Abstract
In children with a hearing loss who receive cochlear implants (CIs) under the age of 2, regular assessments are conducted to monitor auditory and linguistic progress. However, the collection of authentic, representative, and reliable expressive language data on young children with CIs remains a challenge. The purpose of the study was to determine whether data from parental report, language diary, and spontaneous language sample are equally representative of the development of expressive vocabulary over the first 12 months of CI use. Nine French-speaking children and their families participated in the study. We collected data at 3, 6, 9, and 12 months post-implantation, and we measured parental satisfaction regarding the use of a language diary. All three methods showed a progression in the total number of different words expressed over time and captured grammatical diversity. The findings suggest that when vocabulary size is still small, the diary might provide a more comprehensive picture of development than a vocabulary checklist.
Introduction
Children with hearing loss need early intervention to mitigate the difficulties associated with hearing impairment, including speech perception and production, language acquisition, socialization, academic achievement, and social participation (Hoffman, Quittner, & Cejas, 2014; Lederberg, Schick, & Spencer, 2013; Marschark, Rhoten, & Fabich, 2007; Punch & Hyde, 2011; Tomblin et al., 2015). Nowadays, many parents of children with a severe to profound hearing loss choose cochlear implantation for their child. When families have access to an early hearing detection and intervention program, cochlear implantation usually takes place within the first 2 years of life. For families and health care professionals, this is good news, as research has shown that early intervention, including cochlear implantation, is beneficial to language development (Ching, 2015; Connor, Craig, Raudenbush, Heavner, & Zwolan, 2006; Geers, Nicholas, & Sedey, 2003; Yoshinaga-Itano, 2003), despite wide variability in language outcomes (Niparko et al., 2010). The increase of cochlear implantation under 2 years of age may present new challenges to professionals who are not necessarily familiar with family-centered models of service delivery (see Moeller & Mixan, 2016, for an overview). When working with infants and toddlers, the first challenge for cochlear implant (CI) team is to adapt their interventions to very young children and their families to make it more family-centered. This adaptation requires a partnership with parents to work collaboratively in the observation of the child’s language development (Stredler-Brown, 2010). The second challenge is to assess language abilities and to monitor early linguistic progress during the first months of CI use. An international consensus on best practices for children with a hearing loss recommends systematic and comprehensive assessments at regular intervals to ensure that intervention goals are well adjusted and that the child is developing as expected (Moeller, Carr, Seaver, Stredler-Brown, & Holzinger, 2013; Szarkowski & Hutchinson, 2015). For young children with CIs, regular assessments are conducted primarily to monitor auditory and linguistic progress and to guide intervention; however, it is not clear what approach or combination of approaches are most appropriate to systematically collecting authentic, representative, and reliable expressive language data on young children. Hearing loss restricts access to spoken language input (Tomblin, Oleson, Ambrose, Walker, & Moeller, 2014). Moreover, past research involving children with hearing loss has suggested that vocabulary is vulnerable to a delayed exposure to language (Gilbertson & Kamhi, 1995; Stelmachowicz, Pittman, Hoover, & Lewis, 2004).
Importance of Early Vocabulary
Many studies have emphasized the importance of early vocabulary development in children with typical hearing (e.g., Duff, Reen, Plunkett, & Nation, 2015; Reilly et al., 2007). For example, Lee (2011) found that vocabulary size at age 2 predicted later language and literacy skills, after having controlled for ethnicity, gender, birth order, and socioeconomic status. Marchman and Fernald (2008), in a longitudinal follow-up study, showed that children (N = 28) who had larger vocabularies in preschool showed better language and reading abilities at age 8.
Research describing the development of early expressive vocabulary after cochlear implantation is still emerging (Nott, Cowan, Brown, & Wigglesworth, 2009). Although later language outcomes have been studied extensively, the emergence of expressive language (i.e., early vocabulary size and exact composition) has been investigated less frequently (Koşaner, Uruk, Kilinc, Ispir, & Amann, 2013). In a study by Fagan (2015), nine participants who received a CI between the age of 8 and 14 months showed a vocabulary delay of approximately 6 months when measured at 26 months of age, compared with a sample of children with typical hearing. An increased knowledge of vocabulary development within the very first months of implant use will also improve early intervention practices, especially regarding speech-language therapy (Välimaa et al., 2017). Thus, there is a need for a better understanding of early expressive vocabulary development in children with CIs.
Assessment of Emerging Language Skills
Various methods such as norm-referenced assessments, language sampling, parental questionnaires, and language diaries are used to assess early expressive language after cochlear implantation. Jamieson and Simmons (2011) provided an overview of formal and informal approaches to the language assessment of children with hearing loss. Prezbindowski and Lederberg (2003) have outlined cautions regarding the methods used to assess vocabulary in deaf and hard-of-hearing children. The following paragraphs review the research base relating to these methods.
Language tests
Research on language outcomes in children with CIs shows that standardized tests assessing expressive vocabulary are rarely used with children younger than 4 years old. For example, the Expressive One-Word Picture Vocabulary Test (EOWPVT), a test of expressive vocabulary that can be administered to hearing children aged 2 through 18, was used in a study involving 153 children who received CIs before the age of 5 (Geers, Moog, Biedenstein, Brenner, & Hayes, 2009). However, children were aged 5 and 6 at the time of testing and had used their CIs for at least 1 year (duration of use ranged from 1 to 5 years). Boons et al. (2013) also used the EOWPVT with 70 school-aged children with CIs and compared the results with those of a matched group of hearing peers. Again, age at the time of test ranged from 5 to 13 years of age. According to Blaiser and Shannahan (2018), the fact that most language tests do not include children with hearing loss in their norming population prevents their widespread use with deaf and hard-of-hearing children.
Language samples
Another option is the collection of spoken language samples in natural play-based situations at regular intervals. Many researchers in the field of cochlear implantation in children have used this method to describe language evolution over time, to identify specific elements of development such as the emergence of expressive vocabulary (Moreno-Torres, 2013; Schopmeyer, Mellon, Dobaj, & Niparko, 2000) and the use of specific grammatical morphemes (Szagun, 2004), and to closely monitor the fulfillment of language milestones (Ertmer, Strong, & Sadagopan, 2003). For example, Moreno-Torres (2013) studied the transition from babbling to words and the one-word stage in eight children who received CIs by the age of 2. He collected samples of interaction with an adult (parent or speech-language pathologist) every 6 weeks during the first year of CI use (and after 15 and 18 months of use). Moreno-Torres (2013) reported that the variability in the outcomes might be associated with environmental factors, particularly those related to the socioeconomic status of the family (i.e., living in “privileged environments,” p. 22, vs. less stimulating environments).
Parental questionnaires
Previous studies have shown that the use of parent reports can promote parental involvement in the assessment process of children with hearing loss (Szarkowski & Hutchinson, 2015). For example, the LittlEars® Auditory Questionnaire (LEAQ) assesses pre-verbal and early verbal auditory skills in children with CIs (Coninx et al., 2009). Parent reports are also widely used to evaluate early language skills. The MacArthur–Bates Communicative Development Inventories (MBCDIs; Fenson et al., 2007) exist in more than 70 languages, including Quebec French (Trudeau, 2008). It is extensively used in research and in clinical settings. Moreover, Thal, DesJardin, and Eisenberg (2007) established that the MBCDI is valid to use with children with CIs who are in the early stages of language development even if their age exceed the norming age range of the tool. Duchesne, Sutton, Bergeron, and Trudeau (2010) also highlighted the utility of the Quebec French version of the MBCDI to measure the expressive vocabulary level of 11 French-speaking children aged between 20 and 44 months who had received a CI between the age of 8 and 25 months.
Each of the three above methods has strengths and weaknesses, which may be modulated by the age of the child being assessed. For instance, formal testing does not provide a comprehensive description of expressive language abilities, for example, the size and content of the vocabulary and the emergence of two-word utterances, as it includes only a limited number and variety of items (Dockrell, 2001). With this limitation in mind, one may still choose a standardized test to measure specific aspects of language development, in which case, the test must be appropriate for use with children under age 2 and sensitive to levels of language during the initial stages of development. These criteria considerably limit the available options, considering that attention and concentration skills, the ability to understand instructions, and the degree of collaboration required for conventional structured testing are not sufficiently developed in very young children (Bagnato, 2005; Nott, Cowan, Brown, & Wigglesworth, 2003; Szarkowski & Hutchinson, 2015).
Language samples are recognized as part of the “best practices” in the evaluation of young children with hearing loss (Blaiser & Shannahan, 2018). However, this method of investigation is demanding in terms of the time required to complete the transcription and analysis (Nott et al., 2003). In addition, children with a hearing loss can be more difficult to understand because of their speech intelligibility (Prezbindowski & Lederberg, 2003). Moreover, the data obtained might not always be representative of the child’s abilities, as it can be influenced by the conversational context as well as personality factors (e.g., shy temperament; Prior et al., 2008). For example, Hoff (2010) showed that hearing children aged 17 to 26 months (N = 20) who were engaged in dyadic mother–child conversation across three settings (mealtime, toy play, and book reading), used richer vocabularies in book reading contexts than during toy play and mealtime.
Finally, there is no doubt that parental questionnaires and checklists provide valuable information on specific aspects of child development. Despite the strength of parental questionnaires, they do not allow a continuous monitoring of language development (Nott et al., 2003). As these questionnaires list a finite set of linguistic behaviors based on a predetermined list of words, they do not take into account lexical idiosyncrasies related to cultural or family habits. Indeed, Nott et al. (2003) found that parents reported that their child was using a substantial number of words that were not listed in the vocabulary inventory. In addition, Ertmer and Mellon (2001) pointed out that the frequent and repeated administration of a vocabulary checklist increases the possibility that certain words are introduced and reinforced to the detriment of other words not included in the inventory.
Language diaries
Reznick and Goldfield (1994) have argued that a language diary is the most effective tool to document early language development because of its capacity to capture emerging linguistic behaviors (Ertmer & Mellon, 2001) and to describe “the individual nature of each child’s early language development” (Nott, Brown, Cowan, & Wigglesworth, 2005, p. 100). This method was used in a study on the emergence of spoken vocabulary in a child who received a CI at 20 months (Ertmer & Mellon, 2001). In this study, parents were also required to complete the MBCDI every other month. Results showed differences according to the evaluation method used (diary or parent report). Other studies have shown that the diary technique is especially effective for documenting early lexical development in children with a profound hearing loss (Nott et al., 2005; Nott et al., 2003). The Di-EL (Diary of Early Language) was developed at the University of Melbourne (Nott et al., 2003) and specifically designed for children with a hearing loss. It allows the description of early lexical acquisition and the emergence of word combinations. In an initial study, the parents of nine children completed a language diary for a variable amount of time (between 1 and 2 years), from the moment of the first fitting of the device until the attainment of 100 words (Nott et al., 2003). Results of this study showed that the Di-EL is effective in providing descriptive information about early lexical development. In addition, the Di-EL was correlated with other measures of early expressive language, such as the MBCDI and the Rossetti Infant–Toddler Language Scale (RITLS; Nott et al., 2003). In a follow-up study, the language diary data from three children were further analyzed. Results showed the capacity of the Di-EL to monitor the rate of lexical acquisition, lexical content, and the emergence of word combinations, and to portray accurate individual patterns of language acquisition that may guide early practitioners in developing appropriate interventions (Nott et al., 2005).
Triangulation of Different Sources of Early Language Data
Gatt, Grech, and Dodd (2014) discussed the benefits of using multiple sources of data to describe child language development. In their study involving 44 typically developing children, they collected data from a caregiver report, a picture-naming task, and a spontaneous sample. Gatt and her colleagues concluded that the use of multiple methods of investigation provides a comprehensive description of expressive vocabulary and ensures validity of the measures. In addition, Wood and Dockrell (2010) highlighted the importance of obtaining a comprehensive picture of a child’s functioning through the use of a variety of assessment tools with children who have a hearing loss.
Although various assessment formats can be used in conjunction to provide information about children’s progress, they are likely to be used at different moments in development (e.g., language sampling during a specific time period or diary kept over longer periods of time). To date, very few studies on early vocabulary development in children with CIs have presented results that were based on the triangulation of multiple sources of information. For example, Spencer (2004) triangulated data from parent interviews with those from direct child assessments and observations to document language skills as well as parental involvement before and after cochlear implantation. This method allowed Spencer to find associations between the two variables.
In the present study, we applied the use of multiple methods of investigation with a sample of young children with CIs. Our study aimed to go beyond previous findings in exploring how each particular measure might contribute different information on early language development of children with CIs. Moreover, we kept in mind the necessity of collecting authentic and representative language data. This issue is important: Professionals involved in early intervention need to closely monitor language development to determine the appropriateness of intervention goals. Designing timely and appropriate intervention goals is particularly important for this group of children (Ching, 2015; Nott et al., 2005; Stredler-Brown, 2010; Välimaa et al., 2017).
Objectives
The purpose of our study was to determine whether a language diary (Di-EL) was equally representative of children’s lexical functioning as a parental questionnaire (MBCDI), and as a 20-min language sample, over a 12-month period. In other words, we asked whether data from parental report, language diary, and spontaneous language samples are equally representative, both quantitatively and qualitatively (i.e., distribution of word categories) of the development of expressive vocabulary over time during the first year of CI use. A secondary purpose of this study was to survey parents’ satisfaction and motivation regarding the use of a language diary, to explore its potential merits as part of standard language intervention procedures with children who newly received a CI and their families. Specifically, were parents motivated to record single-word and multiword utterances in a diary during the first year of their child’s CI use? Were they satisfied with the tool?
Our goal was not to determine whether one assessment tool was better than the other, but rather we sought to examine how these tools might differ in their particular relevance and clinical usefulness as well as the pertinence of triangulation of different sources of language data for this population. Consequently, we were interested in knowing which tool or combination of tools might be the most sensitive to change (or lack thereof) in the early stages of language development and more efficient to capture the course of expressive vocabulary acquisition within the first 12 months with a CI. We hypothesized, based on the Gatt et al. (2014) findings, that the three methods would show similar trends in terms of the general evolution of expressive vocabulary development. However, we expected qualitative differences, as suggested in Nott et al. (2005). We also expected large inter-individual variability, as in many studies involving young children with CIs (Houston & Miyamoto, 2010; Spencer, 2004; Svirsky, Robbins, Kirk, Pisoni, & Miyamoto, 2000). Whereas some factors have been shown to explain a relative proportion of the variance (e.g., age at implantation), a large amount of the observed variance remains unexplained (see Szagun & Schramm, 2015, for a succinct overview).
Method
Ethics boards at the institutions involved in the research project approved the study (MP-IRDPQ-13-303).
Participants
Nine children with CIs living in French-speaking homes and their families participated in the study (eight girls and one boy). All parents were hearing. The inclusion criteria were as follows: (a) initial programming of a CI in the last 6 weeks, (b) complete insertion of the electrodes in the cochlea, (c) French as the main language at home, and (d) no spoken vocabulary at the time of initial fitting of the device. We initially recruited 10 children, via the Cochlear Implant Center. One family had to withdraw from the study because of postoperation complications. We continuously recruited children over a time frame of approximately a year, as families were entering the CI intervention program. Children were enrolled in the study at the time of their initial fitting (chronological age at the beginning of the study corresponds to age at implantation plus approximately 4 weeks) and had virtually no expressive vocabulary (JD01 had one spoken word), although some of the participants were using some signs to communicate. Mean age at implantation was 19 months (range = 10-27 months). All children received a Nucleus Freedom® CI24RE device. Eight participants eventually received a second implant; four of them started to use the second implant during the course of the study (i.e., before 12 months of use; see Table 1). Prior to implantation, all participants had a hearing-aids trial for a mean period of 5 months (ranging from 2 to 9 months; mean age at first fitting was 13 months (range = 4-22 months).
Characteristics of the Participants (n = 9).
Note. MBCDI = MacArthur–Bates Communicative Development Inventories.
During the course of the study, all children and their families participated in a standard intensive post-implant (re)habilitation program (five or six therapy sessions apportioned among 4 weekdays [for a total of 4-5 hr a week], including auditory training and speech and language therapy) during the first 3 months of device use. They continued to receive speech-language and audiology intervention once or twice a week for the remainder of their first year of CI use. All families were encouraged to use sign language to support the emergence of spoken language; parents were basic users of sign language (i.e., elementary level of proficiency). Table 1 presents the characteristics of the participants.
Materials and Procedure
Data collection points were the following: at the time of initial programming (T0), after 3 months (T3), 6 months (T6), 9 months (T9), and, finally, after 12 months of device use (T12). At each data collection point, we gathered the data for each child within a 2-week frame to ensure that we were capturing the same moment in language development with the three assessment methods.
Cognitive and social-emotional functioning
At T0, we administered the cognitive and social-emotional subtests from the Bayley Scales of Infant and Toddler Development–3 (BSID-III; Bayley, 2006), to establish the developmental level of each child and detect the possible presence of additional disabilities that could influence the results. Mean percentiles were, respectively, 48 for cognitive (range = 1-91) and 22 for social-emotional (range = 2-37). One child (JD06) obtained particularly low scores, which skewed the group mean. At this point, outliers were not removed from the study given the small number of participants—and also to acknowledge the inherent heterogeneity of the population. Table 2 shows the results of cognitive and social-emotional functioning (expressed as percentiles according to chronological age).
Results from the BSID-III (Percentiles) and LEAQ (Total/35).
Note. BSID-III = Bayley Scales of Infant and Toddler Development–3; LEAQ = LittlEars Auditory Questionnaire.
LEAQ data were missing for three participants at T12.
Auditory development
The LEAQ (Tsiakpini et al., 2004) is a short questionnaire, composed of 35 yes–no questions, that has been validated with more than 3,000 children with normal hearing (Coninx et al., 2009). In a study involving 122 children, Obrycka, Lorens, García, Piotrowska, and Skarzynski (2017) established that the LEAQ is a consistent and reliable tool for the assessment of auditory development in children with CIs. We administered the LEAQ at each data collection point to monitor auditory development and to detect possible difficulties with implant use. Table 2 presents the evolution of auditory development over time. Mean LEAQ total scores (maximum score of 35) over time show an adequate progression of auditory skills in all participants.
Language diary
First, a graduate student translated and adapted the materials accompanying the Di-EL (Nott et al., 2005; Nott et al., 2003) to French, under the supervision of the first author. Both these francophone individuals are proficient in English. Specifically, the original training video was replaced by a written document, and a meeting was set with each family at the time of their enrollment in the study to explain the procedure and to answer their questions. We previously reported the detailed procedure of adaptation and the results of a single-case feasibility study (Pelletier-Bergeron & Duchesne, 2012). Parents were asked to record each new word spoken by their child for a period of 12 months. We followed the recording and coding procedures described in Nott et al. (2003): Each diary entry included the child’s production (transcribed orthographically), the date, and the situation in which the production occurred; parents also had to enter the date of the second spontaneous use of each production. Parents entered these items in the support of their choice (notebook, worksheet, digital file, scrapbook, etc.) and forwarded their records to us on a quarterly basis, for coding and analysis purposes. Following the procedure outlined in Nott’s study (Nott et al., 2003), a research assistant coded entries either as a single word (e.g., “camion” [truck]), a frozen phrase, that is, a fixed combination of two or more words which do not occur separately (Lieven, Pine, & Dresner Barnes, 1992; Pine, Lieven, & Rowland, 1996; e.g.,“c’est chaud” [it’s hot]) or a true word combination (i.e., a multiword utterance, for example, “ferme la porte” [close the door]; see Nott et al., 2003, for a detailed description of coding criteria) at each data collection point. Finally, we had monthly telephone contact with the parents to offer support and encouragement with the use of the diary and to respond to their questions.
MBCDI
Parents were also asked to fill out the French (Quebec) version of the MBCDI every 3 months (Trudeau, 2008). We used the Words and Gestures form for all children regardless of their chronological age because they all had virtually no expressive spoken vocabulary at the time they entered the study. At the last data collection point (12 months of CI use), parents of JD01 and JD07 filled out the Words and Sentences form. Research assistants entered individual word data for each child to compute total vocabulary size and vocabulary composition.
Language samples
A free-play session took place at 3-month intervals at the parents’ home or at the clinic, to collect language samples. Children mostly played with their own toys and interacted with a trained research assistant who was following the child’s interests. At that moment, we also filled out the LEAQ with parental input. Sessions were video-recorded to also include the communicative context. A trained research assistant transcribed the language samples, using a broad phonetic transcription. We used Phon 1.5, a software program designed for the transcription, coding, and analysis of phonological corpora (Rose, Hedlund, Byrne, Wareham, & MacWhinney, 2007). A graduate student in speech-language pathology reviewed the transcriptions that were identified as “uncertain” by the research assistant. Finally, a third listener (an experienced speech-language pathologist) verified and confirmed all the transcriptions. To be considered as a word, a child’s production had to be clearly associated with an intentional communicative attempt (Moreno-Torres, 2013) and to have a minimal degree of phonetic resemblance with the intended target word (i.e., sharing at least a vowel or a combination of vowel and consonant(s) with the target). If a sign was produced with voice, it was considered as a word. Linguistic behaviors observed during the play session were coded similarly as in the language diary (i.e., single word, frozen phrase, or true word combination). Spontaneous productions, as well as the coded diary entries, and the checked words in the MBCDI for each participant were also all listed in Excel sheets.
Parental satisfaction questionnaire
At the end of the 12-month period of data collection, we sent a short questionnaire to the parents to survey their appreciation of the language diary. We asked parents to describe, on a scale of 1 to 5 (ranging from very unsatisfied to very satisfied), their level of satisfaction regarding four aspects of the diary: the format, the ease of use, the relevance to monitor language progress, and the time required to complete the diary. We also included three open-ended questions regarding (a) their self-perceived involvement in their child’s intervention program related to the use of the diary (“Do you have the impression that you were playing a more important role in your child’s intervention program during the time you were completing the diary?”), (b) the strengths and weaknesses of the tool, and (c) any suggestion of improvement they might have.
Analyses
Total number of word types and lexical distribution, in terms of grammatical categories (i.e., nouns, verbs, adjectives, function words, and “sounds and games,” including interjections, greetings, and onomatopoeic words) were examined across data collection points for each language assessment method. Results from parental satisfaction questionnaires were analyzed descriptively, in terms of parents’ overall satisfaction and perception of involvement in the intervention program.
Results
Number of Word Types Produced at Each Data Point With Each Method
As reported in Table 3, all three methods showed a progression in the total number of different words expressed over time. The mean number of words recorded both in the MBCDI and the diary (for all participants, as a group) doubled every 3 months, from 3 months up to 9 months post-implantation. At 3, 6, and 9 months after cochlear implantation, the diary and MBCDI data showed no substantial differences in the mean number of words. The mean number of words expressed according to the MBCDI was 8, 17, and 40; it was 5, 18, and 34 according to the diary. At 12 months, the mean number of words captured by the MBCDI more than doubled compared with the previous data point (from 40 to 96), whereas the progress showed by the diary was less striking (mean number of different words went from 34 to 48). At each measurement occasion, the language sample captured fewer words. As we expected, variability was large (see standard deviations [SDs] in Table 3). Due to this variability and the small sample size, we performed a nonparametric analysis of variance (ANOVA; see Noguchi, Gel, Brunner, & Konietschke, 2012, for a detailed account of the it Longitudinal Data [LD] F2 procedure). We found a significant effect for time (ANOVA-Type Statistic [ATS] = 29.508, degrees of freedom [df] = 1.88, p < .0001) and assessment method (ATS = 5.998, df = 1.96, p = .002). We found no interaction between time and method (ATS = 1.63, df = 3.00, p = .179).
Mean (SD) Total Number of Single Words Produced at Each Point With the Three Assessment Methods for the Whole Group (n = 9).
Note. SD = standard deviation; CI = cochlear implant; MBCDI = MacArthur–Bates Communicative Development Inventories.
Individual Development Curves
Figure 1a to 1i shows individual curves that illustrate expressive vocabulary progression over time with each assessment method. For all participants except JD05 and JD06, we observed a rapid rise in the MBCDI score, especially around 9 months of CI use, with a more gradual increase with the two other measures. For two of the participants (JD01, Figure 1a; and JD10, Figure 1i), the rapid increase started at 6 months of use. Both these participants had sign vocabularies at the beginning of the study (Table 1). Two other participants, JD05 and JD06 (Figure 1d and 1e), made very little progress over the course of the study: Regardless of the assessment method, they did not reach the 10-word milestone after 12 months of CI use.
Individual curves: Expressive vocabulary progression with each assessment method.
Word Combinations and Frozen Phrases
Depending on the method and on the time of data collection, very few children produced frozen phrases and word combinations. The two parents who filled in the grammar section of the MBCDI words and sentences at 12 months reported that their child combined words “sometimes” (JD07) and “often” (JD01). Diary data showed a similar trend: Parents recorded no frozen phrases and combinations during the first 3 months. At 6 months, only two parents recorded frozen phrases and combinations: JD01 (two frozen phrases and one combination) and JD03 (one frozen phrase). At 9 months, four parents recorded one frozen phrase (JD01, JD08, JD09, JD10) and one parent recorded one word combination (JD04). At 12 months, only JD01’s parent recorded one frozen phrase and nine word combinations. Finally, during the play session, no child produced frozen phrases or combinations at the 3 and 6 months data collection points. At 9 and 12 months, JD01 had one frozen phrase and one combination. Only one participant (JD04) produced a significant amount of combinations during the last two play sessions: We recorded 16 and 11 different word combinations at 9 and 12 months, respectively. This particular child (JD04) had 147 signs at the beginning of the study (see Table 1).
Vocabulary Content: Lexical Distribution of Words
We also examined the various grammatical categories (e.g., nouns, verbs) expressed at each data collection point and with each method. We classified words according to the following categories: (a) nouns (including proper nouns), (b) verbs, (c) adjectives, (d) function words (including pronouns, conjunctions, prepositions, adverbs, and questions), and (e) sounds and games (including interjections, greetings, and onomatopoeic words). Figure 2a, 2b, and 2c illustrates the lexical distribution at each data collection point for each assessment method.
Mean proportion (n = 9) of words in each grammatical category with each method: (a) MBCDI, (b) diary, and (c) sample.
As shown in Figure 2, all three methods captured grammatical diversity. Generally, all three figures show an increase over time in the proportion of nouns, accompanied by a decrease in the proportion of sounds and games words. In the MBCDI, proportions of verbs, adjectives, and function words remain relatively low over the course of the study. For example, at 6 months, the mean proportion of verbs expressed was lower in the MBCDI (6%) than in the diary (8%) and the sample (15%). A closer examination of individual data shows that at 6 months, only two children had verbs checked in the MBCDI (JD01 and JD10). Also, the mean proportion of function words was higher in the diary than in the MBCDI at 3, 6, and 9 months of CI use (9%, 12%, and 16% in the diary vs. 7%, 6%, and 6% in the MBCDI). Examination of individual data at 6 months revealed that six children (i.e., all children except JD06, JD07, and JD08) had function words recorded in the diary, whereas only three children (JD01, JD03, and JD10) had function words checked in the MBCDI. A similar trend was observed for adjectives: In the MBCDI, the proportion of adjectives never exceeded 7% over the course of the study, whereas the proportion of adjectives at 12 months was 13% in both the diary and the sample.
Parental Satisfaction
Only five families (JD01, JD03, JD04, JD06, and JD07) returned the questionnaire at the end of the study. According to the parents who responded, the diary was generally easy to use and not too time-consuming: Parents were either “quite satisfied” or “very satisfied” with the format, the ease of use, and the time required to complete the diary. Three parents were either “quite satisfied” or “very satisfied” of the usefulness of the diary to observe their child’s linguistic evolution over time, whereas the two remaining parents were only a little satisfied. Two parents found that the recording of each new word was somewhat demanding. Only one parent reported feeling much more involved in her child’s CI intervention program while completing the Di-EL (“I definitely played a more important, concrete role in the rehab”). Two other parents felt equally involved (“as much”), and the other two did not feel they were more involved in the intervention program (“Didn’t change anything”). Two parents suggested a mobile application for the diary.
Discussion
The purpose of the present study was twofold. First, we aimed to determine whether the three methods of assessment were quantitatively and qualitatively representative of the development of expressive vocabulary over time during the first year of CI use. The methods used were parental report, language diary, and spontaneous language sample. Second, we aimed to document parents’ satisfaction and motivation regarding the use of a language diary.
Regarding our first aim, the results reveal that, as hypothesized, all three methods showed evidence of a progression in the number of words expressed over the course of the study. Similar results were found in Gatt et al. (2014) with 44 typically developing children: All three measures used in that study showed a rise in vocabulary production as chronological age increased, with substantial differences in the mean vocabulary production scores obtained with a parental report and with a language sample (Gatt et al., 2014).
In the present study, over the course of development, quantitative differences between the three methods became obvious: The MBCDI captured more words than the two other methods. During the very first stages of lexical development, it seems that the three methods are equally representative in quantitative terms, at least for some of the children (up to 9 months post-implantation), but overall, the language sample seemed less sensitive to children’s abilities. Nevertheless, for seven of the nine participants, the spontaneous language curve mirrored closely the one obtained with the diary. The poor productivity of the language sample is likely due to the fact that free-play situations are too dependent on a child’s willingness to engage in an interaction and on their particular interests (e.g., JD04 played with a fire truck and a police car for 20 min at T9 and JD09 wanted to play alone for the most part at T12). Rescorla and Ratner (1996) stated that the analysis of naturalistic interactions does not always allow distinguishing “propensities from capacities” (p. 163). Although the idea of designing a standard play-session protocol seems attractive, it may be difficult to engage a 2-year old child in an imposed situation. Sachse and Von Suchodoletz (2008) pointed out the difficulty of obtaining an accurate portrait of language abilities in natural contexts: In an unfamiliar setting or with an unfamiliar communication partner, young children may even refuse to engage in interaction.
Results also show that on average, children produced at least three different words and up to eight, depending on the assessment method, as soon as after 3 months of CI use. As shown in Figure 2, many of these early productions are animal sounds and games and routines words (e.g., miaou [meow] and bye-bye), but a few novel nouns were also reported or observed (e.g., bébé [baby] and maison [house]). Similar results were also reported in Fagan (2015), who found that 4 months after cochlear implantation, nine early implanted children (mean age at CI: 12.46 months) produced, on average, one spoken word, either observed in a videotaped interaction or reported by the parent.
The examination of individual developmental patterns over time showed, as expected, individual variability. We decided not to remove JD05 and JD06 from the study (both participants obtained lower scores than the rest of the group; social-emotional and cognitive subtests of the BSID-III) because both participants represented the large variability that exists in this population. Although JD05 and JD06 were among the youngest to receive a CI in our sample (at 13 and 14 months old, respectively), both children showed virtually no progression in their expressive vocabulary over 12 months. JD05’s deafness was caused by cytomegalovirus (CMV), a condition that is likely to cause additional disabilities, and JD06 was later diagnosed with a global developmental delay. Apart from JD05 and JD06, individual development curves look quite similar. This suggests that regardless of age at CI, vocabulary development seems to follow a pattern related to the duration of use. This finding resonates with a recent study involving 20 bilaterally implanted children at a mean age of 12.9 months who were assessed with the MBCDI (Finnish version) at 1, 3, 6, 9, and 12 months after CI activation (Välimaa et al., 2017). Their results showed that 3 months after activation, children had reached the norms of 12-month-old children with a typical hearing and those of 14-month-old children after 6 months of CI use. Neither age at first hearing-aids fitting nor age at CI activation was associated with early expressive vocabulary development (Välimaa et al., 2017).
Finally, even if the language sample was the least fruitful method in terms of number of single words, one child (JD04) produced a significant amount of word combinations during the play sessions at 9 and 12 months, for example, gros camion pompier (big fire truck), lumière rouge (red light), and creuse un gros trou (dig a big hole). Interestingly, JD04 was one of the oldest when he received his implant (26 months), was the only boy enrolled in the study, and was the only child who consistently used signs along with spoken language. Moreover, he was using the largest number of signs prior to cochlear implantation (see Table 1). It is thus possible that the use of signs had a scaffolding effect on spoken language development. Three other children were using a number of signs when the study started (JD01, JD03, and JD10) and made significant progress over the course of the study (see Figure 1)—so did JD08, who did not use any signs prior to cochlear implantation.
The issue of whether the use of signs might have supported the development of early spoken vocabulary in our small sample remains difficult to decide, and current research has yet to provide us with definitive answers. Nevertheless, there is growing evidence suggesting that sign language input promotes subsequent spoken language acquisition in children with CIs. For example, Szagun (2001) pointed out that the use of signs, by providing early access to symbolic communication, could positively contribute to spoken language development. In a more recent study, Davidson, Lillo-Martin, Chen, and Pichler (2014) showed that five CI users who were native signers obtained spoken language scores comparable with those of 20 children with typical hearing who were also native signers (all participants had at least one deaf, signing parent who was communicating with the child using sign language). Conversely, other authors, most notably Geers et al. (2017), argue that there is no advantage in using sign language with children either before cochlear implantation or immediately after. Finally, Fitzpatrick et al. (2016), in a systematic review of studies conducted between 1995 and 2013, could not determine whether oral language therapy alone was superior to sign language combined with oral language. One might conclude from this systematic review that there is no evidence that the use of sign hinders spoken language acquisition in children with CIs (see Humphries et al., 2017, and Mellon et al., 2015, for expert opinions on the subject).
In terms of the content of early vocabulary, our results show that in the early stages of expressive vocabulary development (especially the first 9 months), the diary and to a lesser extent the language sample allow a refined description of the use of different lexical classes. For example, more children had more function words recorded in the diary at 6 and 9 months than in the MBCDI. These findings suggest that when vocabulary size is still small, the diary might provide a more comprehensive picture of development than a vocabulary checklist. Szagun and Schramm (2015) discussed the importance of function words in early vocabulary development in children with CIs. In their study, the diversity of both determiners and lexical words at the early stages of post-implantation development was correlated with grammatical complexity up to 2 years later (Szagun & Schramm, 2015). In addition, Le, Normand, Moreno-Torres, Parisse, and Dellatolas (2013) found that in French-speaking children with a typical development, the diversity of grammatical words better predicted the increase of grammatical complexity than the diversity of lexical words. Thus, a tool that is more sensitive to grammatical variety during the early stages of expressive vocabulary development can be of value to guide intervention. Szagun and Schramm (2015) stated that the monitoring of early indicators of subsequent language development, such as grammatical words, allows the design of specific intervention strategies very early on, thus maximizing the sensitive period for language acquisition.
The language diary seems to be a useful tool for the monitoring of the emergence of vocabulary, when vocabulary size is still small, and we want to comprehensively describe the use of different lexical classes (e.g., grammatical words vs. lexical words). In contrast, the MBCDI appears best suited for documenting the evolution of vocabulary size over time once a certain number of spoken words are acquired. These findings are in line with Ertmer and Mellon (2001) who described early oral development in a child who received a CI at 20 months. They concluded that a diary had been more useful for describing the development of vocabulary in the very first months after cochlear implantation, whereas a parental report (MBCDI) allowed the assessment of vocabulary growth after the child had reached the 10 spoken words milestone (Ertmer & Mellon, 2001).
Regarding our second aim, parents were generally satisfied with the use of a language diary. However, only five families returned the questionnaire at the end of the study, and one family did not send us the diary data at 12 months. The five respondents did not necessarily complain about the time required to record new words daily, but the fact that four other parents did not return the questionnaire is worth noting. Prezbindowski and Lederberg (2003) as well as Ertmer and Mellon (2001) have discussed the difficulty of obtaining diary data from parents, whether a child has a hearing loss or not.
Conversely, parents in Nott et al. (2003) used the Di-EL for 2 years after implantation. According to these authors, the assiduity of the parents in maintaining a language diary was related to a higher motivation and interest regarding language development because their child had a hearing loss. For that reason, they could have been more attentive to their child’s lexical progress (Nott et al., 2003). In the present study, only a few parents reported feeling more involved in their child’s CI intervention program while completing the Di-EL. Nevertheless, some studies have shown that parents perceive the rehabilitation process as highly demanding (Sach & Whynes, 2005; Zaidman-Zait & Most, 2005). Thus, clinicians should be careful with their demands to the parents, as many face numerous everyday problems, primarily related to the maintenance of the device, communication difficulties, and behavior problems (Zaidman-Zait, 2008). Program requirements for a proactive and intense parental involvement can put unnecessary pressure on mothers (Mauldin, 2016), who are traditionally more involved (than fathers) in intervention programs for children with hearing loss (Ingber, Al-Yagon, & Dromi, 2010).
Limitations of the Study
Limitations include the small number of participants and the fact that two of them had additional disabilities, thus adding to the expected variability. In addition, although we did not observe one, we cannot completely rule out the possibility of a “boosting” effect of the second implant on vocabulary development, as four of the participants received a second implant before the end of the study. Guo, McGregor, and Spencer (2015) have suggested the possibility that bilateral implantation make a difference in early lexical acquisition by increasing the sensitivity to word statistics, although it might not be visible at 12 months post-implantation. Finally, only five parents returned the satisfaction questionnaire (55%) without apparent reason. Thus, our second aim that was to document parental views on the use of a language diary was only partially attained.
Conclusion
Multidimensional assessment, for children who are deaf/hard of hearing, particularly those receiving CIs early in life, is essential to obtaining a complete description of a child’s linguistic profile (Prezbindowski & Lederberg, 2003). In the present study, triangulation of data from three sources provided a thorough description of early expressive vocabulary over the first year of implant use. Results suggest that a language diary might be a solution for documenting the first 10 to 50 words, or for a period of approximately six months, depending on the rate of acquisition of spoken words in each child. While parents were generally appreciative of their experience with the use of a language diary, we are not yet ready to add this tool to our standard language intervention procedures with young children who receive CIs and their families. Although parents should be our primary source of language data in the early stages of lexical development after cochlear implantation, clinicians should take into account the demands imposed by the extended use of a language diary.
Footnotes
Acknowledgements
We thank Julie-Anne Pelletier-Bergeron, Émilie Lacroix, Maude Lemieux, and Claudine Lévesque, who participated in the French adaptation of the Di-EL (Diary of Early Language), data collection, and analyses. Thanks to Jean Leblond, PhD, who provided support with the statistical analyses. We also thank Robert Cowan (and Isabelle Boisvert) from the HEARing Cooperative Research Centre (CRC) for granting us the permission to adapt and use the Di-EL.
Authors’ Note
Some of the data discussed in this article were presented at the 13th International Conference on Cochlear Implants and Other Implantable Auditory Technologies (2014) and at the American Speech-Language and Hearing Association (ASHA) Conference (2015).
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a Grant from the Réseau Provincial de Recherche en Adaptation-Réadaptation (REPAR) (Partnership in Hearing Impairment Institut de Réadaptation en Déficience Physique de Québec [IRDPQ]/Institut Raymond-Dewar [IRD]/REPAR).
