Variables affecting speech intelligibility in prelingual Arabic speaking cochlear-implanted children

Background Speech intelligibility (SI) is usually expressed as the degree to which a speaker′s intended message can be recovered by other listeners. It is determined by many factors that may be affected by the degree of hearing loss. The use of cochlear implants (CIs) can facilitate the development of speech and language skills of prelingually deaf children. Thus, improvements in SI after CI fitting can provide indirect evidence of sensory aid benefits. Aim This cross-sectional study aimed to explore the different variables that affect SI in prelingual CI children in order to achieve the maximum benefit for improving SI in such candidates. Participants and methods This study included 30 prelingually hearing-impaired participants, with ages ranging from 6 to 10 years. They used bilateral hearing aids for a duration of 2.6 ΁ 1.7 years before undergoing CI, which was performed at ages ranging from 4 to 9 years. They were enrolled in specific language intervention programs after implantation for a duration ranging from 1 to 3.8 years. Then, the Arabic Speech Intelligibility test was used to assess their SI. The effect of different variables on their SI was assessed through both comparative and correlative analysis. Results SI scores showed highly significant differences in patients at younger age at implantation (aged ≤5years) and in those who had received language therapy for more than 2 years after implantation. Meanwhile, preimplantation therapy for more than 1 year and usage of hearing aids for more than 1 year before CI had significant effects on the SI scores. SI was highly significantly negatively correlated with age at CI and highly significantly positively correlated with the duration of postimplantation therapy. The SI scores were also affected by the three groups of the SI test. The front consonants had higher significant scores compared with both back consonants and sentences, and back consonants had significantly higher scores compared with sentences. Conclusion The SI of prelingual cochlear-implanted children was affected by all of the studied variants. However, for best SI, age at implantation and postimplantation therapy duration should be considered to enable better SI in these children.


Introduction
'Speech intelligibility' (SI) refers to the degree to which a speaker's intended message can be recovered by other listeners [1], or the comprehensibility of the specifi cal linguistic information encoded by a speaker's utterances [2].
Measuring SI is, however, problematic in patients with hearing impairment because it is aff ected by a number of factors including the experience of the listener, the context in which the sentence was spoken, the visibility of the speaker, the familiarity of the speaker with the material spoken, and the articulation/phonological aspects, suprasegmental factors, contextual and semantic/morphologic/syntactic features, and voice quality [3]. Th e extent to which these parameters are aff ected depends on the degree of hearing loss; the worse the hearing, the more these parameters are aff ected.
Hearing loss aff ects a large range of consonants including stridents, velars, nasals, and glides. Th e error types are mostly substitutions, distortions, and omissions of word initials that mostly aff ect SI. Poor pitch control or excessive pitch variation may also contribute to reduced SI [4]. Many studies considered listener's experience to be the best known factor that can infl uence the intelligibility of speech [5]. Studies completed during the 1960s-1980s revealed that the speech of children with severe to profound hearing losses was 20% intelligible on average [6].

Background
Speech intelligibility (SI) is usually expressed as the degree to which a speaker's intended message can be recovered by other listeners. It is determined by many factors that may be affected by the degree of hearing loss. The use of cochlear implants (CIs) can facilitate the development of speech and language skills of prelingually deaf children. Thus, improvements in SI after CI tting can provide indirect evidence of sensory aid bene ts.

Aim
This cross-sectional study aimed to explore the different variables that affect SI in prelingual CI children in order to achieve the maximum bene t for improving SI in such candidates.

Participants and methods
This study included 30 prelingually hearing-impaired participants, with ages ranging from 6 to 10 years. They used bilateral hearing aids for a duration of 2.6 ± 1.7 years before undergoing CI, which was performed at ages ranging from 4 to 9 years. They were enrolled in speci c language intervention programs after implantation for a duration ranging from 1 to 3.8 years. Then, the Arabic Speech Intelligibility test was used to assess their SI. The effect of different variables on their SI was assessed through both comparative and correlative analysis.

Results
SI scores showed highly signi cant differences in patients at younger age at implantation (aged 5years) and in those who had received language therapy for more than 2 years after implantation. Meanwhile, preimplantation therapy for more than 1 year and usage of hearing aids for more than 1 year before CI had signi cant effects on the SI scores. SI was highly signi cantly negatively correlated with age at CI and highly signi cantly positively correlated with the duration of postimplantation therapy. The SI scores were also affected by the three groups of the SI test. The front consonants had higher signi cant scores compared with both back consonants and sentences, and back consonants had signi cantly higher scores compared with sentences.
Cochlear implants (CIs) have been proven to provide an access to the speech signal in profoundly deaf children who derive no benefi t from hearing aids (HAs). CIs do not restore normal hearing, but are individually programmed to provide the recipient with hearing sensitivity within the speech range [7]. A substantial number of studies have demonstrated that the use of CIs can facilitate the development of speech and language skills in children who are prelingually deaf.
Th e extent of SI is likely to be an indicator of the auditory perceptual benefi ts received from sensory aids in young sensory aid recipients [8]. Th us, improvements in intelligibility after CI fi tting can provide indirect evidence of sensory aid benefi t. In contrast, a lack of improvement in SI might indicate that children are not receiving full benefi t from their sensory aids. In the population of prelingually deaf CI recipients, the frequently considered factors to individual variability include age at implantation, duration of CI use, and communication mode [9].
Intelligibility scores can provide important supplemental information for decisions regarding the adjustment or replacement of sensory aids [10]. To our knowledge, no previous study has been conducted to assess intelligibility scales of speech of Arabic speaker prelingual CI children until now. Th erefore, the aim of this study was to explore the diff erent variables that might aff ect the intelligibility of speech in prelingual CI candidates to achieve the maximum benefi t for SI.

Participants and methods
Th e current study was conducted on 30 prelingual children, 13 (43.3%) boys and 17 (56.7%) girls, with an age range of 6-10 years and a mean age of 7.9 ± 1.3 years at the time of performing the study. Th e participants were born deaf or had lost their hearing in the fi rst 2 years of life. Th eir severe to profound sensorineural hearing loss was confi rmed by agematched hearing evaluation. All candidates were recruited from Ain shams University hospitals from March 2011 to August 2013 after obtaining consent from their parents. All candidates were of average mentality according to the Stanford-Binet Intelligence Scale test [11]. Th eir total language age was 3 years or more and they were capable of speaking a mean of at least three-word-length sentences as assessed by the Arabic language test [12]. Patients with visual impairments or brain damage were excluded.
All participants were fi tted with bilateral HAs before performing the CI with a mean duration of 2.6 ± 1.7 years and had been receiving preimplantation language therapy for a mean duration of 2 ± 1.6 years.
All children were unilaterally implanted either with Medel Sonata or Cochlear Freedom CI system. Th eir ages at implantation ranged from 4 to 9 years, with a mean of 6.2 ± 1.5 years. Th ey had been receiving regular language sessions after implantation twice weekly at the Phoniatric Unit, Ain shams University hospitals, for a duration ranging from 1 to 3.8 years, with a mean of 2.2 ± 1 years.
Th e participants were divided into four subgroups on the basis of the following variable parameters: (1) Age at implantation (A): this was further divided into two subgroups, subgroup A1 (≤5 years) and subgroup A2 (>5 years). All candidates were evaluated according to Ain Shams Phoniatric Unit's protocol for language assessment [13]. Aided response was evaluated in all children using CIs in free fi eld to ensure within-normal aided response. Children with unsatisfactory aided response were excluded from this study.
All participants were subjected to the Arabic Speech Intelligibility test [14]. Th e test is designed to provide an estimation of the overall SI. It provides the examiner with a fi nal score in percentage, which is meant to be an objective measure of a child's SI. It is composed of three picture sets (front consonant monosyllabic words, back consonant monosyllabic words, each consisting of 20 cards, and 10 cards of action verb sentences). Th e child is asked to name the pictures (which the listeners cannot see) correctly enough that the listener can identify which word among the cards the child is trying to say. Th e score may denote unintelligible speech (0-29%), poorly intelligible speech (30-50%), fairly intelligible speech (51-66%), good intelligible speech (67-84%), and excellent intelligibility (85-100%).

Statistical analyses
Th e collected data were revised, coded, tabulated, and introduced into a PC using statistical package for social sci ence (SPSS 15.0 for windows; SPSS Inc., Chicago, Illinois, USA). Data were presented and suitable analysis was carried out according to the type of data obtained for each parameter. Th e numerical data were presented as means, ± SD, and minimum and maximum values (range).
Th e non-numerical data were presented as frequency and percentage. Th e independent-samples t-test was used to assess the statistical signifi cance of the diff erence between two study group means, whereas one-way analysis of variance was used to assess the statistical signifi cance of the diff erence between more than two study group means. Once we had determined that diff erences existed among the means, one-way analysis of variance, post-hoc tests, and pairwise multiple comparisons were applied to determine which means diff ered. Range tests were used to identify homogenous subsets of means that were not diff erent from each other. Pairwise multiple comparisons were used to test the diff erence between each pair of means, which yielded a matrix in which asterisks indicated signifi cantly diff erent group means at an level of 0.05. Pearson Correlations were used to assess the strength of association between two quantitative variables. Th e correlation coeffi cient denoted symbolically as 'r' defi nes the strength and direction of the linear relationship between two variables. P values less than or equal to 0.05 were considered signifi cant.
Th ere was no statistically signifi cant diff erence between the two sexes with respect to the mean scores of the SI test (P > 0.5), as shown by the independent-samples t-test (Table 1 and Fig. 1). Diff erent variants including the age at CI, duration of therapy, duration of fi tting the HAs before the CI, and duration of therapy after the CI were all compared with respect to the SI scores. (Table 2). Th e independent-samples t-test revealed that there were highly statistically signifi cant diff erences between SI and age at CI and duration of postimplant therapy (P ≤ 0.001). Meanwhile, each of the two subgroups, duration of preimplant therapy (>1 year) and preimplant HA fi tting (>1 year), were associated with higher signifi cant SI (P ≤ 0.05).
Analysis of the correlation between SI and the diff erent variables revealed a highly signifi cant negative correlation between SI and the age at CI (r = −0.9; P ≤ 0.001). Signifi cant positive correlations were detected between SI and duration of preimplant HA and preimplant therapy (r = 0.39 and 0.42, respectively; P ≤ 0.05). Meanwhile, the duration of postimplant language therapy correlated positively and highly signifi cantly with the SI score (r = 0.89; P ≤ 0.001) ( Table 3).
Th ere was a highly statistically signifi cant diff erence between the mean scores of the three subgroups of SI for all groups -namely, the front consonant words, back consonant words, and sentences. Th e most signifi cance denoting high intelligibility was for the front consonant words followed by the back consonant words, and the least was for sentences (Table 4 and Fig. 2).
Th e mean percentage diff erences in SI for back consonants versus front consonants, for sentences versus front consonants, and for sentences versus back Comparison between sexes regarding speech in telligibility.

Figure 1
Comparison between the speech intelligibility scores of the front consonant words, back consonant words, and sentences. consonants were statistically highly signifi cant. Using back consonant words rather than front consonant words reduced the mean percentage diff erence in SI by −11%, whereas using sentences rather than front and back consonant words reduced the mean percentage diff erence in SI by −18 and -6%, respectively (Table 5 and Fig. 3).

Discussion
Readily intelligible connected speech is the ultimate goal of speech intervention for children with hearing loss. It seems reasonable then that children's progress toward this important goal would be monitored closely [15]. Integration into mainstream elementary classrooms is one of the key motivations for CI use in very young children. Successful integration requires the development of many social, communicative, preacademic, and academic skills before regular school placement [16]. SI of prelingual CI children might aff ect this challenge.
Th is study showed that the overall mean SI score of all participants after regular postimplantation therapy with a mean duration of 2.2 ± 1 years was 60 ± 26%. Th is is nearly commensurate with the fi ndings of Chin et al. [17], in whose study participants received their implants at a relatively young age (M = 38 months) and achieved a mean intelligibility score of 34.5% after an average of 28 months of CI use.
Th e statistically signifi cant diff erences detected in this study on comparing SI scores in relation to the diff erent variants including age at implantation of CI, duration of preimplantation therapy, and duration of wearing HAs before CI and duration of postimplan tation    Comparison of the speech intelligibility scores between the differe nt subgroups. therapy emphasize the positive role of the CI and its marked impact on the linguistic competence of profoundly hearing-impaired children. Our fi nding was comparable to that of Shu-Chen et al. [18], who showed that CIs before age 5 years resulted in greater improvement in speech production skills compared with CIs after that age, which could be related to neural plasticity. It was also reported by Sharma and Dorman [19] that when children are implanted early in childhood the central auditory development becomes age appropriate within 3-6 months after implantation and continues to be normal throughout their childhood years. Decline in neural plasticity with age was also proven by Zwolan et al. [20], who focused on the inability of adults to form necessary neural structures required to process language and speech.
Moreover, the age at implantation and duration of postimplant therapy showed highly signifi cant negative and positive correlative values, respectively, in our analysis of the variable aff ecting the SI of the prelingual CI children in this study.
Our results were comparable to that of Fryauf-Bertschy et al. [21], who suggested that young children represent the best candidates for a CI. Delayed implantation after the early onset of deafness predicts lower levels of speech reception. Moreover, Kirk et al. [22] reported that children with congenital deafness who were younger than 4 years and received CIs developed signifi cantly better speech and language skills compared with children who received implants after 6-7 years of age. Th erefore, the age of the child at the time of implantation is critical for SI progress, as the age of 4 years is the age at which a normal developing child's spontaneous speech should be intelligible to unfamiliar adults, even though some articulation and phonological diff erences are likely to be present [23]. Choosing a higher age limit (5 years) for subgrouping of our studied children, compared with other studies [24,25], could be explained by the availability and cost of CI in a developing country compared with developed countries.
Preimplantation therapy duration is also an important factor that had an eff ect on SI. Our results revealed that children who underwent therapy for 1 year or more before implantation achieved a statistically signifi cantly higher percentage of SI compared with those who attended therapy for less than 1 year. Th is analysis is unique to our study and therefore should be validated in a larger group of children.
It was recently speculated that central auditory processing modifi cations might begin with the onset of hearing loss. It has been shown in a functional MRI study of a cohort of 10 postlinguistically deaf individuals that specifi c brain reorganizations associated with phonological processing in the right posterior superior temporal gyrus/supramarginal gyrus were infl uenced by the duration of severe to profound HL [26] as well as the duration of HA use [27]. Accordingly, early binaural amplifi cation (aided by speech therapy) is suggested to result in slowing down of the related central reorganization that is the likely cause of the decreased performance. Th is was proven in the current study, as children who wore HAs for 1 year or more before CI had higher intelligible speech compared with those who wore HAs for less than 1 year. Whitehill [28] described intelligibility of speech as the most important measure of any speech disorder, and improving it should be the primary goal of therapeutic intervention. Post-CI therapy is the last variable evaluated to determine its eff ect on SI. Th e results of the current study proved that, longer the duration of postimplantation therapy (>1 year), better the SI outcomes, which has been attributed to be due to both time factor and intervention strategy. Th is fi nding is supported by many studies including that of Tobey et al. [8], who reported that speech production in children with CI is infl uenced by implant characteristics, including the length of time, using the newest speech processing strategies, as well as the educat ional programs emphasizing oral-aural communication.
Use of front consonant words was proven in the current study to be the most intelligible in the CI children's speech, followed by back consonant words, and the least intelligible were sentences. Th ese fi ndings give extremely important information that must be considered in clinical applications. Th ey agree with the results of Lach et al. [29], who declared that young hearing-impaired children produce front consonants more often than they do back consonants and they use front consonants with greater frequency than do normally hearing children. Th ey often produce more visible consonants (i.e. labials vs. coronals or dorsals) [30]. Further, phonemes produced at the front of the mouth are more often produced correctly than phonemes produced at the back of the mouth. Th is makes sense when one considers that the relative visibility of articulators should be important to hearing-impaired children [31]. Th is general trend of better production of phonemes has been found not only for production of isolated words and sentences but also for spontaneou s speech [32].
McGarr [33] has shown that intelligibility scores for hearing-impaired speakers may vary considerably depending on speech material (sentences or words). As the SI of the speech material (sentences or words) is infl uenced by the length of the utterance, presence or absence of familiar vocabulary, and the presence of less complex syntax [5], it is expected to fi nd words more intelligible than sentences, as proven in the current study.
Correlation analyses were conducted in this study to investigate separately the eff ect of each variable on the SI, apart from the remaining studied variants. Th ese revealed that age at implantation was highly signifi cantly negatively correlated SI. Th e duration of postimplantation language therapy was highly signifi cantly positively correlated t o SI (P < 0.001) compared with the duration of p reimplantation therapy and HA usage, which were signifi cantly positively correlated to the SI (P < 0.05). On the basis of these fi ndings we concluded that, although these two variants -duration of preimplantation therapy and HA -have signifi cant eff ect on the SI, they should not negatively interfere or delay the age at CI and consequently miss out the more eff ective factors: CI at earlier age and the following postimplantation therapy.
Factors that were thought to be major contributors to SI -such as duration of wearing HAs and duration of preimplantation therapy -did not have the highly signifi cant eff ect on SI as the other two variants -age at CI and duration of po stimplantation therapy -and therefore these last two variants should be given greater priority while considering prelingual implantation of the cochlea. It is not only early fi tting of HA but also early CI that has been hypothesized to aff ect central auditory development -a benefi t that should not be missed in our vision for best SI in prelingual CI children.
We recommend not only a longer follow-up comparative study of SI in prelingual cochlear-implanted Egyptian children but also one that is multicentric in order to compare the eff ect of the diff erent types of CIs on SI in this categor y of children.