Speech Perception Skills of Hindi Speaking Children with Pre-lingual Hearing Loss Using Hearing Aids and Cochlear Implants

Abstract

Very few published studies have reported auditory speech perception in Hindi children with pre-lingual hearing loss. The study is aimed at comparing the speech perception skills of Hindi speaking children with pre-lingual severe to profound hearing loss using hearing aids and cochlear implants. Forty-three 6 to 8-year old children were included as participants, of which 22 were bilateral behind-the-ear hearing aid (HA) users and 21 were unilateral cochlear implant (CI) users. Speech perception was assessed through a forced-choice, picture-pointing task using recorded stimuli presented at 70 dB HL in the sound field. The skills assessed include: (a) pattern perception, (b) bisyllabic word identification, (c) monosyllabic word identification, (d) sentence identification and (e) minimal pair identification. Children using CI consistently performed significantly better than those with HA on all tasks. For the skills assessed, best performance was seen in pattern perception and poorest performance was seen in monosyllabic word identification. One participant from the CI group obtained ceiling scores for pattern perception and bisyllabic word identification. There was no statistically significant difference in the performance of 6 to 7 and 7 to 8-year-old children for any of the tasks. Children fitted with CI have better access to the cues important for perception of speech and hence perform consistently better than those using hearing aids. Recorded speech perception test can be used with children using cochlear implants and hearing aids.

Introduction

Speech Perception is the process of receiving the verbal component of language. Hearing impairment at any age affects the ability of sound detection, discrimination, identification and speech perception. The various acoustic cues in the speech of the persons around are not available to the child with hearing impairment, thus making it difficult to extract these crucial features to develop perception and production of speech. These difficulties in speech perception and production are more pronounced in children with bilateral congenital or pre-lingual severe to profound hearing loss. The outcomes of auditory speech perception in children with hearing loss are dependent on several factors like degree of hearing loss, its configuration, nature, onset, type of hearing device the child if fitted with, age of fitting, benefit with the device and quality of training the child receives.

Of the factors that determine speech perception outcomes, type of hearing device used by the child is an important factor. Several studies have compared speech perception in children with hearing impairment using different types of devices [1–4]. Studies [5–7] indicate that the individuals with hearing impairment do not get access to all the information necessary for speech perception even after using their hearing devices. Also from these studies one can deduce that individuals having similar type and degree of hearing impairment and individuals using same type of hearing devices have variable outcomes as far as speech perception is concerned. Studies comparing speech perception benefits in children using hearing aids and those using cochlear implants [3, 4, 8–14] generally report better speech perception skills with cochlear implants.

Very few large-scale studies on speech perception in children with hearing loss have been reported in India. Hindi is a widely spoken and the national administrative language of India. There are 422,048,642 speakers using Hindi as first language, i.e. 41.03% of 1028.61 million population uses Hindi as first language [15]. Due to this, the clinical population using Hindi language is very frequently encountered in audiology clinics across the country. However, there is a scarcity of published literature reporting speech perception skills such as pattern perception, bisyllabic word identification, monosyllabic word identification, minimal pair identification and sentence perception in Hindi children with hearing loss.

Aim

The present study aims to compare the following speech perception skills in 6 to 8-year-old Hindi children with pre-lingual hearing impairment using bilateral BTE hearing aids and those using unilateral CI: (a) Pattern perception, (b) Bisyllabic word identification, (c) Monosyllabic word identification, (d) Word identification in minimal pairs and (e) Sentence identification.

Materials and Methods

The protocol for the study was approved by Ethics Committee of AYJNISHD(D) and data was collected at the institute. All procedures were in strict adherence to the protocol.

Participants

Table 1 shows the details of participants included in the study. A total of 43 participants were included from the local special schools for children with hearing impairment, hospitals, CI centres and from among clients coming to the institute. Children in the age range of 6–8 years with congenital severe to profound hearing loss and belonging to Hindi speaking backgrounds (Hindi spoken at home and school) were included. Children using appropriately fitted trimmer digital or digitally programmable BTE hearing aids bilaterally or unilateral cochlear implant for minimum duration of 2 years consistently, with aided or CI assisted PTA (500, 1000 and 2000 Hz) of 60 dB HL or better (lower) in at least one of the ears were included. Children with abnormal otoscopic findings, history of middle ear infections, multiple impairments and from multilingual background were excluded from the study.

Table 1.

Details of participants included in the study

	Age range (years)	No	Sex	Average PTA (dB HL)		Average Aided/CI assisted PTA (dB HL)		Mean duration of usage (years)
				Right	Left	Right	Left	Right	Left
Group-1 (HA)	Subgroup-1 6–7 (Mean-6.20)	10	5 M, 5 F	> 95.16 (80–> 108.3)	> 94.17 (85–> 111.67)	57.49 (40–86.67)	51.83 (45–60)	2.65 (2–4)	2.55 (2–4)
	Subgroup-2 7–8 (Mean-7.27)	12	7 M, 5 F	> 103.58 (93.3–> 118.3)	> 101.25 (90–113.33)	63.75 (50–78.33)	58.44 (43.33–63.33)	3.28 (2–6)	3.28 (2–6)
Group-2 (CI)	Subgroup-1 6–7 (Mean-6.11)	10	7 M, 3 F	> 103.96 (90–> 115)	> 107.56 (96.67–> 118)	38.66 (31.6–48.33)		2.69 (2–3)
	Subgroup-2 7-8 (Mean-7.39)	11	8 M, 3 F	> 113.3 (93.33–> 120	> 113.15 (105–> 120	42.42 (30–55)		2.74 (2–6)

Stimulus material

The study aimed at assessing speech perception skills at different hierarchical levels using forced-choice task with a picture-pointing response mode. Stimuli for each skill were chosen taking into consideration that they should be familiar to children with severe to profound hearing impairment in the age of 6–8 years and also picturable. For this purpose, stimulus items were selected from the curricular books used with this age group in schools for hearing impaired as well as State Education Board syllabus. The stimulus items for each task were chosen carefully taking into consideration factors such as requirements of each task, characteristics of Hindi language, the vocabulary of the children to be tested and the pictorial representation of the stimuli.

To select the requisite number of stimuli for each task, a pool of words and sentences was collated and given to parents of ten children with severe to profound hearing impairment in the age range of 6–8 years for familiarity rating. Parents rated items as very familiar, somewhat familiar, and not familiar. The items that were very familiar as per all parents were included as stimuli. Some stimulus words had alternatives that were commonly used by the children and parents gave their suggestions to use these alternatives; e.g. for /ænΛk/, /viman/ for , /motΛr/ for /gadi/, etc. Some words like ball, pen, bat, bus, etc. were used, though in English, as they came out to be very familiar in standard usage by children.

For the first four tasks, i.e. pattern perception, bisyllabic word identification, monosyllabic word identification and minimal pair identification the stimuli used were words, while for the fifth task, stimuli used were sentences. After familiarity rating of stimulus items, appropriate pictures were selected, mainly from Internet sources (without copyright issue) and by photographing some objects. For sentences, pictures were mainly clicked using a digital Sony camera T900 and some of the pictures were taken from commercially available clip art databases. Microsoft Office Picture Manager was used to perform requisite editing of the photographs. Microsoft Power-point was used for preparation of test plates. For sentence identification, to evaluate the suitability of each sentence-picture pair, these pairs were given to two ASLP and two special educators for validation. The sentence-picture pairs that had minimum of 75% approval were appropriately chosen for the test. The selected pictures were then administered on five children with pre-lingual severe to profound hearing impairment for picture naming. The pictures that were not identified correctly were replaced. Picture plates and practice plates were constructed for each of the tasks by using the selected pictures. After arranging the pictures on the plates, colour prints were taken on hard photo papers (A4 size) and arranged properly in a folder.

Table 2 shows the number of picture plates and the total number of stimulus items used for each of the five tasks:

1. For pattern perception, 24 pictures were selected and placed on two test plates containing 12 pictures each (4 monosyllabic, 4 bisyllabic and 4 multisyllabic), e.g. Monosyllabic: /kek/ /kan/, Bisyllabic: Multisyllabic: . On each test plate, out of 12 items, 6 were target items and 6 were foils.

2. Similarly, for bi-syllabic word identification, 24 bi-syllabic words were chosen with 12 words (6 target items and 6 foils) on each of the two plates, e.g. /tivi/, /ghoda/, /kela/, /chaku/ etc.

3. Monosyllabic word identification had two picture plates; for the first plate of this subtest 12 very familiar monosyllabic words (6 target items and 6 foils) starting with /b/ consonant were selected e.g. /bΛs/, /bæt/, /bæg/, /bed/, etc. and for the second plate 9 monosyllabic words (6 target items and 3 foils) with /p/ consonant were selected, e.g. /pen/, /pet/, /pao/, /pær/ etc. The pictures for the first three tasks were randomly placed on A3 sized paper due to higher number of test items on one plate.

4. For minimal pair identification, 42 test plates were made each with 4 items on it (2 target items and 2 foils), labeled (a), (b), (c), (d), e.g. /mala/, /kala/, /nala/, , here /mala/- /kala/ is the target pair assessing nasal/voiceless stop contrast; /bΛs/, , /rΛs/, /hΛs/, here is the target pair assessing place of articulation.

5. For sentence identification, 40 picture plates depicting four sentences on each plate were prepared, each sentence labelled (a), (b), (c), (d) (1 target sentence and 3 foils). Sentences are varying on either nouns, verbs, adjectives or prepositions, keeping other parts of the sentence constant. These were used in various combinations to make different sentence types such as subject, object, verb, adjective, subject + object, subject + verb, subject + adjective, subject + preposition, adjective + object, adjective + adjective and object + preposition; E.g. for Subject; for Object; for Subject + Verb; for Object + preposition; etc.

Table 2.

Number of picture plates and stimulus items for each task

Sr.	Task	No. of picture plates	No. of stimulus items/total score obtainable
1	Pattern perception	2	12
2	Bisyllabic word identification	2	12
3	Monosyllabic word identification	2	12
4	Minimal pair identification	42	84
5	Sentence	40	40

After the pictures were selected and test plates were prepared, pilot study was done on four 6–8 year old children with severe to profound hearing impairment to ensure familiarity and non-ambiguity of the pictures, also that the children could attend adequately to test plates having a large number of pictures and the response time of the children to facilitate estimation of the required inter-stimulus interval. The selected speech stimulus material was then recorded by a male speaker in a professional recording studio by a sound engineer using Neundo Version 4.0. An interval of 6 s was maintained between two items on each of the test lists. This was to provide sufficient response time to the participant.

Procedure

Consent was obtained from the parents for participation in the study after explaining the procedure. A case history was taken in line with the inclusion and exclusion criteria mentioned. Otoscopic examination with a Welch Allyn otoscope was undertaken to rule out cerumen, ear discharge, or any apparent abnormality. Immittance testing was done using Grason-Staedler Inc. (GSI) Tympstar middle ear analyser to rule out middle ear pathologies and ensure that all the participants had type ‘A’ tympanograms. Acoustic reflex threshold screening at 500, 1000, 2000 and 4000 Hz was done to ensure that acoustic reflexes were absent. Behavioural pure tone thresholds were obtained in a standard sound treated one room setup; with noise levels within permissible limits, as per American National Standards Institute [16]. Voyager 522 (Madsen Electronics) one and a half channel audiometer with TDH 49 earphones and Radioear B-71 bone vibrator (specification for audiometers ANSI S3.6-1989) was used. Pure tone air conduction (AC) thresholds were determined for octave frequencies from 250 Hz to 8000 Hz and for bone conduction (BC) at octave frequencies from 250 Hz to 4000 Hz using the modified Hughson and Westlake procedure [17]. Behavioural aided and CI assisted thresholds were obtained in a two-room setup using Interacoustics Diagnostic Audiometer AD229 in sound field conditions. Thresholds were obtained for octave frequencies from 250 to 4000 Hz using warble tones and speech, speech noise and 6-sounds of the Ling test. The testing was done separately for each of the two ears with hearing aid or cochlear implant on, using the modified Hughson and Westlake procedure [17].

Speech Perception Testing

The recorded stimulus material was administered via single channel Interacoustics Diagnostic Audiometer AD229. From the headphone output of a laptop computer a connection was made to the CD input of the audiometer. The connections were made using a stereo cable with 3.5 mm RC pins. Testing was done in a sound field setup with stimuli presented at 70 dB HL with child’s hearing device placed on. Each child was comfortably seated on a chair, with two loudspeakers placed 45° at the left and right of the midline, at 1-m distance. For subjects using hearing aids, the speaker on the side of the ear with better aided average thresholds was used to present the stimuli, while for children with cochlear implants; the speaker on the side of the implanted ear was used to present the stimuli. Each child was instructed as follows: “You will hear words/sentences from the loudspeaker and will be shown pictures in front of you, listen carefully to what is said and point to the respective picture.” A practice plate of 6 words was administered first in similar condition which was not scored. It only formed a basis to judge whether the participant had understood the task and to give a sample of what could be expected. The participant’s responses were recorded on the score sheet. For pattern perception, score of ‘1’ was given for correct word identification, ‘0.5’ was given for identifying correct pattern and ‘0’ for wrong response. For all other tasks, ‘1’ was given for correct response and ‘0’ was given for a wrong response. The total time taken for the entire procedure (including preliminary tests) was approximately 45 min per participant. Reinforcer was given to the subject at the end of the testing in the form of sticker, chocolate, etc. The findings were communicated to the parents.

Results

Out of the 43 participants, one subject who was a hearing aid (7–8 years) user could not perform the minimal pair identification task and hence his scores for the same were not included in analysis. The mean scores, SD, range and maximum obtainable scores for the five speech perception tasks for both participant groups are shown in Table 3 and the corresponding percentage scores are shown in Fig. 1.

Table 3.

The Mean Scores, SD and Range of Scores for each task

	Age (years)	Max. score	Group 1—HA				Group 2—CI
			N	Mean	Range	SD	N	Mean	Range	SD
Pattern perception	6–7	12	10	7.45	1.5–11.5	± 2.77	10	8.95	2.5–11.5	± 3.14
	7–8	12	12	5.92	3–11.5	± 2.61	11	8	3–12	± 3.18
Bisyllabic word identification	6–7	12	10	6.1	0–10	± 2.92	10	7.9	1–12	± 3.9
	7–8	12	12	3.5	1–9	± 2.65	11	6.18	1–12	± 4.33
Monosyllabic word identification	6–7	12	10	4.2	0–9	± 3.26	10	6.5	1–11	± 2.83
	7–8	12	12	3.75	1–12	± 3.31	11	5.73	3–11	± 2.61
Minimal pair identification	6–7	84	10	42.6	20–55	± 10.3	10	50.9	29–78	± 16.05
	7–8	84	11	39.64	20–52	± 9.38	11	47.27	30–75	± 14.30
Sentence identification	6–7	40	10	17.4	8–22	± 4.57	10	26	11–33	± 6.566
	7–8	40	12	15.92	9–28	± 4.98	11	23.73	13–33	± 9.931

Fig. 1.

Percentage of scores obtained by the two participant groups (HA and CI) for different levels of speech perception

It can be seen from Table 3 that participants from the HA group did not obtain ceiling scores for any of the tasks, except for one participant who obtained ceiling score on monosyllabic word identification task. The 6 to 7-year-old participants have higher mean scores as compared to the 7 to 8-year olds for all the tasks. It is further evident from the percentage scores that for both age ranges in the HA group, pattern perception was the highest scoring task. The difference between the two age groups was with reference to the bisyllabic word identification task, which was the least scoring for 7 to 8-year group, but second highest in scoring for the 6 to 7-year old group.

Among the CI group, one participant obtained ceiling scores for the tasks of pattern perception and bisyllabic word identification, but not for the other three tasks. Again, the 6 to 7-year-old sub-group has higher mean scores as compared to 7 to 8-year-old sub-group. Pattern perception is the highest scoring task for both 6 to 7 and 7 to 8-year-old CI users, while monosyllabic word identification is the least scoring task. As with the HA group, bisyllabic identification task is more difficult for the 7 to 8-year sub-group as compared to the 6 to 7-year sub-group.

Age Differences

There was no significant difference in the two age ranges i.e. 6–7 and 7–8 years in all the five tasks. However, on closely perusing the mean scores obtained by the two age groups, a consistent pattern is observed, wherein the younger group is obtaining higher scores as compared to the older age group. This pattern is seen for both, the hearing aid group as well as the cochlear implant group.

Device Differences

The difference in scores of HA group and CI group was compared using t-tests for each of the five tasks, the results of which are shown in Table 4. The differences were statistically significant for the sentence identification task for both 6–7 and 7–8 year sub-groups, where children using CI performed significantly better than children using HA. The differences were not statistically significant for the other 4 tasks.

Table 4.

Results of the t test for significance of difference between the two participant groups

Task	Age range	t value	df	p value
Pattern perception	6–7 years	− 1.132	18	0.272
	7–8 years	− 1.72	21	0.099
Bisyllabic word identification	6–7 years	− 1.168	18	0.258
	7–8 years	− 1.77	21	0.095
Monosyllabic word identification	6–7 years	− 1.566	18	0.135
	7–8 years	− 1.74	21	0.097
Sentence recognition	6–7 years	− 3.398	18	0.003**
	7–8 years	− 2.351	21	0.033*
Minimal pair identification	6–7 years	− 1.378	18	0.185
	7–8 years	− 1.481	20	0.154

*Significant at the 0.05 level (1-tailed)

**Significant at the 0.01 level (1-tailed)

Discussion

It is the first time that speech perception evaluation was done using recorded stimuli for Hindi children with hearing impairment. The first three tasks, pattern perception, bisyllabic word identification and monosyllabic word identification, parallel the standard form of Early Speech Perception test ESP [18], except a variation in Test 2 was required because spondaic words are not present in Hindi. The material included recorded speech stimuli presented from a computer attached to the audiometer in order to maintain consistency in presentation [19, 20]. Zheng et al. used a recorded administration procedure during the development of Mandarin ESP for 2–5-year-old developmentally normal children. It was observed that the recorded test also can be performed on children using hearing aids and CI successfully [21].

The stimulus material was constructed as a closed-set task. Most open-set tests are not appropriate for young paediatric populations because open-set tests carry higher perceptual, linguistic and memory demands than closed-set tests [22]. Thus, for children who cannot be tested with open test speech materials, material such as the ones used in the current study is a viable option. The response format used for administration of all the stimulus material was picture pointing task. Many children with hearing loss are at risk for language and speech production problems, the use of nonverbal response modes such as picture pointing or the use of manipulatives is considered as a good option. Therefore, any speech perception tests for paediatric populations e.g., Word Intelligibility by Picture Identification test; [23], especially children with hearing loss and other special needs, use closed set stimuli and picture pointing responses. All the children in 6-8 years (target population) with severe to profound hearing loss could perform all the five closed-set tasks without significant difficulty.

The results of the statistical analysis showed no statistically significant difference between the scores of 6–7 and 7–8-year participants for both HA and CI groups. It was however seen that the scores of the 6–7 group were consistently higher than the 7–8-year group for both the HA and CI users. This can be attributed to factors like residual hearing, benefit from the hearing device, duration of usage, as described in Table 1. In the group of hearing aid users, 4 out of 10 subjects in 6–7 years age range had severe hearing loss in one ear and profound on the other ear and 2 subjects had bilateral severe hearing loss. In the older group, out of 12 children, 1 had severe hearing loss in one ear and profound hearing loss on the other, and 1 had bilateral severe hearing loss. The mean bilateral PTA for the younger hearing aid group was > 94.67 dB HL, while that for the older hearing aid group was > 102.42 dB HL. This indicates that the younger group had more subjects with residual hearing as compared to the older group. Also, the children in the younger group had better hearing with the hearing aids than the older group which correlates with their better performance on HTSP. Speech perception testing is used by professionals to find the benefit of hearing aids, cochlear implants, FM devices or any combination of technology [24]. It is said that the better the aided threshold, the better is the performance on the speech perception test [25]. There is a correlation between the average aided thresholds and performance on the stimuli, lower aided average (better hearing) relates with higher scores on all tasks. This validates the stimulus items used. In the group of cochlear implant users, out of the 10 younger subjects, seven had experience of usage of CI for 3 years and three children had for 2 years. This means majority of the subjects were implanted at the age of approximately 3 years. On the other hand, in the older group, eight of the subjects had experience with CI for 2 years and three children had the experience of 3, 5 and 6 years. Majority of the subjects in this age group were late implantees, implanted approximately by the age of 5 years. Several studies have documented that the time duration for which the child has been using the amplification device is an important factor that influences speech perception and production skills. Anderson et al. [26] reported the results of 41 children with pre-lingual hearing impairment with a minimum of 3 years’ CI experience, which suggested that cochlear-implanted children develop open-set speech recognition soon after implantation, and these skills develop over a long period of time, highlighting the need for continued therapy to maximize listening and learning.

When children using HA were compared to those using CI, it was found that children using a cochlear implant performed significantly better than children using hearing aids. Similar results were reported by Somers [9] and Ranjan [12]. Also, higher scores are reported on ESP adapted to Hebrew language for cochlear implant users as compared to hearing aid users [27]. The results of documented and validated studies corroborate the results of the present study that in the group of profound hearing losses, children using cochlear implant perform better than those using hearing aids.

Conclusions

Like several earlier studies, the current study supports the view that children using CI perform better that those fitted with HA. The recorded stimuli developed during the study can be used for other purposes like: ascertaining candidacy for cochlear implant surgery, pre- and post-operative CI assessment, comparison of benefit from cochlear implant or hearing aid, prognosis of the child using HA or CI and planning of therapy goals. Considering the dearth of standardized test material in Indian languages, the stimulus material developed for this study needs to be standardized for use with normal hearing as well as children with hearing loss from a wide age range. It also would be helpful to correlate the speech perception skills to the outcomes to children’s language, speech, visual perception and cognitive skills.

Compliance with ethical standards

Conflict of interest

None.