stanford Achievement Test for educators of deaf and hard-of-hearing students. Washington, DC: Gallaudet Research Institute. Jensema, C., McCann, R., & Ramsey, S. (1996). Closed-captioned television presentation speed and vocabulary. American Annals of the Deaf, 141(4), 284–292. Johnson, C. E. (1997). Enhancing the conversational skills of children with hearing impairment. Language, Speech, and Hearing Services in Schools, 28, 137–145. Ling, D., & Ling, A. (1978). Aural habilitation: The foundations of verbal learning in hearing-impaired children. Washington, DC: A. G. Bell Association for the Deaf. Massaro, D. W. (1998). Perceiving talking faces: From speech perception to a behavioral principle. Cambridge, MA: MIT Press. Musselman, C., & Kircaali-Iftar, G. (1996). The development of spoken language in deaf children: Explaining the unexplained variance. Journal of Deaf Studies and Deaf Education, 1(2), 108–121. Paul, P. V. (1996). Reading vocabulary knowledge and deafness. Journal of Deaf Studies and Deaf Education, 1(1), 3–15. Pollack, B. J. (1997) Educating children who are deaf or hard of hearing: Additional learning problems [Web Page]. Retrieved May 2000 from http://ericec.org/digests/e548.htm Quigley, S. P., & Paul, P. V. (1989). English language development. In M. C. Wang, M. C. Reynolds, & H. S. Walberg (Eds.), Handbook of special education: Research and practice (vol. 3, pp. 3–21). New York: Pergamon Press. Roffe, S. (1998). The ‘dumbing down’ of language. Hearing Health, 14(3), 24–27. Smith, C. B. (1997). Vocabulary instruction and reading comprehension. Bloomington, IN: ERIC Clearinghouse on Reading, English, and Communication. Tarulli, N. J. (1998). Using photography to enhance language and learning: A picture can encourage a thousand words. Language, Speech, and Hearing Services in Schools, 29, 54–57. Vess, S. M., & Douglas, L. S. (1995). Program planning for chil-

198

Journal of Deaf Studies and Deaf Education 8:2 Spring 2003

dren who are deaf or severely hard of hearing. Best practices in school psychology III (pp. 1123–1133). Washington, DC: National Association of School Psychologists. Yoshinaga-Itano, C. (1997). The challenge of assessing language in children with hearing loss. Language, Speech, and Hearing Services in Schools, 28, 362–373. Received March 12, 2001; revisions received December 29, 2001; accepted January 14, 2002