Book Chapter
Reference
Language development in children with mild-to-moderate hearing loss: an overview
DELAGE, Hélène, TULLER, Laurice
DELAGE, Hélène, TULLER, Laurice. Language development in children with mild-to-moderate hearing loss: an overview. In: Most, T. & Ringwald-Frimerman, D. Theoretical and applied aspects of rehabilitation and education of hard-of-hearing and deaf children and adolescents. Tel-Aviv, Israel : MOFET Publishing House, 2014.
Available at:
http://archive-ouverte.unige.ch/unige:154076
Disclaimer: layout of this document may differ from the published version.
Language development in children with mild-to-moderate hearing loss: An overview Hélène Delagea and Laurice Tullerb
a Université de Genève, Equipe de Psycholinguistique, Switzerland
b Université François-Rabelais de Tours, UMR Inserm 930, France
We propose in this chapter to provide a current overview of language development in children and adolescents with mild-to-moderate hearing loss. Relatively few studies address this population, although it is the largest one among deaf children. After presenting results from the literature, we will focus on syntactic development, which is highly likely to be impaired (compared to other domains) in this population. Our specific goal is to contribute to better understanding of the prevalence, nature, and specificity of morphosyntactic impairment in participants with mild-to-moderate hearing loss, and, in particular, of the commonly found high inter-subject variability. To achieve these goals, we have selected specific aspects of French which we believe are particularly indicative of language impairment, such as accusative pronoun production or rates of production of clausal embedding. The different studies we will present involve a large number of participants ranging in age from childhood to adolescence and a variety of methodologies, including comparisons with other types of pathology.
1. Acquiring Language with a Mild-to-Moderate Hearing Loss
What is the effect of prelingual mild-to-moderate hearing loss on language development in children and adolescents? Although the scientific literature is sparse, it is clear that many of these children experience considerable difficulty acquiring oral language. We report in this chapter on a series of studies we have conducted involving a large number of French-speaking children and adolescents aged 6 to 16, all of whom have bilateral sensorineural, prelingual hearing loss. Mild-to-Moderate Hearing Loss (MMHL) corresponds to an average hearing
loss (as measured by pure tone average) in the range of 21 to 70 decibels (21-40 dB for mild hearing loss, 41-70 dB for moderate hearing loss, according to the official audiometric classification of the International Bureau for Audiophonology). This degree of congenital, sensorineural hearing loss is much more frequent in children than severe and profound
hearing loss. So, for example, Russ et al. (2003) found that 75% of a group of 134 Australian children fitted with hearing aids between the age of 0 and 6, had mild or moderate hearing loss (respectively 42% and 33%), compared to severe and profound hearing loss (16% and 9%). Since not all children with mild hearing loss use hearing aids, this means that the total proportion of children with mild hearing loss is necessarily bigger than the figure reported in this study. We can also report results of Fortnum et al. (2002) who found 53% moderate hearing loss in a group of 17 160 British children ages 3 to 18 with hearing loss over 40 dB.
The technical improvement, and generalization of cochlear implantation in children with profound (90-120 dB), and even severe (70-90 dB) hearing loss means that that the number, and the proportion, of children functioning with mild or moderate HL is now even greater1. Indeed, hearing gains due to cochlear implantation result in PTAs of 30 to 45 dB (Verbist, 2010), and thus to levels equivalent to mild or moderate HL.2 The question of language development in the context of MMHL therefore currently takes on new significance, in terms of the number of children involved, but also in terms of the relevance of this population as a control group in studies evaluating children with cochlear implants. Understanding language development in this context thus has broader implications and applications.
In countries such as France where neo-natal hearing screening is not yet universal, MMHL is frequently detected relatively late: around the age of five (Delage, 2008; Delage &
Tuller, 2007, 2010; Tuller & Jakubowicz, 2004). And, even in countries where there is
1 Of course, aided hearing in children with MMHL means that their hearing thresholds are no longer in the mild- to-moderate range. However, not only is their condition often be detected relatively late, aided hearing may in many case remain impaired.
universal neo-natal screening, as in Israel, it should be emphasized that this screening can only detect hearing loss greater than 40 dB (Govaerts, 2002; Vohr et al, 2012), and thus it cannot identify children with mild hearing loss.3 Late detection of hearing loss in children with MMHL thus means that many of these children experience several years of degraded language input, years which fall squarely within the temporal window typically regarded to constitute a critical period for language acquisition (Bishop et al., 1993; Bortfeld &
Whiteburst, 2001; Newport et al., 2002). Studies of the classical context for critical period effects, that of second language acquisition (see Johnson & Newport, 1989, 1991), have shown that language acquisition after the age of 6-7 does not resemble native acquisition in its outcome, particularly in formal aspects of language (phonology, morphosyntax). Studies of children with cochlear implants have also found strong correlations between language performance and age of implant, highlighting the existence of critical periods for both maturation of language and maturation of audition (Sharma et al., 2002, 2005).
Another, obvious, but important fact about children with MMHL is that, all other things being equal, they have a normal language acquisition device4, meaning that, putting aside for the moment the question of co-morbidity, any language impairment found in this population would be a consequence of factors external to Universal Grammar and thus this population constitutes a particularly interesting comparison group for children with Specific Language Impairment (SLI), for which proposals about the source of language impairment include the suggestion that language competence is altered in these children (van der Lely, 2011). Such comparative studies are so far rather scarce (Briscoe et al., 2001; Norbury et al., 2001, 2002; Hansson et al., 2004, 2007), though results obtained so far point to remarkable similarities.
2 Children with MMHL are not CI candidates since conventional hearing aids result in aided hearing at/below level obtained by CI, and, of course, destruction of a (partially) functioning cochlea raises clear ethical questions.
3 Current technology entails an unacceptable number of false positives if levels below 40 dB are used.
Previous studies focused on language performance in children with MMHL have shown significant rates of language impairment, especially in the areas of phonology and
morphosyntax, although all of these studies have pointed out considerable inter-subject variability (see for example Gilbertson and Kamhi, 1995; Briscoe et al., 2001; Norbury et al., 2001, 2002; Tuller and Jakubowicz, 2004; Hansson et al., 2004, 2007, Friedmann &
Szterman, 2006, 2011; Delage and Tuller, 2007 and Friedmann & Haddad-Hanna, in press).
Studies specifically focusing on oral language development in children and adolescents with MMHL are in fact quite rare (see Delage & Tuller, 2007 for a detailed review). Moreover, in terms of the language domains explored in previous work on language in children with MMHL, studies targeting specific linguistic variables are rather limited. Most studies have used global language measures, and it is thus known that there are language disorders in this population, but little is known about their nature and in particular about the common
denominator behind morphosyntactic properties that are difficult for children and adolescents with MMHL. Finally, exploring the nature of impairment entails comparing the language difficulties in this population with those observed in other studied populations, and in
particular in SLI. However comparative studies are few (and, in the domain of morphosyntax, they are restricted to comparisons with SLI and epilepsy in our own work (Tuller and
Jakubowicz, 2004; Delage and Tuller, 2007; Tuller et al., 2011) and with SLI by Bishop, Norbury and colleagues (Briscoe et al., 2001; Norbury et al., 2001, 2002)).
The goal of this paper is to present data addressing these little studied questions about language development in children and adolescents with MMHL. We will draw on different studies we have conducted on French-speaking children and adolescents with MMHL
(altogether, these include 82 participants aged 6 to 16, with prelingual, bilateral, sensorineural
4 Notice that we analyze our data under the assumption of the nativist theory (in the framework of universal grammar) and then use its terminology.
HL, using their hearing aids consistently and with PTAs ranging from 27 to 69 dB5). A
growing body of literature has presented results which support that hypothesis that similarities between language properties which take longer to emerge in young typically developing (TD) children and those which cause particular difficulties in children developing language in unusual circumstances (such as SLI) are due to the complexity of the linguistic computation underlying these properties (Jakubowicz, 2005, 2011; Jakubowicz and Tuller, 2008; Prévost et al., 2010; Jakubowicz and Strik 2008; Soares, 2006). In these studies, complexity has been measured by reference to linguistic operations: number of instances of syntactic movement operations or of agreement operations, and more generally, level of syntactic embedding (Hamann et al., 2007). We will pursue this line of investigation here to explore the nature of morphosyntactic difficulties experienced by children with MMHL.
Section 2, after presenting general measures of morphosyntax derived from
standardized language assessment tools, focuses on different measures of computationally complex aspects of morphosyntax in French: production of accusative clitics, which has been argued to be a clinical marker of language impairment in French, and then performance on syntactic embedding in general. In keeping with the points outlined above, we will be comparing language development in children with MMHL to TD children and adolescents, but also to individuals developing language in other atypical situations, with pathology (SLI) and without (second language acquisition).
Section 3 addresses the inter-subject variability found in the studies presented in Section 2, which mirror in this respect other studies of language in children with MMHL. We examine in particular the variables of degree of hearing loss and age. A general discussion is presented in Section 4, which returns to the basic question of the nature of the link between MMHL and language impairment, and suggests that this link may be the result of interface between
5 Notice that we have compared results of children with mild and those with moderate HL.
language and areas of executive systems. Finally, we conclude our comment by addressing some recommendations about education and clinical therapies conducted with these children.
2. Morphosyntactic Impairment
This section reports on morphosyntactic performance in several groups of French- speaking children and adolescents with MMHL. First, we will present results indicative of general performance, as measured by standardized tests. We will subsequently focus on particular syntactic structures, reporting on results using a variety of different kinds of measures: elicited production of accusative clitics and frequency and type of syntactic embedding in spontaneous language samples. In all of these different language evaluations, children worse their hearing aids. Language in children with MMHL is sometimes described as giving the illusion of normality (see Delage 2008). We have sought to use different types of measures in an effort to understand what might be behind this observation. Do (some of) these children avoid structures which cause them difficulties?
2.1 General Performance
How do children with MMHL fare on standardized language tests designed to assess children for language impairment? We compared 32 children aged 6;1 to 11;11 (M = 9;2, SD
= 1;9) with MMHL ranging from 29 to 69 decibels to 10 children with SLI aged 8;0 to 11;11 (M = 10;1, SD = 1;4) (Delage, 2008). The participants were administered a computerized oral language battery (the BILO battery, Khomsi et al., 2007). In that battery, expressive grammar is evaluated using a sentence completion task: The child, listened to a sentence corresponding to a picture on a computer screen and then had to complete the end of the sentence to fit with
the picture6 (e.g., Ici, le chien boit son lait; là les chiens… boivent leur lait ‘Here the dog is drinking his milk; There, the dogs… are drinking their milk’). This test assessed ability to produce a variety of specific grammatical morphemes: nominal, adjectival and verbal inflexion, irregular plurals, prepositions, passives, and pronominal clitics. The
morphosyntactic comprehension test consists of a task assessing ability to judge the well- formedness of sentences presented orally with a related picture. The child has to decide whether the sentence heard with the picture is grammatically correct, by giving a yes or no answer (e.g., a picture of several birds sitting on a tree branch is shown and the child hears the following (ungrammatical sentence): *Les oiseaux est sur la branche ‘The birds is on the branch’).
Although the children with MMHL scored significantly better than the children with SLI on each of these two tasks (for expressive grammar, U = 45, p <.001; for receptive grammar, U = 65.5, p <.01), their mean scores showed that morphosyntax, as tested by such general measures, is impaired in these children: Their mean z-score was -1.1 SD below test norms7 for expressive grammar (-3 for the SLI group) and -1.4 for receptive grammar (-4 for the SLI group). There was high inter-subject variability: Some children had severe
impairment, others were not severely impaired, and some were not impaired at all. To illustrate this point, consider Figure 1 which presents individual z-scores of the 32 children with MMHL for the expressive grammar task; it shows that many of the low-scoring children obtained very low scores: 37% of them had SDs between -1.65 and -6. Over-all, 50% of the children displayed some degree of language impairment (they obtained low scores, as defined by a SD equal or below -1), a rate which is typical for studies of children with MMHL (see for example Gilbertson and Kamhi, 1995, Briscoe et al., 2001 or Hansson et al., 2007). In
6 As children are hearing impaired, several repetitions of the stimuli were authorized: on request for the majority of children and almost systematically for the youngest children or those who demonstrated some fluctuating attention.
7 As we mention z-scores, the normal value equal 0.
addition, it is interesting to note that there was no significant difference between children with mild HL and children with moderate HL, for either production (U = 72, p = 0.13) or
comprehension (U = 89.5, p = 0.40).
Fig. 1. Expressive grammar: Standard deviations obtained by children with MMHL compared to mean SD obtained by children with SLI
These two morphosyntactic tasks were two of the six tasks of the oral language battery all participants were administered. The other tasks tested phonology (a word repetition task), expressive vocabulary (a picture naming task), lexical judgment (a word-picture matching task) and sentence comprehension (a sentence-picture matching task)8. Interestingly, if we focus exclusively on the children with MMHL who had low global results on the entire language battery (the 16 children having at least 2 scores < -1.65 SD among the 6 scores in the battery), children we have dubbed as “MMHL with Language Impairment (MMHL-LI)”
in Figure 2, the group of children with MMHL no longer statistically differs from the group of children with SLI for the two morphosyntactic tasks (for expressive grammar, U = 43, p = 0.08 and for receptive grammar, U = 57.5, p = 0.33), as illustrated by Figure 2. In other words, a subgroup of children with MMHL had performance quantitatively very similar to that of children with SLI. This subgroup of ‘MMHL-LI’ is significantly younger than the
16/32 participants
≤ -1SD Mean SD of children with SLI
other participants with MMHL (U = 74.5, p <.05), but, interestingly, not deafer (U = 93, p = 0.20).
Fig. 2. Expressive and Receptive Grammar: Mean standard deviations obtained by the 16 children with MMHL with language impairment (MMHL-LI) and by the 10 children with SLI
Summarizing, standardized language measures of morphosyntax identified
approximately 50% of this group of children with MMHL as having impaired morphosyntax, a result which joins those of previous studies, including our own. Furthermore, as a group, the children with MMHL with over-all impaired language resembled children with SLI, as a group, in terms of the severity of their morphosyntactic impairment.
2.2 A Clinical Marker of Language Impairment in French: Accusative Clitic Production Standardized tests of the type presented in the preceding section give a general picture of the degree of impairment. However, they are not designed to explore which areas of morphosyntax cause difficulties and which ones do not. Tasks focused on specific language properties allow for such exploration and thus hold the promise of furthering understanding of the nature of language impairment in children with MMHL. We would like to know which areas of morphosyntax are affected and whether these can be directly linked to hearing loss.
8 As expected, children with MMHL displayed more difficulties in phonology (47% obtained scores < - 1.65 SD) than in vocabulary (25% with such low scores) and oral comprehension (3% with such low scores).
Tuller and Jakubowicz (2004) explored the question of whether perceptual limitations resulting from hearing loss lead to difficulty with grammatical elements whose surface form is less salient. To do this, we compared several such grammatical items, and, in particular, definite determiners and third person accusative clitic pronouns (see Table 1). These
particular morphemes were chosen since, although their phonological form is identical, they differ in terms of the complexity of the linguistic computation required to derive them.
Notably, while definite determiners appear in the canonical position for determiners in French (at the beginning of the noun phrase), accusative clitics do not appear postverbally in the canonical direct object position, but rather preverbally (see Tuller et al, 2011, for detailed discussion of the complex properties of third person accusative clitics).
Table 1. Definite Determiners and Direct Object Pronouns (Accusative Clitics) in French
Definite Determiners Direct Object Pronouns (= Accusative Clitics)
Singular Plural Singular Plural
Masculine Feminine Masculine Feminine
le la Les le la les
le garcon
the boy la fille
the girl les filles
the girls Max le voit
Max him sees Max la voit
Max her sees Max les voit Max them sees
Elicited production of determiners was compared to that of third person accusative clitics in 20 children with MMHL aged 6;0 to 13;9 compared to 10 children with SLI aged 6;0 to 8;6 and to 36 young typically-developing 3-, 4-, and 6-year-olds (12 children per age group). The children with MMHL were divided into two age groups: 10 children aged 6 to 8 and 10 children aged 9 to 13. An experimental probe was used in which children were shown a picture and asked a question, as in (1), eliciting a definite determiner, and (2), eliciting an accusative clitic.
(1) [picture of a girl sprinkling Minnie with a garden hose]
Experimenter: Qui arrose Minnie? ‘Who is sprinkling Minnie?’
Expected response: la fille ‘the girl’
(2) [picture of a girl aiming a water pistol at Minnie]
Experimenter: Que fait la fille à Minnie? ‘What’s the girl doing to Minnie?’
Expected response: Elle la vise ‘She’s aiming at her’
As illustrated in Figure 3, accusative clitics were less well mastered than definite determiners by children with MMHL, as by children with SLI and by 3-year-old control children.
Performance in all groups of children reached ceiling for determiners, for which production rates were above 95%, whereas accusative clitics were produced at much lower rates by children with MMHL (40% in younger children and 68% in older children), by TD 3- and 4- year-olds (44% and 79%, respectively), and by children with SLI (only 29%). Tuller and Jakubowicz (2004) concluded that these results showed that perceptual salience is not the factor which determines morphosyntactic performance in children with MMHL (or with SLI), since only one of two phonologically identical morphemes with low salience9 is a source of difficulty and suggested therefore that the increased syntactic complexity involved in one of them (accusative clitics) is the key factor in accounting for difficulty, in learners with both typical and atypical context of language development.
Fig. 3. Production (%) of Determiners and Accusative Clitics: MMHL, SLI, TD-3, TD-4, TD-6
As mentioned above, it has been argued that accusative clitic production is a robust marker of morphosyntactic difficulties in children acquiring French. Not only do difficulties persist into adolescence (see Tuller et al. 2011), they are found in the context of very different pathologies, in MMHL and SLI, but also in benign childhood epilepsy (see Tuller et al. 2011;
Monjauze 2007) or in children acquiring French as a second language in Canadian immersion schools (Paradis and Crago 2003; Paradis 2004). Given the circumstances of language
acquisition in children with MMHL, and in particular the fact that timing issues are involved (accessible input beginning late for many of these children due to late detection, and
remaining less than optimal due to limitations of hearing aids), we hypothesize that
comparison with TD children learning a second language (L2) after early childhood would be interesting in that the second language input (French) for these children starts late. Is
performance in these two groups comparable for this marker of morphosyntactic difficulty in French?
We present here therefore a second comparative study of accusative clitic production, in a different group of children with MMHL. 32 children with MMHL aged 6 to 11 (Delage, 2008) were compared to 3 groups of TD children (12 6-year-olds, 12 8-year-olds, and 12 11- year-olds), to 18 children with SLI (aged 6-12), and to 29 first language British English- speaking children acquiring L2 French (aged 6-12), naturally, after family immigration to France.10 In this study, we tested accusative clitic production, but also production of two other types of pronominal clitics, nominative and reflexive. Nominative clitics do not involve movement to a non-canonical position, and, reflexive clitics have a local and non-ambiguous antecedent, whereas accusative clitics involve both appearance in a non-canonical (preverbal)
9 Notice that determiners were not always produced at the beginning of the sentences. This fact contrasts with Slobin’s operating principles (Slobin, 1973) for which morphemes appearing at the beginning of a sentence are more salient.
10 The mean age of first exposure to French, upon arrival in French was 7;0 (SD 1;10 and range 4;5 to 10;9), and the mean length of exposure was 2;5 (SD 1;2). See Haiden, 2011.
position and their antecedent is necessarily non-local (see Jakubowicz et al., 1998; Tuller et al., 2011). Given these differences in the complexity of these different elements, we expected to find, like previous researchers, that accusative clitics would be less well produced than the other two types of clitics. To elicit these clitics, we developed a Production Probe for Pronoun Clitics (PPPC), which elicits production of third person nominative, reflexive, and accusative clitics, by requiring a response to a question about a drawing appearing on a computer screen, as in (3) and (4).
(3) Experimenter: Que fait Thomas? ‘What’s Thomas doing?’
Expected response: Il se pèse. ‘He’s weighing himself’
(4) Experimenter: Que fait Marie avec le chien? ‘What’s Mary doing to the dog?’
Expected response: Elle le lave. ‘She’s washing it’
Table 2 presents the results obtained by each group tested. Accusative clitics were produced at lower rates than both nominative and reflexive clitics, in each of the “atypical”
groups (i.e. MMHL, SLI, and L2). This was also the case for the youngest TD group, i.e. the 6-year-olds, but not for the TD8 and the TD11 groups, which displayed ceiling performance.
Inter-group comparisons revealed a significant group effect, with the MMHL performing worse on accusative clitics than the TD8 (U = 79, p <.01) and the TD11 (U = 55, p <.001), though not than the TD6, and better than the SLI (U = 182.5, p <.05). Interestingly, there was no significant difference for production of this clinical marker between children with MMHL and L2 children.
Table 2. Production (%) of Nominative, Reflexive and Accusative Clitics: MMHL, TD, SLI, and L2
Group Age range Nominative Reflexive Accusative MMHL (N=32) 6;1-11;11
(M = 9;2) 91%
(SD = 13) 87%
(SD = 27) 65%
(SD = 33)
Accusative < Nominative (p
<.001)
Accusative < Reflexive (p
<.001) SLI (N=18) 6;6-12;11
(M = 9;5) 76%
(SD = 36) 72%
(SD = 36) 42%
(SD = 36)
Accusative < Nominative (p <.01)
Accusative < Reflexive (p
<.01) L2 (N=29) 6;4-12;7
(M = 9;5) 85%
(SD = 21) 72%
(SD = 35) 61%
(SD = 35)
Accusative < Nominative (p < .01)
Accusative < Reflexive (p
< .05) TD6 (N=12) 6;1-6;7
(M = 6;4) 98%
(SD = 2) 100% 84%
(SD = 11)
Accusative < Nominative (p
<.01)
Accusative < Reflexive (p
<.01) TD8 (N=12) 7;9-8;7
(M = 8;2) 98%
(SD = 6) 100% 96%
(SD = 7) TD11 (N=12) 11;1-11;9
(M = 11;4) 94%
(SD = 9) 100% 98%
(SD = 3)
In summary, two different groups of French-speaking children with MMHL (a total of 52 children, aged 6 to 13, mean age 9) displayed the same specific difficulty with third person accusative clitics found in younger TD children, in children the same age with SLI, and in children the same age learning French as a second language after age four. While children with MMHL, as a group, performed better than age-matched children with SLI, their performance was quite comparable to that of the L1 English/L2 French children.
2.3 Embedded Clauses in Spontaneous Language Samples
Accusative clitics are notoriously difficult for different learners of French, and this difficulty includes children with MMHL, as we have just seen. Investigation of other structures entailing high levels of syntactic complexity will allow us to see to what extent the context of MMHL leads to language impairment. We then turn to the production of different types of clausal embedding in spontaneous language samples.
We examined the production of embedded clauses in spontaneous language samples in the group of 32 children with MMHL aged from 6 to 11 (M = 9;2, SD = 1;9) presented in the
preceding subsections (Delage, 2008). These samples were compared to those from 18 children with SLI aged from 6 to 12 (M = 9;5, SD = 1;10) and to 36 control children aged from 6 to 11 (M = 8;8, SD = 2;1). 60 utterances per child were transcribed and coded in CHAT and analyzed with CLAN tools (McWhinney, 2000). Figures 4 and 5 present results for MLU and for rate of embedded clauses (= N subordinate clauses / N verbal utterances).
Fig. 4. MLU: MMHL, SLI and TD groups Fig. 5. Rate of embedded clauses:
MMHL, SLI and TD groups
Although children with mild HL do not differ from those with moderate HL for MLU (U = 74.5, p = 0.15), they do produce significantly more embedded clauses (U = 48.5, p <.05). The mean MLU of the total group of children with MMHL was significantly lower than that of the control group (U = 200, p < .001). They also used fewer embedded clauses than controls (U = 296, p < .001). However, these rates were very convergent in the MMHL and the SLI groups.
This similar pattern was also found for rates of production of relative clauses: Children with MMHL and those with SLI produced on average 2.7 and 3.4 relative clauses, respectively;
these numbers did not differ significantly, but they were significantly lower than the mean number of relative clauses produced by TD children (M = 5.7, U = 317, p <.01). Another point of convergence between the MMHL and SLI groups concerned the use of avoidance mechanisms. Indeed, a large number of subordinate clauses attempted by children with MMHL (14%) and by SLI children (18%) did not surface as embedded clauses, a
phenomenon which was less frequent in TD children (only 3%). We suggest that the higher rate of failed subordination can be interpreted as avoidance of embedding. Failed attempts at subordination consisted of two predominant strategies: self-interruption of a subordinate clause, which was then abandoned, as in (5), and avoidance of a relative clause, by use of juxtaposition of two simple clauses, as in (6).
(5) Il se passe des histoires d'une reine qui portait un collier qui…
‘There are stories about a queen who was wearing a necklace that…’
(6) Y a Samy il prend d(e) la Chantilly
‘There’s Sammy he’s having whipped cream’
Turning to production of erroneous utterances (utterances containing at least one morphosyntactic error), presented in Figure 6, all rates measuring error production significantly differentiated the children with MMHL from the children with SLI (simple erroneous utterances, U = 119.5, p <.001, and complex erroneous utterances, U = 124, p
<.01). Although children with MMHL produced fewer errors than children with SLI, they produced significantly more erroneous utterances than TD controls, both for simple and complex utterances (U = 94, p <.001). Finally, the MMHL and SLI groups showed the same over-all pattern: They produced more complex erroneous utterances than simple erroneous utterances (MMHL: T = 45, p <.001; SLI: T = 21, p <.01).
Fig. 6. Rates (%) of Simple and Complex Erroneous Utterances: MMHL, SLI and TD Controls
3. Variability in Language Performance in Children with MMHL
In the preceding section, we saw that the French-speaking children with MMHL that we have studied, as a group, performed below TD children on standardized measures of language, and also on specific properties of French that are known to be difficult for young TD children, and for children and adolescents with SLI. Children with MMHL look very similar to children with SLI, displaying difficulties in the same areas. However, these are group results, and though children with MMHL performed below TD children of the same age, they generally performed better than children with SLI. This intermediate position could have two sources:
all children with MMHL have language impairment, but this impairment is simply less severe than it is in children with SLI, or, alternatively, groups means are masking very different performance, some children performing very much like children with SLI and other children performing very much like TD children. Inspection of individual results clearly shows that the latter scenario is the relevant one. Indeed, this has been one of the stable results in studies of language in this population (Briscoe et al., 2001; Norbury et al., 2001, 2002; Tuller and Jakubowicz, 2004; Hansson et al., 2004; Delage and Tuller, 2007; Hansson et al., 2007). Our recent studies confirmed this result; we illustrate this by returning to the study of the
production of third person accusative clitics in 32 children with MMHL (Delage, 2008).
Figure 7 presents the individual results for these 32 children. It can be seen there that 13 children had production rates superior to 87%, close to ceiling. The other 19 children produced these clitics at rates varying from 0% to 75% (whereas 8- and 11-year-old normal hearing children have a mean performance on this task between 93% and 98%).
Fig. 7. Production rates of accusative clitics produced by children with MMHL ranked by age
Which clinical variables can account for this variability? We explored several clinical variables which might be expected to explain this heterogeneity in this group of children with MMHL. Some of these turned out to be entirely irrelevant (i.e. no correlations were found with any language measures): socio-economic status, non-verbal capacities (evaluated via Raven’s Progressive Matrixes, Raven et al., 1998) and age of detection of HL/of hearing aid fitting.11 One logical possibility is that of co-morbidity with SLI in some children with MMHL; these children would thus be deaf children who also happen to have SLI. However, as has been previously pointed out (Briscoe, 2001; Delage & Tuller, 2007), it is clear that co- morbidity cannot fully account for the proportion of children with MMHL who display
morphosyntactic impairment (around 50% in standardized tests, as was seen above, in Section 2.1), as prevalence for SLI is estimated at 7% (Tomblin et al., 1997). In other words, co- morbidity could only explain a very small part of the variability, and thus we are left with the initial problem of explaining why it is that some children with MMHL seem to have far greater difficulty with language than others (including those who have no difficulties at all).
11 Notice however that this can be masked by the degree of HL: A more severe HL may be identified earlier than a milder one. In the same way, children who demonstrate obvious difficulties in acquiring oral language may be detected earlier than children with less evident impairment.
At or near ceiling:
13/32 children
In the remainder of this section, we will focus on two variables which deserve deeper consideration: severity of hearing loss and age.
3.1 Severity of hearing loss
The considerable linguistic heterogeneity reported in studies focusing on children with MMHL has generally, and unexpectedly, not been correlated with degree of HL, except on phonological measures (Briscoe et al., 2001; Sahlén and Hansson, 2006; Hansson et al., 2007). Delage and Tuller (2007) presented evidence suggesting that the link between language impairment and severity of HL might be easier to detect in adolescents than in children. In that study, we found a significant (negative) correlation between degree of HL and performance in expressive grammar in 19 adolescents with MMHL aged from 11 to 15.
We then suggested that absence of such correlations in children might due to differences in developmental rhythms, that naturally occur in childhood, obscuring the effects of variation in hearing loss during childhood. As linguistic development is completed at adolescence, the effect of degree of HL can then emerge. Using more refined measures, we have since been able to detect an influence of degree of HL on morphosyntactic performance, even in children (Delage, 2008). Figure 8 shows that there is a (negative) link between rates of complex utterances (i.e. rate of utterance containing one or more embedded clauses) produced by the group of 32 children with MMHL aged 6-11 and degree of HL (rs = -0,51, p <.01), i.e. the greater the HL of the children, the fewer complex utterances they produced. This correlation was found for a few other syntactic measures, such as for Z-scores obtained on the
standardized task of expressive grammar (rs = -0,44, p <.05). However, it was not found for all measures. For instance, in the standardized receptive grammar task, Z-scores obtained by the 32 children with MMHL were not linked to the severity of HL, as illustrated by Figure 9.
Fig. 8. Production rates of complex utterances produced by children with MMHL ranked by severity of HL
Fig. 9. Receptive grammar: Standard deviations obtained by children with MMHL, ranked by severity of HL
Summarizing, the influence of severity of HL on performance on language measures is not consistent. Some measures have been found to be correlated, but others are not. This pattern would seem to suggest that the link between hearing loss and language is not a simple one.
3.2 Age effects and the issue of language normalization
Studies focusing on children with MMHL have also found an age effect on syntactic performance (Norbury et al., 2001, 2002; Tuller and Jakubowicz, 2004), which might explain part of the variability (older children performing better than younger). This would suggest that
children with MMHL have only delayed language development, and that they will catch up later on, at the end of childhood or in adolescence. This question raises the issue of language normalization with age. Delage and Tuller (2007) showed that language does in fact not normalize with age in subjects with MMHL. This study examined language performance in 19 adolescents with MMHL aged 11;9 to 15;1, using standardized tasks (the BILO battery, see section 2.1) and elicited production of pronominal clitics (see section 2.2). No age effect on language performance was found in this study, although inter-subject variability was still high: More than 50% of the adolescents displayed clear long-lasting language impairment.
So, for example, 63% of these adolescents obtained standard deviations (SD) less than -1.65 on a word repetition task which assesses phonology (a task which is, of course, specially hard of hearing impaired children due to disturbed auditory perception of the stimuli) and 32% of them did on expressive and receptive grammar tasks.12 Many of these adolescents also displayed the same difficulty with accusative clitics that we reported in Section 2.1 for children with MMHL: half of the adolescents produced third person accusative clitics with rates lower than 75%, whereas 11-year-old control children were entirely at ceiling (mean rate of 98%).
4. General Discussion
Our studies have shown that limited HL has significant effects on the acquisition of morphosyntax in French-speaking children with MMHL, but that there is high inter-subject variability. Some children with MMHL show difficulties as severe as those displayed by children with SLI, whereas others appear to have completely normal language. Detailed investigation of specific structures in the language performance of these children showed that children with difficulties have particular trouble with constructions that entail complex
12 The adolescents with MMHL were compared to 12 adolescents with SLI (aged from 11 to 19), who displayed this same pattern, namely a predominance of phonological and morphosyntactic impairment.
syntactic computation in French, such as accusative clitics, as well as, more generally, clausal embedding, and these are manifested not only in errors, but also in use of avoidance strategies (production of structures which are computationally simpler). Our cross-population
comparisons have also demonstrated that the areas of morphosyntax which cause difficulty for children with MMHL cannot be directly related to HL, since these are the very same areas which cause difficulties for young TD children and for participants in other atypical contexts of language development, with and without pathology (SLI and L2 acquisition). We take this to be support for the hypothesis that areas of difficulty are above-all related to the complexity of the computation underlying their derivation.
Summarizing, three clear results have emerged from the different studies reported here:
1) MMHL very often leads to morphosyntactic impairment which does not disappear with age, 2) specific areas of difficulty are shared with other acquisition contexts (such as
accusative clitic production or clausal embedding in spontaneous language, also impaired in individuals with SLI), and 3) these areas can be defined by those constructions which entail complex linguistic computation. What is much less clear is why it is that some individuals with MMHL seem to develop language normally or with very little difficulty, whereas others experience considerable and long-lasting difficulty. While obvious language-external
variables, such as non-verbal intelligence or family socio-economic status, have failed to provide any explanations for this variability, both age and degree of HL appear to be at least somewhat linked to language performance in MMHL, although neither of these variables satisfactorily accounted for the degree of observed heterogeneity.
Regarding the influence of the severity of HL on language performance, significant correlations between these are not at all consistent across language measures. Moreover, some studies have failed entirely to find correlations between syntactic performance and degree of HL (see for example Norbury et al., 2001, 2002 or Tuller and Jakubowicz, 2004). The results
reported here included children which displayed opposite patterns of language performance despite having the same degree of HL. It would seem therefore that the link between HL and morphosyntax is not a direct one. We believe that the relevant factors could be found in performance systems which interact with linguistic competence. We agree with Borg et al.
(2007) that working memory is one rather obvious candidate. It seems safe to assume that HL places a burden on attention, in that more resources are required for attending to auditory input, and thus fewer are available for other types of processing, such as working memory.
Such a scenario put into a developmental context may have as a result that working memory capacity, having been “deprived”, does not mature normally, or not as rapidly as in normally- hearing children. According to this line of reasoning, insufficient availability of resources essential to the maturation of working memory capacities would result in immaturity of this performance system, which could in turn have long-term effects on language performance.
This is supported by the well-known fact that working memory capacities are impaired in children with more severe HL (Alegria et al., 1999; Burkholder and Pisoni, 2003). It has moreover been suggested that working memory capacities are sensitive to the syntactic complexity of a given linguistic derivation (Jakubowicz, 2005, 2007; Jakubowicz and Tuller, 2008). Thus, if some children with MMHL are indeed impaired in working memory, they may have difficulties computing more complex syntactic constructions (as we have seen here), which require more working memory resources. Of course, this hypothesis requires further investigation on working memory capacities in children with MMHL
5. Educational and clinical perspectives
In this chapter, we have provided an overview of the linguistic development of children with MMHL. We now consider various therapeutic repercussions of the identified
characteristics. To begin, we have underlined the extreme variability in the linguistic abilities
of these children. For some with this condition, language skills appear to be intact; for others with the same degree of hearing loss, a linguistic profile similar to that attested in children with Specific Language Impairment has been observed. Two pitfalls are therefore to be avoided by professionals treating children with MMHL: 1) On the one hand, one should take care not to underestimate the potential difficulties in language development faced by these children: It is not because some are capable of compensating for their hearing loss that all are able to do so. Failing to consider their possible language delays can lead to the often-
erroneous hypothesis of an intellectual deficit or of an insufficiently supportive family
context. 2) On the other hand, one must be careful not to systematically conclude that children with MMHL are necessarily classifiable as children with language difficulties due to their hearing loss, even if it must be taken into account that those children that are able to perform at the same level of their classmates are generally achieving these results due to an increased effort. As Bourland-Hicks & Tharpe (2002) have illustrated, children with MMHL make considerably more effort than normally-hearing counterparts during tests involving listening, not only in the presence of background noise but also when there is no such noise, a fact which may partially explain their academic challenges.
The most appropriate approach for children with MMHL clearly remains the early implementation of educational and speech intervention, tailored to their needs. The aim for therapists is then to identify those children whose schooling and speech are truly affected by their hearing loss, and to provide intervention that will allow them to compensate for their handicap as early as possible. What specific compensations can be implemented? In addition to an appropriate hearing aid, emphasis should be placed on audio-visual perception, through the development of speech-reading skills, so that children can benefit from additional visual information. Moreover listening skills also need to be refined so as to improve the perception and discrimination of noises, and in turn of phonemes, syllables and words. In addition, given
that a majority of this population faces phonological and morphosyntactic difficulties, it is also advisable to work on strengthening both the auditory-phonation loop as well as sentential structure. Younger children in greater difficulty may also benefit from the symbolizing of words and their grammatical function (for example ‘subject’, ‘verb’, ‘complement’…) so as to allow them to gain a better grasp of these notions. Finally, the training of metaphonological skills should further allows these children to become more conscious of the units of oral language, which in turn better prepares them for acquiring the art of writing. In this chapter, we have seen as well that phonological and morphosyntactic skills do not spontaneously attain a normal level of oral language in older children with MMHL. It therefore seems unrealistic to target this goal for adolescents. Instead, it appears better to train these individuals in avoidance and compensatory strategies. With respect to syntax, for example, it is much less stigmatizing for an adolescent to produce simpler yet error-free sentences, rather than to attempt complex speeches that are possibly more prone to error.
It is particularly essential to take the above considerations into account when setting up pedagogical support for children with MMHL. The impact of hearing loss on education can be significant: illiteracy rates among young French adults with this condition are found to be three times higher than those attested amongst the normally hearing population (Bentolila, 1996). Davis et al. (1986) have also evaluated the academic performance of children and adolescents with MMHL and have illustrated their difficulty on scholastic tests of reading and mathematics. As such, they conclude that MMHL lowers one’s chance of academic
achievement. Finally, in our own work (Delage & Tuller, 2007), we have underlined that among 19 high school students with MMHL aged 11 to 16, the proportion of children experiencing academic failure (repeating two grades) is as high as 21% while the national level in France is otherwise 8,2% at age 14. In closing, we would therefore like to draw the
attention of professionals working with children with MMHL to the potentially non-negligible impact this condition may have both on linguistic and academic development.
References
Alegria, J. et al., 1999. Surdité. In: Rondal, J.A., Séron, X. (Eds.), Troubles du langage, bases théoriques, diagnostic et rééducation. Mardaga, Bruxelles, pp. 551-587.
Borg, E. et al., 2007. Speech and language development in a population of Swedish hearing- impaired pre-school children, a cross-sectional study. International Journal of Pediatric Otorhinolaryngology, 71, 1061-1077.
Bortfeld, H. & Whiteburst, G.J., 2001. Sensitive period in first language acquisition. In:
Bailey, D.B. et al. (Eds.), Critical thinkings about critical periods. Paul H. Brookes Publishing Co, Baltimore, pp. 173-193.
Bourland-Hicks, C. & Tharpe, A.M., 2002. Listening effort and fatigue in school-age children with and without hearing loss. Journal of Speech, language, and Hearing Research, 45, 573- 584.
Briscoe, J., Bishop, D.V.M. & Norbury, C.F., 2001. Phonological processing, language, and literacy: A comparison of children with mild-to-moderate sensorineural hearing loss and those with specific language impairment. Journal of Child Psychology and Psychiatry, 42, 329-340.
Burkholder, R.A. & Pisoni, D.B., 2003. Speech timing and working memory in profoundly deaf children after cochlear implantation. Journal of Experimental Child Psychology, 85, 63- 88.
Davis, J.M., Elfenbein, J., Schum, R.& Bentler, R., 1986. Effects of mild and moderate hearing impairment on language, educational, and psychosocial behavior of children. Journal of Speech, Language, and Hearing Research, 51, 53-62.
Delage, H., 2008. Evolution de l’hétérogénéité linguistique chez les enfants sourds légers et moyens : Etude de la complexité syntaxique. Doctoral thesis, François-Rabelais University, Tours.
Delage, H. & Tuller, L., 2007. Language Development and Mild-to-Moderate Hearing Loss:
Does Language Normalize with Age? Journal of speech and language and hearing research, 50, 1-14.
Delage, H. & Tuller, L., 2010. Evolution of Syntactic Complexity and Avoidance Strategies in Children and adolescents with Mild-to-Moderate Hearing Loss. In: Costa, J. et al. (Eds.), Language Acquisition and Development: Proceedings of GALA 2009. Cambridge Scholars Publishing.
Fortnum, H.M., Marshall, D.H. & Summerfield, A.Q., 2002. Epidemiology of the UK population of hearing-impaired children, including characteristics of those with and without cochlear implants-audiology, aetiology, comorbidity and affluence. Int. J. Audiol., 41, 170- 179.
Friedmann, N., & Haddad-Hanna, M. (in press). The comprehension of sentences derived by syntactic movement in Palestinian Arabic-speaking children with hearing impairment.
Applied Psycholinguistics.
Friedmann, N. & Szterman, R., 2006. Syntactic movement in orally-trained children with hearing impairment. Journal of Deaf Studies and Deaf Education, 11, 56-75.
Friedmann, N., & Szterman, R. (2011). The comprehension and production of wh questions in children with hearing impairment. Journal of Deaf Studies and Deaf Education, 16(2), 212- 235.
Gilbertson, M., Kamhi, A., 1995. Novel word learning in children with hearing impairment.
Journal of Speech and Hearing Research, 38, 630–642.
Govaerts, P.J. et al., 2002. Outcome of Cochlear Implantation at Different Ages from 0 to 6 Years. Otology & Neurotology, 23, 885-890
Hamann, L. et al., 2007. (Un)successful subordination in French-speaking children and Adolescents with SLI. In: Caunt-Nulton, H., Kulatilake, S., Woo, I. (Eds.), Proceedings of the 31st annual Boston University Conference on Language Development. Cascadilla Press, Somerville, MA, pp. 286-297.
Hansson, K. et al., 2004. Working memory and novel word learning in children with hearing impairment and children with specific language impairment. International Journal of
Language and Communication Disorders, 39, 401-422.
Hansson, K., Sahlén, B. & Mäki-Torkko, E., 2007. Can a ‘single hit’ cause limitations in language development? A comparative study of Swedish children with hearing impairment and children with specific language impairment. International Journal of Language and Communication Disorders, 42, 307-323.
Jakubowicz, C., 2005. The Language Faculty: (Ab)normal development and interface constraints. Talk presented at GALA 2005, University of Siena.
Jakubowicz, C., 2007. Grammaire universelle et trouble spécifique du langage. In: Brimond, J., Franck, J. (Eds.), Noam Chomsky. Cahiers de l’Herne, pp. 164-175.
Jakubowicz, C., 2011. Measuring derivational complexity: New evidence from typically developing and SLI learners of L1 French. Lingua, 121, 339-351.
Jakubowicz, C. et al., 1998. Determiners and Clitic Pronouns in French-Speaking Children with SLI. Language Acquisition, 7, 113-160.
Jakubowicz, C. & Strik, N., 2008. Scope-marking strategies in the acquisition of long distance wh-questions in French and Dutch. Language and Speech, 51, 101-132.
Jakubowicz, C. & Tuller, L., 2008. Specific Language Impairment in French. In: Ayoun, D.
(Ed.). Benjamins, Amsterdam.
Johnson, E.L., Newport, S.J., 1989. Critical period effects in second language learning: the influence of maturational state on the acquisition of English as a second language. Cognitive Psychology, 21, 60-99.
Johnson, E.L., Newport, S.J., 1991. Critical period effect on universal properties of language:
the status of subjacency in the acquisition of a second language. Cognition, 39, 215-258.
Khomsi, A. et al., 2007. Bilans Informatisés du Langage Oral (BILO 2 ; BILO 3C). ECPA, Paris.
McWhinney, B., 2000. The Childes project: Tools for Analysing Talk, 3rd ed. Lawrence Erlbaum, Mahwah, NJ.
Monjauze, C., 2007. Langage et épilepsie à pointes centro-temporales : déficits et séquelles.
Doctoral thesis, François-Rabelais University, Tours.
Newport, E. L., Bavelier, D. & Neville, H.J., 2002. Réflexions critiques sur les périodes critiques: le cas de l’acquisition du langage. In : Dupoux, E. (Ed.), Les langages du cerveau, Odile Jacob, Paris, pp 477-498.
Norbury, C.F., Bishop D.V.M. & Briscoe J., 2001. Production of English finite verb
morphology: A comparison of SLI and mild-moderate hearing impairment. Journal of Speech, Language, and Hearing Research, 44, 165-179.
Norbury, C.F., Bishop, D.V.M. & Briscoe, J., 2002 Does impaired grammatical
comprehension provide evidence for an innate grammar module? Applied Psycholinguistics, 23, 247-268.
Paradis, C. & Crago, M., 2003. What can SLI Tell US About Transfer in SLA? In: Liceras, J.
M. et al. (eds.) Proceeding of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002), Cascadilla Proceedings Somerville, MA, pp. 219-226
Paradis, C., 2004. The relevance of specific language impairment in understanding the role of transfer in second language acquisition. Applied Psycholinguistics, 25, 67-82.
Prévost, P. et al., 2010. Computational complexity effects in the acquisition of wh-questions in child L2 French. In: Dominguez L., Guijarres-Fuentes, P. (Eds.), New directions in
language acquisition: Romance languages in the generative perspective, Cambridge Scholar Publisher, Cambridge, pp. 415-443.
Raven, J.C., Court, J.H. & Raven, J., 1998. Progressives Matrices Couleur. EAP, Paris.
Russ, S. et al., 2003. Epidemiology of congenital hearing loss in Victoria, Australia. Int J Audiol., 42, 385-390.
Sahlén, B. & Hansson, K., 2006. Novel word learning and its relation to cognitive and
linguistic skills. A comparison between children with SLI and children with mild-to-moderate hearing impairment. Journal of Multilingual Communication Disorders, 4, 95-107.
Sharma, A., Dorman, M. & Kral, A., 2005. The influence of a sensitive period on central auditory development in children with unilateral and bilateral cochlear implants. Hearing Research, 203, 134-143.
Sharma, A., Dorman, M. & Spahr, A.J., 2002. A sensitive period for the development of the central auditory system in children with cochlear implants: implications for age of
implantation. Ear Hear, 23, 532-539.
Soares, C., 2006. La syntaxe de la périphérie gauche en portugais européen et son acquisition.
Doctoral dissertation, University of Paris VIII.
Tomblin, J.B. et al., 1997. Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language, and Hearing Research, 40, 1245-1260.
Tuller, L. & Jakubowicz, C., 2004. Développement de la morphosyntaxe du français chez des enfants sourds moyens. Le Langage et l’Homme : Logopédie, Psychologie, Audiologie, 14, 191-207.
Tuller, L. et al., 2011. Clitic pronoun production as a measure of atypical language
development in French: A comparative study of SLI, mild-to-moderate deafness and benign epilepsy of childhood with centrotemporal spikes. Lingua, 121, 423-441.
van der Lely, H. K.J., Jones, M. & Marshall, C.R., 2011. Who did Buzz see someone?
Grammaticality judgement of wh-questions in typically developing children and children with Grammatical-SLI. Lingua, 121, 423-441.
Verbist, A., 2010. The acquisition of personal pronouns in cochlear-implanted children.
Doctoral thesis.
Vohr et al, 2012. Language outcomes and service provision of preschool children with congenital hearing loss, Early Human Development, doi:10.1016/j.earlhumdev.2011.12.007.
