Consonant Development in Pediatric Cochlear Implant Users Who Were Implanted Before 30 Months of Age

Abstract

This study provided a yearly record of consonant development for the initial 4 years of cochlear implant (CI) use and established a precedent for using a standardized articulation test, the Goldman–Fristoe Test of Articulation—2 (Goldman, R., & Fristoe, M. [2000]. Goldman–Fristoe Test of Articulation—2. Circle Pines, MN: American Guidance Services). The study used CI age as a referent for 32 children who received their CI before 30 months of age. Consonants produced by 70% of the children were listed, as were the most common error types, which were consonant omissions and substitutions. Using consonant repertoire lists and standard scores, the study revealed that children with CIs had acquisition patterns that were similar to their peers when the duration of CI experience was similar to the chronological age norms of typically developing children. The results revealed that CI users need time to coordinate their articulatory organizing principles with the input they receive from their CI. It is appropriate to use length of CI use as a proxy for chronological age during the first 4 years when comparing articulation development with hearing peers.

A major issue in the field of cochlear implants (CIs) is whether children with CIs will acquire speech and language skills comparable to typically developing children. Research that focused on speech sound development in children with CIs has attempted to address this issue by evaluating the overall accuracy (e.g., Tobey, Geers, Brenner, Altuna, & Gabbert, 2003; Tomblin, Peng, Spencer, & Lu, 2008), speech sound inventory (e.g., Serry & Blamey, 1999), and/or error patterns (e.g., Warner-Czyz & Davis, 2008) in this population from spontaneous language samples or elicited productions. Though the previous studies were informative in the speech sound development of children with CIs, the applicability of the findings to clinical work remains an open question because of the methodological limitations. For instance, Blamey and colleagues (Blamey, Barry, & Jacq, 2001; Serry & Blamey, 1999) documented the time course of mastery of Australian English phonemes in nine children who were implanted between the ages of 1;7 (years;months) and 5;2 from preimplantation to 6 years postimplantation. The findings from Blamey et al. may not be generalized unambiguously to the current population of CI users because these children tended to receive cochlear implantation when they were older than current children. To date, clinicians who work on the speech sound development in children with CIs do not have sufficient information obtained from children with CIs representative of the current population to guide their clinical decision making (e.g., Can/Should the clinician target /s/ in a child with 2 years of CI experience?). As an initial step to address this issue, the present study reported the yearly development of consonants in a group of children who were implanted before 30 months of age over a 4-year span by using a standardized test (i.e., Goldman–Fristoe Test of Articulation—2 (GFTA-2); Goldman & Fristoe, 2000) with two major objectives. First, we evaluated the overall development of consonants by comparing children’s performance based on their chronological age and length of experience in using cochlear implant devices (i.e., CI age). Second, we documented acquisition of specific consonants at the word-initial and word-final positions separately with reference to children’s CI age by using a 70% criterion (i.e., 70% of children with a particular CI age produced a specific sound). In what follows, we first review the studies of consonant development in children with typical development and children with CIs and then present the scope of the current study.

Consonant Development in Typical Children

Several cross-sectional, normative studies had been conducted to document the age of acquisition of English consonants in children with typical development (e.g., Prather, Hedrick, & Kern, 1975; Smit, Hand, Freilinger, Berthal, & Bird, 1990; Templin, 1957). Because of the difference in the criteria of mastery (e.g., 75% criterion in Templin; 90% criterion in Smit et al.), these studies may show discrepant ages of acquisition for specific consonants, but there were some general trends across studies. In terms of place of articulation, anterior sounds (e.g., /p/) tended to be acquired earlier than posterior sounds (e.g., /k/). In terms of manner of articulation, stops (e.g., /t/) and nasals (e.g., /n/) tended to be acquired earlier than affricates (e.g., /tʃ/) and fricatives (e.g., /s/). Overall, most typical children were able to produce /p, b, m, n, h, w/ by age 3 and produce all English consonants by age 8. Factors that accounted for the acquisition order of English consonants may include, but are not limited to, articulatory complexity and the environment input (e.g., token frequency) of a given sound (Stokes & Surendran, 2005).

The normative studies of consonant acquisition offered clinicians a reference as to which consonants we might expect children of a particular chronological age to produce and hence guided the clinical decision making, such as setting up therapy goals (Bauman-Waengler, 2012). However, comparing children with CIs to the normative data based on their chronological ages may not always be appropriate given the deprivation of auditory input in the early year(s) in these children. Children with CIs do not receive significant auditory input from the ambient language until they are implanted, which does not typically happen until they are at least 12 months of age. The deprivation of auditory input also leads to reduced opportunities for using their articulators to practice the sounds they hear, leading to the delayed development of oromotor coordination (Ertmer & Goffman, 2011; Oller & Eilers, 1988; Tye-Murray, 1992). Given the unequal perceptual and oromotor development, it may not be legitimate to directly compare children with CIs to the normative data based on their chronological age. An alternative is to compare children with CIs to the normative data based on their length of experience in using the CI devices (i.e., CI age). For instance, consonant production of a 5-year-old child with a CI who has 3 years of CI experience would be compared to that of typical 3-year-olds in the norm (i.e., typical children who have matched hearing experience). However, given that children with CIs receive electrical signals rather than acoustical signals (Wilson, 2006), it is not clear whether children with CIs acquire consonants in a similar pattern to typical children who have matched hearing experience. This question must be answered before we can validate the CI age comparison, but the existing studies, as are reviewed below, do not provide clear answers.

Consonant Development in Children With Cochlear Implants

Studies that focused on overall accuracy.

The overall production accuracy of consonants in children with CIs has been documented by using spontaneous language samples and elicited production (e.g., standardized tests) cross-sectionally and longitudinally. Tobey, Geers, Brenner, Altuna, and Gabbert (2003) investigated percent consonant correct in 108 English-speaking, 8- to 9-year-olds who received CIs between the ages of 1;8 and 5;4 and had an average of 5.5 years of experience in using CI devices. The participants produced consonants with 68% accuracy in the conversational language samples, which is lower than typical 5-year-olds (94%) with reference to the norm of Shriberg, Austin, Lewis, McSweeny, and Wilson (1997). Tomblin, Peng, Spencer, and Lu (2008) characterized speech sound development in 27 pediatric CI recipients who were implanted between the ages of 2;7 and 7;5. Percent correct of phonemes (i.e., consonants and vowels) in a story-retell task was computed annually for 10 years from most of the children. As a group, children with CIs showed steady improvement of phoneme accuracy during the first 6 years after implantations, but the improvement slowed down after 6 years of CI experience, leading to an asymptote around 81% accuracy. Taken together, these two studies showed that, as a group, pediatric CI recipients may show improvement in consonant production over time but did not reach typical levels of consonant accuracy after more than 5 years of CI experience. However, because these studies included children with a wide range of age of implantation, it is not clear whether the findings can be generalized to the current population who typically received CIs before 24 months of age.

Connor, Craig, Raudenbush, Heavner, and Zwolan (2006) investigated the effect of age of implantation on consonant production accuracy in 100 children from four groups with different ages of implantation: 1–2.5 years, 2.6–3.5 years, 3.6–7.0 years, and 7.1–10.0 years. Each child received a standardized test of phonology yearly during the first 5 years after implantation. The result most relevant to the current study was that children who received CIs before 30 months showed early bursts of growth in consonant production accuracy, whereas the early bursts were not found in those who received CIs at older ages. At 5 years postimplantation, the predicted consonant production accuracy was about 85% for children who received CIs before the age of 30 months and was 70% for children who received CIs between the ages of 30 and 42 months. These findings stressed the importance of controlling the factor of age of implantation in examining speech sound development of children with CIs. However, because the standard scores were not reported, it was not possible to determine whether the consonant production accuracy in children who were implanted before 30 months of age was within typical range (e.g., standard score ≥85).

Schorr, Roth, and Fox (2008) evaluated the consonant production of 39 children who were implanted between the ages of 1;3 and 8;2 and had an average of 5.5 years of CI experience (range: 1.7–11.7 years) at the time of testing. Using the Goldman–Fristoe Test of Articulation—2, Schorr et al. reported that the average standard score for this group was 93.54 (SD = 18.94), which was not significantly different from the typical group. In addition, 33 (85%) out of 39 children achieved a standard score that was within typical range (cutoff = 85). Given the wide range of age of implantation and length of CI experience of the participants, it was difficult to draw reliable conclusions from the findings. Nevertheless, this study did reveal that children with CIs had the potential to produce consonants at typical levels as measured by standardized tests.

In summary, children with CIs seemed to produce percent consonant correct at a lower level than the typical children who had similar hearing ages (Tobey et al., 2003), but the result was confounded by the wide range of age of implantation. Nevertheless, the accuracy studies that controlled for the age of implantation (e.g., Connor et al., 2006) may provide the clinicians with overall accuracy levels to track the progress of children with CIs over time. The accuracy studies, however, do not offer the information with reference to which consonants a child with a given amount of CI experience can typically produce.

Studies that focused on consonant repertoire.

Using conversational language samples, Warner-Czyz and Davis (2008) documented the emergence of consonant inventories of four children with CIs monthly for 6 months after they started to produce their first meaningful words. These children started to use the CI device between 1;0 and 1;10 and had 6–10 months of CI experience at the study onset. In terms of manners of articulation, stops and nasals occurred more frequently than fricatives. In terms of place of articulation, labials and coronals occurred more frequently than dorsals. Though this study revealed the general tendency of the emergence of sound categories, it did not document the timing for the mastery of specific sounds.

In a series of studies, Blamey and colleagues (Blamey, Barry, & Jacq, 2001; Serry & Blamey, 1999) documented the time course of the emergence and mastery of phonemes in nine Australian English-speaking children with CIs who were implanted between the ages of 2;6 and 5;2 from 3 months preimplantation to 6 years postimplantation. Conversational language samples were collected every 6 months (i.e., 11 language samples from each child). The age of mastery of phonemes was documented. The age of mastery of a given phoneme was the earliest CI age at which five or more children produced two or more correct tokens of that phoneme and at least 50% of all attempts at the phoneme to be correctly produced. Children with CIs mastered /p, b, m, w, d, n, l, j, ʃ, h/ (ordered by place of articulation) at 3 years postimplantation (i.e., CI age = 3); /f, v, r/ at 4 years postimplantation; /ð, ʒ, k, ŋ/ at 5 years postimplantation; and /g/ at 6 years postimplantation. Five consonants were still not mastered at 6 years postimplantation, including /θ, s, z, ʒ, tʃ/. Overall, the pattern of mastery was consistent with previous studies based on typical children (Prather et al., 1975; Smit et al., 1990): labials and coronals were acquired earlier than dorsals, and stops and nasals were mastered earlier than fricatives.

Though the studies of Blamey and colleagues (Blamey, Barry, & Jacq, 2001; Serry & Blamey, 1999) can potentially serve as a guideline for clinicians to track the progress of children with CIs and select the therapy goals, they are limited in some aspects. In addition to the concerns of small number of children and wide ranges of age of implantation, Blamey et al. used a set of criteria to define age of mastery that was substantially different from previous studies based on typical children (Prather et al., 1975; Smit et al., 1990; Templin, 1957). For instance, ages of acquisition were generally determined as the earliest age at which 75% of children produced a sound correctly at the word-initial and word-final positions (Smit, 1986; Templin, 1957). The mastery criterion that five (56%) out of nine children produced a given sound at least 50% correct in the studies of Blamey et al. was relatively low as compared to other studies, whereas Serry and Blamey (1999) stated that this decision was made because of the small sample size. In addition, Blamey et al. did not separate the age of mastery for a given sound at the word-initial and word-final positions (e.g., /p/ as inp ig and li p). This was probably because they used language samples in their analysis, which did not allow them to control the occurrence of a specific sound in different word positions. Thus, the criteria of Blamey et al. may have led to overestimation of the phonological development in children with CIs. For instance, /l/ was acquired by 6;0 in Templin (1957) but was acquired by 3;0 in Serry and Blamey (1999). Furthermore, the age of mastery could be confounded by token frequency of a given sound in the studies of Blamey et al. One of the criteria of mastery was that children had to produce two or more tokens of a given phoneme. Certain infrequently occurring sounds, such as /ʒ/, may not appear consistently in a conversational sample. When a sound like /ʒ/ had an older age of mastery, this could result from the low frequency of occurrence of that sound.

In summary, a few studies had attempted to document the development of consonant inventory in children with CIs, but they were limited in sample size, age of implantation of the participants, and/or the criteria of defining mastery/acquisition (Ertmer & Goffman, 2011). Thus, how long it takes the current population of CI users to acquire a specific consonant and whether they follow a typical developmental sequence of consonant acquisition remain open questions.

The Present Study

To address the issues of sample size, age of implantation of the participants, and the criteria of defining sound mastery, the present study used a standardized test (i.e., GFTA-2) to document consonant development longitudinally in a group of 32 children with CIs who were implanted before 30 months of age during the first 4 years after implantation. The use of GFTA-2 allowed us to control the word positions of a given sound and hence set up acquisition criteria commensurate with the previous studies based on typical children (e.g., Smit et al., 1990) to document the consonant repertoire by children’s CI age. The present study had three goals. First, we compared the performance of children with CIs to the norm in the standardized test based on their chronological age and their CI age whenever applicable. These comparisons allowed us to determine whether children with CIs produced consonants as accurately as their typical peers matched in either chronological age or CI age. Second, we provided a list of consonant sounds produced in word-initial and word-final position by 70% of children with CIs in the initial 4 years after implantation. This enabled us to examine the extent to which the developmental sequence of consonants in children with CIs was consistent with that in typical children. In addition, we provided a description of the most frequent errors made by children with CIs. By examining the development consonant repertoire and the error patterns, this study may provide information for clinicians as to whether a given child’s consonant production is “typical” of other CI users.

The research questions addressed were: Was the performance of children with CIs on GFTA-2 within typical range (cutoff standard score = 85) when they were compared to the norm by using their chronological age or CI age? Was the developmental sequence of consonants similar in children with CIs and their typical peers who were matched in hearing experience? What were the most common errors of consonant production in children with CIs? Given the differences in perceptual and oromotor development, we predicted that children with CIs would perform below typical range during the initial years after implantation when they were compared with the norm based on their chronological age. In addition, research has shown that the cochlear implant device may provide sufficient, though degraded, auditory information for children to develop typical speech and language, especially for those who were implanted at a young age (Connor et al., 2006; Nicholas & Geers, 2007). Thus, we predicted that children with CIs would perform within typical range when they were compared to the norm based on their CI age and that the development sequence of consonants would be similar to their typical peers who had matched hearing experience. Furthermore, to the best of our knowledge, no published research has presented the error patterns (i.e., omission, substitution, distortion) of consonant production in children with CIs. Given that typical children were more likely to omit a (singleton) consonant at the word-final than at the word-initial position (Bauman-Waengler, 2012; Dodd, Holm, Zhu, & Crosbie, 2003) and children with profound hearing loss (Smith, 1975) showed a similar trend, we predicted that omission errors in the children with CIs would be proportionally higher than substitution errors at the word-final position, but not at the word-initial position. Moreover, omission errors should decrease over time at both word-initial and word-final positions.

Method

Participants

The present study was part of a large, longitudinal study of children with prelingual deafness who received CIs. In the protocol, the children were seen multiple times during the first year after implantation and annually thereafter for device setting and follow-up and for data collection in the areas of speech and music perception as well as speech and language development. One part of the protocol was to use the GFTA-2 to evaluate children’s speech sound development, which was reported in the current study. To date, one other study has been published using GFTA-2 data collected just up through the first year of CI use (Walker & Bass-Ringdahl, 2007).

Thirty-two children (17 boys, 15 girls) with a unilateral CI participated in the present study. All children had prelingual, bilateral, profound hearing loss and had engaged in a trial period using hearing aids and did not detect speech at 65 dB HL or greater. The CI devices in these children were activated between the ages of 11 and 29 months (mean = 19 months, SD = 5 months). Two children, participants 14 and 27, continued to use a hearing aid in the opposite ear after they received the CI. Table 1 summarizes the demographic data for each child. All children came from monolingual English-speaking family and had typical motor and cognitive development as measured by Bayley Scales of Infant Development—Second Edition (Bayley, 1993). No concomitant developmental disabilities aside from the hearing loss were identified in any of the children throughout their time of participation in this study. All of the children were recruited in the Midwest states, including Iowa (26 children), Illinois (2 children), Missouri (1 child), North Dakota (1 child), and Wisconsin (2 children). They all received early intervention programs that provided an average of 60min of intervention per week. Intervention was based on the philosophy of total communication according to the parent report, which indicated that they used both spoken and sign language in communication.

Table 1 .

Demographic information

Child	Gender	CI eara	Device type/processing strategyb	Etiologyc	Age of ID (months)d	Age at CI connection (months)	Number of data points availablee
CI-1	F	R	NU2/ACE	Unknown	6	26	4 (12–48)
CI-2	F	L	NU1/ACE	Meningitis	11	15	2 (24, 36)
CI-3	M	R	NU1/ACE	Unknown	0	19	4 (12–48)
CI-4	M	R	NU1/ACE	Family history	0	12	4 (12–48)
CI-5	F	R	NU1/ACE	CX26	0	15	4 (12–48)
CI-6	M	L	NU1/ACE	Unknown	0	19	2 (24, 36)
CI-7	F	R	NU1/ACE	Family history	0	27	3 (12, 24, 36)
CI-8	F	L	AB7/CIS	Unknown	0	24	1 (36)
CI-9	M	R	NU1/ACE	CMV	0	14	4 (12–48)
CI-10	F	R	NU2/ACE	Family history	0	18	4 (12–48)
CI-11	M	R	NU2/ACE	Meningitis	13	16	3 (24, 36, 48)
CI-12	M	R	AB9/C6	Waardenburg	0	29	1 (24)
CI-13	F	R	NU7/ACE	Family history	0	18	3 (12, 24, 36)
CI-14	M	R	NU2/ACE	Unknown	0	23	1 (12)
CI-15	F	R	NU1/ACE	Unknown	0	23	3 (24, 36, 48)
CI-16	M	R	NU1/ACE	Unknown	0	11	2 (36, 48)
CI-17	F	R	NU2/ACE	Family history	0	26	2 (36, 48)
CI-18	F	L	NU1/ACE	Family history	0	12	4 (12–48)
CI-19	F	L	AB9/C6	Unknown	0	13	1 (24)
CI-20	M	L	NU7/ACE	Unknown	14	25	1 (12)
CI-21	M	R	NU1/ACE	Unknown	0	16	2 (24, 48)
CI-22	F	R	NU1/ACE	Unknown	0	14	1 (24)
CI-23	M	L	NU1/ACE	Family history	0	15	3 (12, 36, 48)
CI-24	M	R	NU2/ACE	Ototoxicity	10	19	3 (24, 36, 48)
CI-25	M	R	NU2/ACE	Unknown	0	19	2 (36, 48)
CI-26	M	R	NU1/ACE	Unknown	0	24	2 (24, 36)
CI-27	M	R	NU2/ACE	Family history	0	29	3 (12, 24, 36)
CI-28	F	L	NU1/ACE	Mondini	0	17	4 (12–48)
CI-29	M	L	NU1/ACE	Unknown	0	16	1 (48)
CI-30	F	R	NU2/ACE	CX26	0	18	2 (36, 48)
CI-31	F	R	NU2/ACE	Ushers type 1	0	18	2 (24, 48)
CI-32	M	R	NU1/ACE	Family history	0	13	2 (24, 48)
Mean					2	19	2.5
SD					5	5	1.5

^aCI ear, the ear that received cochlear implantation; L, left ear; R, right ear.

^bNU1, Nucleus Spectra; NU2, Nucleus Sprint; NU7, Nucleus Esprit 24; AB7, Advanced Bionics Clarion S-Series; AB9, Advanced Bionics Clarion Auria; ACE, ACE Processing Strategy; CIS, CIS Processing Strategy; C6, HiRes-S Processing Strategy.

^cCMV, cytomegalovirus; CX 26, Connexin 26.

^dAge of ID, age of identification.

^eThe number within the parentheses indicated the specific data points in which the GFTA-2 data were available: 12, 12 months postimplantation; 24, 24 months postimplantation; 36, 36 months postimplantation; 48, 48 months postimplantation.

Longitudinal data were collected from the participants as close to the following time points as possible: 12, 24, 36 and 48 months (±3 weeks) after the activation of the device. However, not every child participated at every time point. There were various reasons for missing data, including that some children were unable to complete the standardized articulation test at their first year postimplant appointment due to maturational factors; child factors such as fatigue, fitfulness, late/missed annual appointments; or time constraints that forced the examiner to curtail the number of tests given. The number of children who completed the GFTA-2 at each data point was 14, 21, 23, and 20, respectively. The mean chronological age was 2;11, 3;8, 4;8, and 5;7, respectively, at each data point.

Materials

The GFTA-2 evaluated the children’s abilities to produce sounds in citation forms (i.e., the sounds-in-words section) and in sentences (i.e., the sounds-in-sentences section). Because the GFTA-2 provided normative data for the sounds-in-words section but not for the sounds-in-sentences section, we only included the results from the sounds-in-words section in our analysis.

The sounds-in-words section contained 34 color pictures for eliciting children to produce words that included the target consonant singletons or consonant clusters. Eighteen out of 23 target consonant singletons were evaluated once at the word-initial, word-medial, and word-final positions. The consonants /w, h, j/ were tested only at the word-initial positions; /ð/, at the word-initial and word-medial positions; /ŋ/, at the word-medial and word-final positions. All of the 16 consonant clusters were two-consonant blends and were only tested once at the word-initial position.

Procedures

Each child was tested individually by an examiner in a quiet therapy room by following the administration procedures in GFTA-2. Two examiners were involved in test administration. The first examiner was a researcher specializing in infant perception and language development and had over 5 years of experience testing infants who were profoundly deaf. She tested all participants up to the age of 36 months. The second examiner (i.e., the first author) was a certified speech-language clinician who had over 15 years of experience working with children who were deaf and/or hard-of-hearing. She tested all participants who were over 36 months of age.

Each trial involved the examiner presenting a picture and required the child to talk about the picture by responding to the examiner’s question (e.g., What’s this? What do you call this?). If the child did not respond to the question, the examiner used the suggested cues that were specified on the stimulus book to prompt the child. The whole section was videorecorded for transcription.

Transcription and Data Analysis

The transcriptions were completed by the first author and two graduate students in speech language pathology. Before transcribing the children’s production in the GFTA-2 test, all of the three transcribers were asked to complete a transcription task that required them to transcribe sentences produced by children with CIs who were between the ages of 3 and 17 years. A point-to-point transcription reliability average of 79.05% (SD = 8.87%) was in place for the sentence task between the two research assistants and the first author before the GFTA-2 transcriptions had begun.

A child’s responses in the GFTA-2 test were transcribed by using broad transcription. The child’s production of the target sound was judged as correct or incorrect. We then conducted three analyses based on a child’s production: overall accuracy level, consonant repertoire, and error analysis.

Overall accuracy level.

Based on the scoring rules of GFTA-2, the total number of errors of consonant singletons and consonant clusters were tallied to yield a raw score for each child at each available data point. We then converted the child’s raw score to two standard scores with reference to the GFTA-2 norm: one based on his/her chronological age and the other based on his/her CI age. The CI age was defined as the time interval between the activation of the CI device and the receipt of the GFTA-2 test. It should be noted that because the youngest age in the norm was 24 months, the standard score based on the child’s CI age was not available at 12 months postimplantation.

Consonant repertoire.

We separately documented the consonant singletons 70% of children with CIs could produce at the word-initial and at word-final positions at each data point (i.e., 12, 24, 36, and 48 months postimplantation) and compared the current findings with the norm of Smit et al. (1990). This allowed us to determine how long it took for the majority of the children with CIs to acquire a given sound at word-initial and at word-final positions and to what extent the time course of consonant development in children with CIs was consistent with that in their typical peers who had matched hearing experience. We chose the norm of Smit et al. (1990) for comparison because the participants in their study were from the states of Iowa and Nebraska and were similar to the participants in the current study. Following Smit et al. (1990), we did not examine the development of consonants at word-medial positions.

However, there were a few methodological differences between the current study and the study of Smit et al. (1990). First, although Smit et al. (1990) used the 75% criterion (i.e., 75% of children could produce a given consonant at a specific word position) to determine whether a consonant was acquired/mastered by the majority of the children, we used the 70% criterion. This decision was made because we had a smaller sample size, but we did not want to use a criterion (e.g., 56%; in Serry & Blamey, 1999) that could potentially lead to overestimation of children’s phonological ability. Second, although Smit et al. (1990) presented data for males and females separately, we were unable to do so because of the small sample size. In order to compare our finding with the norm, we took the average of the male/female data for each sound presented in the Smit et al. (1990) study. For instance, 88% of the females produced /ʃ/ at the word-initial position at 3;0, but only 70% of the males produced it correctly. Thus, at age 4;0 when the data were weighted for both males and females, 77% of the children produced /ʃ/ accurately, which was above the 75% criterion of Smit et al. (1990). In our comparison (see Table 2) following the weighted percentages presented by Smit et al. (1990), /ʃ/ at the word-initial position was listed as a sound that was mastered at 4;0 in typical children. Third, the norm of Smit et al. (1990) sampled children at a 6-month age interval, whereas we followed up the children with CIs on a yearly basis. For sounds that were mastered at 3;6 in the norm (e.g., /j/ and /f/ at word-initial positions), we rounded up the age and listed these sounds as those mastered at 4;0 in typical children.

Table 2 .

Consonant repertoire at the word-initial position in children with CIs and their typical peers who had matched hearing experience by data point

CI age/hearing experience	Children with CIs	Typical children who have matched length of hearing experience*
12 months		Information not available
24 months (n = 23)	/b/, /d/	Information not available
	/m/, /n/
	/w/
	/h/
36 months (n = 23)	/p/, /b/, /t/, /d/,/k/, /g/	/p/, /b/, /t/, /d/, k/, /g/
	/m/, /n/	/m/, /n/
	/w/	/w/
	/f/, /s/, /ʃ/, /h/	/h/
48 months (n = 20)	/p/, /b/, /t/, /d/,/k/, /g/	/p/, /b/, /t/, /d/,/k/, /g/
	/m/, /n/	/m/, /n/
	/w/	/w/, /j/
	/f/, /s/, /ʃ/, /h/	/f/, /v/, /s/, /ʃ/, /h/
	/tʃ/	/dʒ/

*Taken from Smit et al. (1990).

Error analysis.

Children’s production errors of consonant singletons were classified into three categories: omissions, substitutions, and distortions (Bauman-Waegler, 2012; Smith, 1975). Omissions were the errors in which a target consonant was not produced (e.g., “house” /haʊs/ → /haʊ/). Substitutions were the errors in which a target consonant was replaced by another sound in English (e.g., “cup” /kʌp/ → /tʌp/). Distortions were the errors in which a target consonant was pronounced as a non-English sound (e.g., “chair” /t ʃɛr/ → /ɕɛr/). We computed the percentage of each type of errors for each word position at each data point by collapsing the data from all children. We also listed the most common substitution errors for each sound whenever applicable in order to document the substitution errors that were “typical” of children with CIs.

Reliability

A subset of 25% of the data was re-transcribed to check transcription reliability for the GFTA-2 test. The three transcribers (two students and the author) had to agree whether the target phoneme response was correct or incorrect. The mean level of agreement for correct responses for the three transcribers was 95.95%. Inter-rater reliability between transcriber A and the first author was 94.6%, whereas inter-rater reliability between transcriber B and the first author was 97.3%.

Results

Overall Accuracy Level

Table 3 presents the mean standard scores of children with CIs in GFTA-2 based on their chronological age or CI age at each data point. One-sample t-tests were conducted to determine whether the standard scores were significantly below 85, the cutoff score of typical range (Schorr et al., 2008). The results showed that the standard scores based on children’s chronological ages were significantly lower than 85 at 36 months postimplantation: t(22) = −1.94, p = 0.03, one-tailed, d = 0.83. There were no other significant results for the comparisons based on children’s chronological age, ts < 1.48, ps > 0.08, one-tailed, ds < 0.78. These findings indicated that children with CIs, as a group, can produce consonant singletons and clusters in words within typical range when they were compared to their typical peers who had matched chronological ages. However, there was a wide range of individual differences. If we use a standard score of 85 as a cutoff, about 50% of the children with CIs performed below the typical range at 12, 24, and 36 months postimplantation. The percentage of children who scored below the typical range dropped to 35% at 48 months postimplantation. This seems to suggest that more children caught up with their typical age-matched peers after 4 years of CI use.

Table 3 .

Children’s standard scores in the GFTA-2 based on the chronological age and the CI age by data point

CI age in monthsa	Scores based on chronological age				Scores based on CI age
	Mean	SD	Range	% typicalb	Mean	SD	Range	% typical
12 (n = 14)	79.57	13.71	48–98	50% (7/14)	n/ac	n/a	n/a	n/a
24 (n = 21)	77.86	23.35	40–115	52% (11/21)	94.76	17.37	65–126	67% (14/21)
36 (n = 23)	75.48	23.57	40–113	48% (11/23)	89.74	21.43	54–124	65% (15/23)
48 (n = 20)	82.45	26.08	40–114	65% (13/20)	93.45	25.11	40–122	70% (14/20)

^aThe number in parentheses indicated the number of children at each data point.

^b% typical, percentage of children who scored within typical range (i.e., standard score ≥85).

^cn/a, not applicable.

In contrast, the standard scores based on children’s CI ages were significantly higher than 85 at 24 months postimplantation: t(20) = 2.58, p = 0.01, one-tailed, d = 1.15. There were no other significant results for the comparisons based on children’s CI age: ts < 1.50, ps > 0.07, one-tailed, ds < 0.45. Approximately 65–70% of children with CIs produced consonants in words within typical range at 24, 36, and 48 months postimplantation when they were compared to their typical peers who had matched hearing experience. In addition, the percentages of children with CIs who scored within the typical range were similar in the chronological age (65%) and CI age (70%) comparisons at 48 months postimplantation.

Consonant Repertoire

Description of consonant development in the CI group.

Tables 3 and 4 present a list of the consonant singletons produced by 70% of the children with CIs (left column) by their CI age (i.e., hearing experience) at word-initial and at word-final positions, respectively. After 12 months of CI use, there were no consonants produced at the initial position of words by 70% of children with CI, although in the final position the bilabial, unvoiced stop /p/ was produced by 70% of the children.

Table 4 .

Consonant repertoire at the word-final position in children with CIs and their typical peers who had matched hearing experience by data point

CI age/hearing experience	Children with CIs	Typical children who have matched length of hearing experience*
12 months	/p/	Information not available
24 months (n = 23)	/p/	Information not available
	/ʃ/
36 months (n = 23)	/p/, /b/, /t/, /k/, /g/	/p/, /b/, /t/, /d/, /k/, /g/
	/m/, / n/	/m/, /n/
	/l/	/f/
	/f/, /s/, /ʃ/
	/tʃ/
48 months (n = 20)	/p/, /b/, /t/, /k/, /g/	/p/, /b/, /t/, /d/, /k/, /g/
	/m/, /n/	/m/, /n/, /ŋ/
	/l/	/f/, /v/, /s/
	/f/, /s/, /ʃ/	/ tʃ/, /dʒ/
	/tʃ/

*Taken from Smit et al. (1990).

After 24 months of CI use, six consonants were produced at the word-initial position by 70% of the children, including /b, d, m, n, w, h/. These included two voiced stops, which had a labial/alveolar distinction, a bilabial and alveolar nasal, one voiced, bilabial glide, and a voiceless postalveolar fricative. At the word-final position, two voiceless consonants were produced by 70% of the children, including /p, ʃ/.

After 36 months of CI use, 70% of the children produced /p, b, t, d, k, g, m, n, w, h, f, s, ʃ/ at the word-initial position, an addition of seven new consonants (i.e., the bold ones). Thus, the inventory expanded to include voice/voiceless stop distinctions in labial, alveolar, and velar positions. The added fricatives were all voiceless at the places of labial-dental, alveolar, or palatal. At the word-final position, 13 consonants were produced by 70% of the children, including /p, b, t, k, g, m, n, l, f, ʃ, tʃ/, adding the voiced stop distinctions, the voiceless palatal fricatives, and the voiceless alveo-palatal affricate.

After 48 months of CI use in the word-initial position, 14 consonants were produced by 70% of the children, expanding the repertoire by one voiceless alveo-palatal affricate /tʃ/. At the word-final position, 12 consonants were produced by 70% of the children, with no elaboration of features noted.

Between 24 and 36 months of CI use, children with CIs increased their consonant repertoire at the word- initial position by 7 sounds, and by 11 sounds in the word-final position. Conversely, between 36 and 48 months of CI use, the consonant repertoire at the word-initial position of words increased by only one sound, and at the word-final position there were no additional sounds added. Thus, the interval between the second and third year of CI experience was important for expanding the consonant repertoire, although growth slowed down between the third and fourth years of CI use.

Comparisons of consonant repertoires between children with and without CIs.

To examine whether the development of consonant repertoire of children with CIs were commensurate with their typical peers with matched hearing experience, Tables 3 and 4 present the normative data from Smit et al. (1990) at the right column. No data were available from typical 1- or 2-year-olds because the youngest age that Smit et al. (1990) reported was 3;0. It should be noted again that although Smit et al. (1990) used the 75% criterion to define mastery, the current study used the 70% criterion. Thus, the consonants listed for the typical group were produced by 75% of the typical children in Smit et al. (1990), whereas those listed for the CI group were produced by 70% of the children in the current study.

At the CI age of 36 months, children with CIs produced 13 consonants at the word-initial position, whereas their typical peers who had matched hearing experience produced 10 consonants (see Table 3). The differences included three voiceless fricatives /f, s, ʃ/ that were found in children with CIs but not in typical 3-year-olds. At the word-final position, children with CIs produced 12 consonants, whereas typical 3-year-olds produced 9 consonants (see Table 4). The differences included that the children with CIs produced /l, s, ʃ, tʃ/, whereas the children with typical development did not. In contrast, typical children produced the voiced alveolar stop /d/, which children with CIs did not produce.

At the CI age of 48 months, the children with CIs were no longer producing more consonants than the typical children. Children with CIs produced 14 consonants at the word-initial position, whereas typical 4-year-olds produced 15 consonants. Three consonants were not in the repertoires of children with CIs, yet were in the repertoires of children with typical development, including the voiced palatal glide /j/, the voiced labial-dental fricative /v/, and the voiced alveo-palatal affricate /dʒ/. At the word-final position, children with CIs produced 12 consonants and typical 4-year-olds produced 13 consonants. Differences in repertoire are that the children with CIs have included the voiced alveolar glide /l/ and the unvoiced palatal fricative /ʃ/, whereas typical children did not. Conversely, typical 4-year-olds have added the velar nasal /ŋ/, the voiced labio-dental fricative /v/, and the voiced alveo-palatal affricate /dʒ/, whereas children with CIs did not.

Error Analysis

Because less than 1% (n = 5) of errors were distortion errors, we focused on the errors of omissions and substitutions below. Tables 5 and 6 present the frequency of omission and substitution errors across children at word-initial and at word-final positions, respectively. Overall, children with CIs were more likely to make substitution errors than omission errors at the word-initial position. In contrast, they were more likely to make omission errors than substitution errors at the word-final positions. They also made more omission errors at the word-final positions than at the word-initial positions across the 4-year span. At both word-initial and word-final positions, the proportion of omission errors did not change much during the first 3 years postimplantation but dropped at 4 years postimplantation. For instance, at 4 years postimplantation, the omission errors dropped from 45% to 25% at the word-initial position and from 80% to 65% at the word-final position, suggesting that children with CIs were making proportionally more overt attempts of the target consonants at this period of time than earlier time points.

Table 5 .

Frequency of omissions and substitutions of consonants at the word-initial position across children with CIs by data point, error type, and consonant

	12 months			24 months			36 months			48 months
	O	S	Most common substitutions	O	S	Most common substitutions	O	S	Most common substitutions	O	S	Most common substitutions
p	2	5	/b, t, h/	4	4	/m, d, k/	3	1	/b/	2	1	/b/
b	1	5	/w, d, m/	0	1	/b/	0	2	/p/	1	0
t	5	6	/n, d/	6	6	/d, p, b/	7	1	/d/	4	2	/d, p/
d	2	4	/j, t, g/	3	0		1	1	/g/	1	0
k	7	3	/t, h/	9	4	/t, h, j/	6	3	/t/	2	5	/t, h/
g	5	4	/k, d/	9	3	/j, d/	4	1	/k/	1	3	/k, h, d/
m	2	3	/b, k, g/	2	2	/b/	2	0		1	0
n	4	6	/d/	5	2	/l, w/	1	2	/d/	1	0
w	2	4	/d/	3	0		2	1	/m/	2	0	/m/
l	4	6	/w, n, d/	5	6	/w/	7	12	/w/	1	7	/w/
r	3	10	/b, w, m/	4	13	/w/	2	9	/w/	1	8	/w/
j	3	6	/l, w/	4	6	/n, d/	2	6	/n, d/	1	5	/n, l, ŋ/
h	6	1	/k/	4	1	/b/	3	2	/t, b/	2	1	/m/
f	4	5	/b, p, v/	3	10	/ʃ, b, p/	3	4	/p, ʃ/	2	4	/p, b/
v	2	7	/b, d, w/	7	9	/b, d/	4	9	/b, f/	1	10	/b, d/
θ	6	1	/f/	7	15	/f, d, t/	3	12	/f, d/	1	12	/f, b/
ð	5	9	/d, b, j/	7	14	/d, f, w/	4	8	/d, n/	0	10	/d, w, s/
s	4	6	/d, j, k/	8	4	/d, t/	4	2	/ʃ/	4	1	/d /
z	4	8	/b, j, w/	5	7	/s, d, p/	7	3	/d, s, h/	2	6	/d, s, j/
ʃ	4	5	/w, d/	4	3	/t, d, p/	3	4	/d, t/	0	5	/s, t, tʃ/
tʃ	4	8	/t, d, j/	7	6	/t, ʃ/	2	7	/ʃ, t, d/	1	6	/ʃ, t/
dʒ	4	8	/d, b, j/	3	12	/d, ʃ/	6	2	/d, ʃ/	1	9	/ʃ, t, tʃ/
Total	83	120	—	109	128	—	76	92	—	36	95	—
%	41%	59%	—	46%	54%	—	45%	55%	—	25%	75%	—

Note. O, omission errors; S, substitution errors.

Table 6 .

Frequency of omissions and substitutions of consonants at the word-final position across children with CIs by data point, error type, and consonant

	12 months			24 months			36 months			48 months
	O	S	Most common substitutions	O	S	Most common substitutions	O	S	Most common substitutions	O	S	Most common substitutions
p	2	0		5	0		1	3	/b/	2	1	/b/
b	7	2	/m, t/	3	1	/m/	7	0		4	2	/l, d/
t	11	0		14	0		5	0		4	0
d	10	1	/s/	12	1	/t/	9	0		9	0
k	7	0		10	2	/t, f/	6	1	/t/	4	2	/t/
g	10	0		14	1	/t/	6	1	/d/	4	2	/j, k/
m	6	2		7	3	/b, w/	2	2	/p, n/	2	1	/b/
n	10	0		9	5	/d, ð/	4	1	/b/	2	2	/m/
ŋ	9	4	/n, b/	12	2	/d, ʌ/	10	2	/n/	5	7	/n, b/
l	7	2	/w/	7	4	/w/	3	1	/w/	2	1	/w/
r	10	2	/w/	10	2	/w/	9	1	/w/	5	3	/w/
f	7	2	/ʃ/	10	4	/ʃ, h/	7	2	/ʃ, p/	6	1	/d/
v	8	2	/s/	12	1	/b/	7	1	/b/	6	0
θ	7	2	/ʃ, t/	15	2	/f/	9	6	/f, ʃ/	2	6	/f, t/
s	10	0		7	2	/ʃ/	6	0		3	2	/f, tʃ/
z	10	2	/d/	12	0		7	1	/d/	5	1	/s/
ʃ	8	1	/b/	6	0		5	0		1	3	/s/
tʃ	5	5	/ʃ, h, s/	8	2	/ʃ/	3	4	/ʃ/	3	2	/f, ʃ/
dʒ	6	3	/ʃ,tʃ/	13	4	/ʃ, tʃ/	6	2	/ʃ, d/	3	3	/ʃ, tʃ/
Total	150	30	—	186	36	—	112	28	—	72	39	—
%	83%	17%	—	84%	16%	—	80%	20%	—	65%	35%	—

Note. O, omission errors; S, substitution errors.

We also documented three most common sounds that children with CIs used to substitute the target consonants in Tables 5 and 6 whenever applicable and conducted a feature analysis (i.e., voicing, place, manner) of the errors. At the word-initial position, 37% of the common substitution errors were characterized by one feature (8% voicing, 21% place, 8% manner) and 64% of the errors involved multiple feature errors. At the word-final position, 54% of the common substitution errors were characterized by one feature (9% voicing, 29% place, 16% manner) and 46% of the errors involved multiple feature errors.

Discussion

This study has extended previous studies of consonant development in children with CIs by using larger sample size, restricting the age of implantation to before 30 months of age, and adopting a more robust acquisition criteria. Using a standardized test (i.e., GFTA-2), we documented the consonant development in 32 children in their first 4 years of CI use. We were able to control the word position of a given sound, which allowed us to set acquisition criteria commensurate with previous studies that were based on typical children. We accomplished three goals. First, we compared the performance of children with CIs to the norm in the standardized test based on their chronological age and CI age. These comparisons revealed that children with CIs produced consonants within typical range regardless of being compared to the norms based on their chronological ages or CI ages. Second, we provided a list of consonants produced by 70% of children with CIs yearly in the initial 4 years after implantation. Though there were some discrepancies, the developmental sequence of consonants in children with CIs was consistent with that in typical children who had matched hearing experience. Finally, we provided a description of most frequent errors made by children with CIs. Children with CIs were more likely to omit consonants at the word-final position than at the word-initial position. The proportion of omission errors decreased over time across word positions. A discussion of the implications of these findings follows.

Children with CIs Performed Within Typical Range in Chronological or CI Age Comparisons

The first objective of this study was to investigate whether children with CIs produced consonants at a level comparable with their peers. We predicted that children with CIs would score below the typical range (cutoff = 85) when they were compared with their age-matched peers because of the perceptual and oromotor constraints. The results did not provide robust evidence for this prediction given that the standard scores of the children with CIs were not significantly below the cutoff in most cases. One interpretation was that children with CIs, as a group, can develop age-appropriate articulatory/phonologicial skills as early as the first year postimplantation. This interpretation may not be valid for three reasons. First, there were only 14 children whose data were available at the first year postimplantation. The null result could have been due to the small sample size and large standard deviation. Second, children with CIs, as a group, produced consonants at a level significantly below the typical range at 3 years postimplantation. Third, about 50% of children scored below the typical range over the initial 3 years postimplantation. Thus, the group data did not truly reflect the consonant development in children with CIs.

On the other hand, when we looked at the individual performance, we found that more children with CIs moved into typical range from the third to the fourth years postimplantation. This finding was consistent with the study of Schorr et al. (2008), in which 85% of children with CIs who had on average 5.5 years of CI experience were achieving standard scores within the typical range on GFTA-2, when they were compared to their age-matched typical peers. Taken together, the current study and that of Schorr et al. (2008) suggest that it may be appropriate to expect children with CIs who receive an implant before 30 months of age to produce consonants within typical range in chronological age comparisons at least after 5 years of CI use.

In the case of children who had fewer than 5 years of experience using CIs, our data revealed that not all children with CIs achieved scores within the typical range in the CI age comparison. This means that the CI age comparison still allows the children to identify those who achieved below the typical range given the length of CI use. Thus, the current study suggests that it is reasonable to use CI age as an alternate way to score the GFTA-2.

Children With CIs Developed Consonant Repertoire Similar to Typical Children With Exceptions

The children with CIs advanced their consonant repertoire in a way that was similar, yet not exactly, to their typically developing peers who had matched hearing experience in both rate and type of consonants acquired with each year of CI experience. The general order of acquisition seen in typical children, where anterior consonants were acquired before posterior consonants and stop consonants were acquired before fricatives, was replicated in this study. For instance, the word-initial consonants that were acquired during the first 2 years were mostly anterior sounds, except /h/. There were a few consonants, however, the children with CIs tended to produce slightly earlier than their typical peers. For instance, children with CIs produced word-initial /f, ʃ, s/ and word-final /l, ʃ, s, tʃ/ earlier than did the typical children. This finding is noteworthy because some of these sounds were included as the “late sounds” (s, z, l, θ, ð, ɹ, ʃ, tʃ,dʒ) to be acquired by typical children (Bleile, 2006; Hoffman, Stager, & Daniloff, 1983).

We can think of three possible explanations for why these sounds are emerging earlier in the children with CIs. First, the fricative sounds are consistently transmitted well by the processing strategies of the CI. Blamey et al. (2001) posed that children using CIs might improve their consistency of acquisition of fricative sounds as coding strategies for the sound processors improve. The findings in the current study give credence to this notion on two counts. The children were not only younger than those in Blamey et al. (2001) but were also using technology that was two generations “more advanced” than the children in his study. The major difference between the current processing strategies and the ones used in the 2001 study was that they use higher stimulation rate per channel than older strategies. All the sounds in question, with the exception of /l/, have concentrations of energy at frequency rates that are quite high, with relatively low levels of intensity.

Another explanation for the early appearance of these sounds in the consonant repertoires of children with CIs is that these are reasonably easy to visualize during production. The sounds /f, ʃ, s, tʃ, l/ are all made in the anterior or coronal portion of the mouth and are quite easy to contrast with one another visually. For instance, /f/ is easy to distinguish from /ʃ/ or /l/ because of the placement of the articulators. The children with CIs may be more acutely aware of these visual aspects of the sounds as they watch speakers produce them. The children in this study were all receiving intervention based on the philosophy of total communication, and it is likely they were cued to watch the speaker’s lips for sound production cues (Ling 2002; Tye-Murray, 2009).

A third explanation may be related to the overall contribution of the maturity of the oral-motor system of children with CIs, as they were chronologically older than the typical children. It is possible that the combination of the visibility of a sound and the speech processing characteristics of the CI rendered some sounds especially perceptually salient for children with CIs. The combination of this perceptual “enhancement” with a mature oral-motor system may have rendered the sounds more amenable to production for children after relatively few years of CI experience. A caveat to this notion, however, is that historically children with profound hearing loss who used analog hearing aids had speech that was characterized by an absence of fricatives (Markides, 1983). Because these sounds are known to be challenging for children with hearing loss, we cannot rule out that the earlier mastery of the sounds was a result of having those sounds specifically targeted by child speech language clinicians.

A final note of discrepancy is that children with CIs were not producing the word-final /ŋ/, by 48 months, yet their typical peers were. The two other nasal consonants /m, n/ were produced by children with CIs at both initial and final positions by CI age of 36 months. One reason that /ŋ/ has not emerged may be that it is a postalveolar sound and is not easily visible, plus the perceptual contrast between /n/ and /ŋ/ is difficult (Narayan, Werker, & Beddor 2010).

Children With CIs Significantly Expanded Consonant Repertoire Between 24 and 36 Months Postimplantation

Another notable finding in this study is that consonant repertoires of the children with CIs grew the most between 24 and 36 months after receiving a CI, with some tapering in consonant repertoire growth between 36 and 48 months. This finding is consistent with the study of Connor et al. (2006) in that percent consonant accuracy in the standardized tests showed early bursts between the first and third year of CI use and slowed down after the third year postimplantation in children who were implanted before 30 months of age. However, our finding is somewhat different from what was reported by Blamey et al. (2001) and Tomblin et al. (2008). Both these studies reported the “slowing” of growth in phoneme inventory development between 5 and 6 years of CI experience. Recall, however, that the children in these studies were older when they first received their CI (i.e, on average 3;9 in Blamey et al. and 4;6 in Tomblin et al.), and the method of data collection was conversation or narration. Thus, given current practice where children tend to receive their CIs between 12 and 30 months of age, we should expect considerable growth in phoneme repertoire and overall accuracy of production in the first 3 years after a child receives a CI; these years are an important period in setting the stage for ultimate sound production achievement.

It should also be noted that the current study took strides to ensure that we did not overinflate age of mastery by increasing the criteria such that a larger proportion of the participants in this study had to produce the sounds than in previous studies (70% vs. 50%). Additionally, because we used a standardized test of sound within words, there is no confound of token that may have been in place in previous studies. The use of standardized tests allowed us to separate the age of mastery for a given sound by word positions. Thus, the current findings are more robust than the previous findings (e.g., Blamey et al., 2001; Serry & Blamey, 1999).

Error Patterns in Children With CIs Varied With Word Positions and Time

In this study, children with CIs were more likely to make substitution rather than omission errors at the word-initial position. In contrast, they were more likely to make omission rather than substitution errors at the word-final positions. These findings supported our predictions and were in concert with developmental patterns seen in typical children (Bauman-Waengler, 2012: Dodd, Holm, Zhu, & Crosbie, 2003) and children with profound hearing loss (Smith, 1975). In addition, at both word-initial and word-final positions, the proportion of omission errors did not change much until 4 years postimplantation. This suggests that although the overall numbers of consonants produced increased considerably between 3 and 4 years after receiving a CI, it takes a bit longer for these children to begin to refine their production system and/or phonological representation.

A feature analysis of the substituted sounds children with CIs used in place of the target consonants (Tables 5 and 6) revealed that at the word-initial position, the majority of common substitution errors (64%) involved multiple feature errors, whereas those at the word-final position (54%) were characterized by one-feature errors. This finding may have therapeutic implications, such that final position substitution errors may be easier to remediate than initial position substitution errors, because there are fewer features in error.

Therapeutic Implications

The current findings reveal that it is logical to expect that if children receive a CI before 30 months of age, they will develop speech sounds in an order that is similar to their typically developing peers. We base this expectation not on chronological age, however, but on length of time with CI use or CI age. This study suggests that speech language clinicians can follow typical principles for articulation therapy with some select additions. They can administer a standard articulation test, such as the GFTA-2, to derive an inventory of speech sounds children can produce. Next, clinicians can compare a specific child’s consonant repertoire to the lists that correspond to the number of years of CI experience provided in Tables 3 and 4 at the word initial and final positions if the child has no more than 4 years of CI experience. For children who have more than 5 years of CI experience, it is reasonable for the clinician to use the normative data (e.g., Smit et al., 1990) based on the child’s chronological age to facilitate clinical decision making.

Clinicians should exercise caution when using standard scores to make decisions on whether to include a child on a therapy caseload. Many school districts use a cutoff score of 1.5 SD below average in order to qualify for speech language services. It is apparent that numerous children with CIs may achieve standard scores that are above this cutoff score if CI age is used as a reference and even in cases when chronological age is issued. Results of this study suggest that if the strict 1.5 SD is upheld, many children with CIs may not qualify for speech language services. In the current study, the children achieved these results with the provision of speech therapy (an average of 60min per week). In determining recommendations for continued therapy for an individual client, a clinician will need to consider whether the inclusion of therapy is necessary in order to maintain growth in sound acquisition over time.

Limitations of the Current Study and Implications for Future Studies

Although the characteristics of this particular group of participants is somewhat typical for the current point in time and for certain educational settings, the results may not generalize to all groups of children with CIs. The children in this study had prelingual, bilateral, profound hearing loss, lived with their parents who were hearing and who spoke English, were implanted before 30 months, and received a moderate amount of therapy provided in a public school setting. The reader must consider these characteristics when comparing an individual child with these results. For instance, given that children in this study all had unilateral CIs and received therapy based on the philosophy of total communication, it is not clear whether the findings are generalizable to those who have had bilateral CIs or who use bimodal communication (CI and hearing aids) or those who receive therapy based on the auditory-oral method. Future studies that investigate the effects of bilateral/bimodal communication are needed to answer these questions. The data provided in this study can serve as a baseline for comparison.

In addition, this study investigated consonant development in children with CIs by using GFTA-2. Because of the use of a standardized test, we were unable to investigate the effect of phonotactics (e.g., the vowel preceding or following the target consonants) on consonant acquisition. Research has shown that how often a target sound combines with the other sound can influence the production accuracy of the target sound (e.g., Storkel & Morrisette, 2002). Future studies that manipulate phonotactics will lead to a better understanding of consonant development in children with CIs.

Funding

This research is supported in part by research grant 5 P50 DC00242 from the National Institute on Deafness and Other Communication Disorders, National Institutes of Health ; specialized center grant M01-RR-59, National Center for Research Resources, General Clinical Research Centers Program, National Institutes of Health; the Lions Clubs International Foundation; and the Iowa Lions Foundation.

Conflicts of Interest

No conflicts of interest were reported.