Semantic and syntactic reading comprehension strategies used by deaf children with early and late cochlear implantation

Research in Developmental Disabilities 49–50 (2016) 153–170

Contents lists available at ScienceDirect

Research in Developmental Disabilities

Semantic and syntactic reading comprehension strategies used by deaf children with early and late cochlear implantation Carlos Gallego a,*, Ma Teresa Martı´n-Aragoneses b,c, Ramo´n Lo´pez-Higes a, Guzma´n Piso´n a a b c

Complutense University of Madrid (UCM), Spain National Distance Education University (UNED), Spain Laboratory of Cognitive and Computational Neuroscience, Centre for Biomedical Technology (CTB), Spain

A R T I C L E I N F O

A B S T R A C T

Article history: Received 25 March 2015 Received in revised form 12 November 2015 Accepted 19 November 2015 Available online 17 December 2015

Deaf students have traditionally exhibited reading comprehension difficulties. In recent years, these comprehension problems have been partially offset through cochlear implantation (CI), and the subsequent improvement in spoken language skills. However, the use of cochlear implants has not managed to fully bridge the gap in language and reading between normally hearing (NH) and deaf children, as its efficacy depends on variables such as the age at implant. This study compared the reading comprehension of sentences in 19 children who received a cochlear implant before 24 months of age (earlyCI) and 19 who received it after 24 months (late-CI) with a control group of 19 NH children. The task involved completing sentences in which the last word had been omitted. To complete each sentence children had to choose a word from among several alternatives that included one syntactic and two semantic foils in addition to the target word. The results showed that deaf children with late-CI performed this task significantly worse than NH children, while those with early-CI exhibited no significant differences with NH children, except under more demanding processing conditions (long sentences with infrequent target words). Further, the error analysis revealed a preference of deaf students with early-CI for selecting the syntactic foil over a semantic one, which suggests that they draw upon syntactic cues during sentence processing in the same way as NH children do. In contrast, deaf children with late-CI do not appear to use a syntactic strategy, but neither a semantic strategy based on the use of key words, as the literature suggests. Rather, the numerous errors of both kinds that the late-CI group made seem to indicate an inconsistent and erratic response when faced with a lack of comprehension. These findings are discussed in relation to differences in receptive vocabulary and short-term memory and their implications for sentence reading comprehension. ß 2015 Elsevier Ltd. All rights reserved.

Keywords: Reading comprehension Syntactic and semantic strategies Key word strategy Cochlear implant Age of implantation

* Corresponding author at: Facultad de Psicologı´a, Universidad Complutense de Madrid, Campus de Somosaguas, Crtra. De Hu´mera s/n, 28223 Madrid, Spain. E-mail addresses: [email protected] (C. Gallego), [email protected] (M.T. Martı´n-Aragoneses), [email protected] (R. Lo´pez-Higes), [email protected] (G. Piso´n). http://dx.doi.org/10.1016/j.ridd.2015.11.020 0891-4222/ß 2015 Elsevier Ltd. All rights reserved.

154

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

1. Introduction Deaf people often exhibit problems in learning to read as a result of their difficulties in developing spoken language (Conrad, 1979; Domı´nguez, 2003; Perfetti & Sandak, 2000). The cochlear implant has changed this situation considerably in recent years, as it pertains to spoken language and, by extension, to reading. The benefits of cochlear implantation (CI) to the spoken language skills of children with prelingual profound deafness are well established (Boons et al., 2012, 2013a, 2013b; Sparreboom, Langereis, Snik, & Mylanus, 2015). Similarly, and probably as a consequence, the use of cochlear implants has increased the reading potential of deaf students. However, this does not seem comparable yet to that of normally hearing (NH) children (see, for example, Archbold et al., 2008; Connor & Zwolan, 2004; Domı´nguez, Pe´rez, & Alegrı´a, 2012; Geers, 2003, 2004; Geers, Tobey, Moog, & Brenner, 2008; Johnson & Goswami, 2010; Lo´pez-Higes, Gallego, Martı´n-Aragoneses, & Melle, 2015; Marschark, Rhoten, & Fabich, 2007; Marschark, Sarchet, Rothen, & Zupan, 2010; Nicholas & Geers, 2008; Spencer, Gantz, & Knutson, 2004). The age at implantation seems to be one of the factors that determine the extent to which implanted children will benefit (Miyamoto, Hay-McCutcheon, Kirk, Houston, & Bergeson-Dana, 2008). There seems to be a sensitive period in early postnatal life, particularly before age three, during which the brain is highly efficient in establishing connections between speech auditory input and the development of linguistic skills, such as lexical and grammatical knowledge (Kuhl, Conboy, Padden, Nelson, & Early, 2005; Markman et al., 2011). In turn, the higher the sound deprivation in these first years of life, the greater the negative impact on the maturation of auditory pathways, nuclei, and centers (Sainz & de la Torre, 2005). The benefit derived from wearing hearing aids is limited in the severe to profound hearing loss range. In fact, when hearing loss is profound, the expected hearing level in prelingually deaf individuals using conventional hearing aids is residual effective, which means that the child is not able to discriminate a verbal message through auditory processing only, and that comprehension using lip-reading improves not more than 50% with auditory support (Jua´rez, Monfort, & Monfort, 2005). The cochlear implant allows these children to access auditory information similar to that received by a child with moderate hearing loss who wears hearing aids (Spencer & Marschark, 2010). However, the exact nature of auditory stimulation that is received through cochlear implants is difficult to predict or describe because it depends on various factors, such as the quality of pre-implant auditory representations, or how brain plasticity contributes to transforming these representations (Sharma, Dorman, & Kral, 2005). For example, it has been found that cortical responses to auditory stimulation in children implanted before the age of three and a half are similar to those observed in children with normal hearing, though the pattern of activation seems to depend on audiological background variables (Sharma, Nash, & Dorman, 2009). Thus, an incomplete spoken language input prior to implantation, combined with the atypical auditory perception following cochlear implant placement, can affect language development not only at the phonological level, but also at lexical-semantic, morpho-syntactic and pragmatic levels. Among other effects, the ability to establish relevant phonological contrasts should be significantly impeded as a consequence of inadequate speech input. This could make it difficult to distinguish between similar words, such as different verb tenses, or nouns with or without a plural marker (Johnson & Goswami, 2010). In fact, it has been proposed that the specific developmental sequence for grammatical skills in children with cochlear implants is determined by the perceptual prominence of the relevant acoustic markers (Svirsky, Stallings, Lento, Ying, & Leonard, 2002). Obviously, this atypical acquisition of morphophonological representations at the auditory comprehension level will consequently be reflected in reading comprehension. Furthermore, as suggested for other romance languages such as Italian (Caselli, Rinaldi, Varuzza, Giuliani, & Burdo, 2012), the language difficulties observed in Englishspeaking implanted deaf children could be increased for children who learn Spanish as their native language, given the complexity of its morphology, and also its phonetic and prosodic features. In a previous study, which is part of a broader research as the present work, Lo´pez-Higes and coworkers (2015) just found that Spanish-speaking deaf children with prelingual severe to profound hearing loss who had been implanted from age 2 to 5 years differed significantly from their normally hearing peers, even from children implanted before 24 months with the same age and educational level, in both nominal and verbal morphology indices of a written morphological awareness test. In addition, verbal inflectional morphology was the most important factor in distinguishing between late implanted children and the other two groups of children in a consistent manner. Moreover, the lack of an early auditory and articulatory experience seems to have negative effects on other cognitive and linguistic factors that could also affect reading comprehension in children with cochlear implants, specifically vocabulary knowledge and working memory processes (Connor & Zwolan, 2004). Obviously, there is a strong association between vocabulary knowledge and reading comprehension. Hearing loss negatively impacts children’s vocabulary development. Deaf children tend to have slower rates of word acquisition and smaller lexicons (Prezbindowski & Lederberg, 2003). Regarding working memory, there is abundant evidence of its involvement in processes related to reading ability, and its influence in the morpho-syntactic comprehension of children with cochlear implants has been proven (Asker-A´rnason, Wass, Gustafsson, & Sahle´n, 2015; Asker-A´rnason, Wass, Ibertsson, Lyxell, & Sahle´n, 2007; Lo´pez-Higes et al., 2015). Some studies have also shown a direct relationship between working memory span and vocabulary size in implanted deaf children (e.g., Harris et al., 2013). In general, it has been noticed that children with cochlear implants have shorter spans than their normally hearing peers (Burkholder & Pisoni, 2003, 2004; Cleary, Pisoni, & Geers, 2001; Harris et al., 2013). According to Burkholder and Pisoni (2003, 2004), this could be due to less efficient short-term memory processes, in particular to those associated with verbal rehearsal and serial scanning of information. In a recent study, Arfe, Rossi, and Sicoli (2015) found that children without hearing impairment exhibited higher scores than deaf children on both reading comprehension and

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

155

forward digit span, but not on reading span, a traditional measure of working memory span. This finding was interpreted as further evidence of a larger gap between hearing and deaf children in verbal rehearsal than in executive working memory processes. But, without doubt, the use of cochlear implants under certain conditions facilitates the development of inner speech and verbal rehearsal (Dillon & Pisoni, 2004). The poor reading performance of deaf individuals has often been also attributed to deficits in the syntactic processing of sentences, leading researchers to question whether deaf children use strategies based on syntactic information to understand sentences, like their NH counterparts do, or whether they develop qualitatively different strategies (Domı´nguez et al., 2012; Miller et al., 2012). The ability to correctly assign thematic roles to the components of a sentence in an effort to extract its meaning (who did what to whom) is crucial in reading comprehension. This ability depends on semantic factors or characteristics, such as the animate-inanimate trait of the subject or object, but also on syntactic factors, like the word order, the hierarchical relationship between them and morphological markers (Ferna´ndez & Anula, 2002). The role that syntactic structure plays in sentence comprehension is especially evident in those statements in which the assignment of thematic roles is not restricted by semantic or pragmatic factors (e.g., semantically reversible sentences, such as The boy kisses the girl). In these cases, understanding the sentence necessarily requires integrating the meaning of the words into a structure that defines the relationship between them. In order to understand the sentence, the reader must consider the syntactic cues present in it, such as the order of the words, the functional words, the concordances, the morphological markers, and so on. The reader must also understand what the words mean. Thus, the integration will rely on both bottom-up as well as topdown processes (Perfetti, Landi, & Oakhill, 2005). Most research on reading in prelingually deaf individuals suggests that they are unable to syntactically process sentences. For example, Miller (2000, 2005, 2006, 2010) examined how well non-implanted deaf readers with hearing aids understood semantically plausible, non-plausible and neutral sentences. His findings indicate that with plausible sentences, most deaf readers attain a good level of comprehension, similar to that of NH individuals, while it drops to chance level or even below when the sentence is implausible (e.g., The truck driver who ran over the pedestrian was wounded) or semantically neutral (e.g., The woman who watched the girl was smiling). These results can also be extended to hard of hearing readers with hearing losses lower than 85 dB in the better ear (Miller, 2005). In light of these findings, this author proposes a reading strategy based on the top-down processing of content words as the source of deaf and hard of hearing individuals’ difficulties understanding neutral and implausible sentences. In such statements, the message being conveyed is not supported by, or can even contradict, the reader’s experience and knowledge of the world, and their syntactic bottom-up processing is essential for proper comprehension. However, confronting the content words from a plausible sentence with prior knowledge may be enough to understand or infer its meaning, which makes the processing of its syntactic structure optional. In a more recent study, Miller et al. (2012) examined the factors that explain the variance in the reading comprehension skills of non-implanted prelingually deaf students. The main purpose was to determine what differentiates skilled from nonproficient deaf readers. The study participants, 213 children in grades six through ten, were recruited from four different language backgrounds (Hebrew, Arabic, English and German) and assigned to three distinct reader profiles (syntactic, semantic or unspecified readers). As in previous studies, a sentence comprehension test that manipulated the semantic sentence’s plausibility was used. In addition, a word processing experiment was performed to explore both phonological and semantic processing skills. This experiment required the participants to determine as quickly as possible whether two stimuli (i.e., two words or a real word and a pseudohomophone) were semantically related. Findings suggest both inefficient syntactic knowledge and deficient knowledge structures (a lack of general knowledge and a lack of well-structured domainspecific knowledge), rather than differences at the phonological level. Various studies have demonstrated the positive effects of CI on reading comprehension, especially after using a cochlear implant for an extended period of time (Asker-A´rnason et al., 2015; Connor, 2006; Lo´pez-Higes et al., 2015). However, the studies are not absolutely conclusive about its benefit. According to review of empirical studies about implanted children’s reading achievement by Marschark and coworkers (2007), the lack of consistent findings might be due to the influence of factors such as age of implantation. Considering this variable and its associated factors, we explored morphosyntactic reading comprehension with the same sample used in this study (Lo´pez-Higes et al., 2015). In that study, we found that late implanted children (i.e., after 24 months) made significantly more errors on a sentence-picture verification task than early implanted or NH children when the foil was syntactic – i.e., in those items where the picture represented an action in which the thematic roles of the arguments had been reversed with respect to the sentence. This indicates that late implanted children have problems assigning thematic roles to constituents in semantically reversible sentences, or that they do not use effective syntactic strategies. Early implanted children, however, performed as well as their control group peers on a grammatical comprehension test, obtaining an accuracy level in syntactic foils that indicates they are able to accurately detect changes based on the thematic role assignment. Domı´nguez and coworkers carried out a study with 71 prelingually and profoundly deaf children aged 6–16 years, and a control group of 326 NH children (Domı´nguez et al., 2012; Soriano, Pe´rez, & Domı´nguez, 2006). In the group of deaf children, 53.5% were implanted at an average age of 3.74 years. Fifty-five percent of implanted children had received the implant early in their development, and the remaining 45% after the age of three years. The children were given a reading efficiency test (Marı´n & Carrillo, 1999) and another test to detect the use of syntactic strategies (Soriano et al., 2006), as well as an orthographic decision task and a test to assess metaphonological skills. Although the results obtained by deaf children with cochlear implants were better than those of non-implanted deaf children, and their progression was closer to the norm, they still had difficulties and made significantly more errors in sentence reading comprehension tasks than NH children did. From

156

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

this, and especially from the results of the test to evaluate syntactic strategies, where deaf children with cochlear implants exhibited greater competence than those without them, the authors concluded that wearing a cochlear implant improves the deaf students’ use of syntactic cues. In contrast, deaf children without cochlear implants would only use a semantic strategy based on content words to understand sentences. The results also showed differences in the group of implanted children depending on the age of implantation. Children who received a cochlear implant before the age of three years exhibited greater skills than those who had received it later in their development. It has been proposed that deaf individuals use a semantic strategy in sentence comprehension that consists of identifying content words and deriving a complete representation of the sentence by ignoring any other information in the sentence, and thus the morphosyntactic relationships between their words. This assumption has been empirically supported by data obtained in studies that examined the sentence reading mechanisms used by deaf adults with secondary or even higher education who, with the exception of one participant, had been taught in special schools for the deaf using pure oral communication methods, although all of them used Spanish Sign Language in their daily lives (Domı´nguez & Alegrı´a, 2010; Domı´nguez, Carrillo, Pe´rez, & Alegrı´a, 2014). This reading mechanism or procedure is called the key word strategy. The hypothesis is that the basic mechanism employed by many deaf adults to understand written sentences involves identifying the most frequent content words, ignoring low-frequency and function words, and using the former to derive a representation for the sentence’s meaning. This strategy is used to a greater extent by deaf than NH individuals at the same reading level (Domı´nguez & Alegrı´a, 2010). The tendency to use the key word strategy is also found in implanted deaf children, though to a lesser degree than in deaf children without cochlear implants (Domı´nguez et al., 2012). The thesis of the key word strategy arose from data derived from the Semantic strategies detection test (Alegrı´a, Domı´nguez, & van der Straten, 2009; Domı´nguez & Alegrı´a, 2010; Domı´nguez et al., 2014, 2012; Domı´nguez & Soriano, 2009; Soriano et al., 2006). The test features a list of sentences of increasing complexity, in which the last word is omitted. Each sentence must be completed using one of four alternatives presented to the reader, among which only one yields a plausible interpretation for the sentence. The other three are words that are semantically related to some of the content words in the sentence or to the context (semantic foils). The rationale of the test is that selecting the proper alternative (target word) requires the use of syntactic and semantic cues. If the reader selects a wrong alternative, this indicates that the sentence is being processed superficially because the reader is guided solely by global semantic cues related to the meaning of some word in the sentence. Errors are thus interpreted as a preference for using semantic strategies as opposed to an accurate analysis of the syntactic relationships among the content words. However, the test has a drawback in that the alternatives to the correct answer are all semantic foils. It is therefore impossible to determine with certainty if there is a preference for using a strategy based on key words, because when the reader does not understand the sentence and makes an error, she/he must forcibly choose a semantic foil, since there is no other alternative among the options. Thus, selecting the wrong alternative can only indicate a lack of comprehension, but not a preference for one strategy or another. In light of the above, the aim of this study was to assess the sentence reading comprehension skills of implanted deaf children with prelingual severe to profound hearing loss in order to verify: (a) whether they rely on the same strategies based on syntactic cues used by their NH peers, and (b) whether the use of syntactic cues is determined by implantation age and associated factors. Certainly, clarifying these issues will help establish best practices for the improvement of reading skills in implanted deaf children. Based on existing evidence, it is expected that children with early-CI will use syntactic cues, as their NH peers do, while children with late-CI will exhibit a profile similar to that of deaf children, and will thus rely primarily on a sentence comprehension strategy that will presumably be based on content key words, instead of a syntactically guided strategy. To test this hypothesis we designed a sentence completion test analogous to that referenced earlier (Domı´nguez & Alegrı´a, 2010; Domı´nguez et al., 2014, 2012; Soriano et al., 2006), but in which the nature of the possible alternatives has been modified to allow us to infer the strategies employed by analyzing the types of errors made. 2. Method 2.1. Participants Fifty-seven primary school children between 8 and 12 years of age were selected for this study from public and private schools in Madrid and Castilla La Mancha (two autonomous communities of Spain). Of the participants, 38 had prelingual deafness and had received a cochlear implant, 19 of them before the age of 24 months (early-CI), and 19 between the ages of 24 months and 5 years (late-CI). The remaining 19 were normally hearing (NH). All of the children with cochlear implants had a history of bilateral severe or profound deafness (loss in excess of 70 dB) and had been diagnosed before the age of 24 months. The causes of hearing loss were viral diseases (5.3% in both CI groups), genetic disorders (57.9% in the early-CI group and 42.1% in the late-CI group) and undetermined causes (36.8% in the early-CI group and 52.6% in the late-CI group). No significant differences between the groups of children with cochlear implants were noted with respect to the etiology of hearing loss (x2 (2) = 1.00, p = .606). The implanted children were recruited by professionals from the institutions cooperating in this research (schools, children’s hospitals, associations for the deaf) according to the above criteria. Children with a low IQ (<85), attention deficit disorder or learning difficulties were excluded from this study, as were those with low birth weight in order to prevent any possible adverse effects associated with potential developmental delays or brain dysfunctions. Also excluded were children with cochlear malformations. The participants did not have any other kind of sensory deficit or neurological condition that could affect their ability to read or understand instructions. Thus, three groups

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

157

Table 1 Means (and standard deviations) of chronological age in months, number of males and females by group, and means (and standard deviations) of scores on receptive vocabulary, short-term memory and nonword reading. Group

Na

NH early-CI late-CI

19 19 19

Age

Gender

M (SD)

M

116.11 (14.75) 116.32 (13.54) 118.58 (14.07)

6 12 11

Vocabulary

FDS

NWR

F

M (SD)

M (SD)

M (SD)

13 7 8

130.32 (19.76) 98.95 (20.67) 85.68 (24.91)

5 (.75) 4.95 (.97) 4.32 (1.25)

35.7 (1.84) 33.6 (3.48) 28.2 (9.48)

Note: Vocabulary = PPVT-III equivalent age (in months); FDS = WISC-IV Forward Digit Span; NWR = Nonword Reading subtest of PROLEC-R. a For NWR test, n = 13 in the NH group, n = 18 in the early-CI group, and n = 17 in the late-CI group.

of 19 participants each were formed: one with children with early-CI, another with children with late-CI, and a control group with NH children. In order to ensure the equivalence of the groups, their participants were matched on nonverbal intelligence as measured by the WISC-IV Perceptual Reasoning Index, so that the score on this index did not differ between groups (x2 (2) = 1.95, p = .378). Likewise, there were no significant differences between groups in age (x2 (2) = .49, p = .781) and gender distribution (x2 (2) = 4.35, p = .113). Table 1 shows demographic information, as well as some cognitive and linguistic characteristic of the participants. Based on the reading accuracy for nonwords, the groups also did not differ significantly in their decoding ability (x2 (2) = 5.72, p = .057), but there were significant differences in their receptive vocabulary (x2 (2) = 25.25, p < .001) and short-term memory span (x2 (2) = 7.17, p = .028). Specifically, the receptive vocabulary level of children with late-CI was considerably lower than that of their control group peers (U = 32.50, Z = 4.32, p < .001, Cliff’s d = .82), and even than that shown by children with early-CI (U = 103.50, Z = 2.25, p = .024, Cliff’s d = .43), who in turn differed from NH children as well (U = 48.50, Z = 3.86, p < .001, Cliff’s d = .73); while the differences in short-term memory span were solely due to a significantly shorter span in the late-CI group with respect to the NH group (U = 95.00, Z = 2.53, p = .012, Cliff’s d = .47). Additional information on hearing and communication in the groups of children with cochlear implants is provided in Table 2, which refers to current hearing aid status, age in months of the first implantation and the first hearing device, and communication modes used in family and school settings according to the age of implantation. As can be observed, most implanted children received hearing aids prior to implantation, and only 21% of children with early-CI and 31.6% of children with late-CI did not initially use another hearing device. In addition, there were no statistically significant differences between the two groups of implanted children in terms of the hearing aids they currently used (x2 (2) = 2.22, p = .329). Although the percentage of children using sign language was higher in the late-CI group than in the early-CI group, all known cases of signers in the sample used spoken language in combination with sign language. Moreover, it should be noted that spoken language was the dominant mode of communication for both groups, especially in the family setting. 2.2. Material The tests described here were part of a broader assessment protocol used to study reading comprehension in school-age children ranging from third to sixth grade (see Lo´pez-Higes et al., 2015). This study focuses on a particular test based specifically on the Semantic strategies detection test created by Domı´nguez et al. (Domı´nguez & Alegrı´a, 2010; Domı´nguez et al., 2014, 2012; Soriano et al., 2006). In addition, other study subjects’ scores obtained from the application of several standardized tests are also used. Specifically, we report here the following: (a) the WISC-IV Perceptual Reasoning Index, a composite score composed of three core subtests: Block Design, Picture Concepts and Matrix Reasoning (Wechsler, 2004); Table 2 Current hearing aid status, mean (and standard deviation) age in months of first implantation and first hearing device, and mode of communication in family and school settings by group. CI group late-CI

early-CI Current hearing aid status (n) Bilateral CI Bimodal (CI plus hearing aid) Unilateral CI Age of first implantation (M (SD), in months) Age of first hearing device (M (SD), in months) Mode of communication (%) Spoken language Spoken & Sign language Unknown

9 4 6

4 4 11

14.68 (5.69) 11.05 (6.87)

41.89 (12.97) 30.42 (16.63)

Family 78.9

School 89.5

21.1

10.5

Family 73.7 10.5 15.8

School 57.9 36.8 5.3

158

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

(b) the WISC-IV Forward Digit Span (FDS) scaled score (Wechsler, 2004); (c) the equivalent age score on the Peabody Picture Vocabulary test (PPVT-III; Spanish adaptation by Dunn, Dunn, & Arribas, 2006); and (d) the raw scores on the Grammatical Structures (GS) subtest and the Nonword Reading (NWR) subtest, both of them from the PROLEC-R test (Cuetos, Rodrı´guez, Ruano, & Arribas, 2007), a Spanish standardized test that evaluates the processes involved in reading using accuracy and speed as measures of performance in different tasks. The NWR subtest contains 40 elements formed by replacing one or two letters in the words employed for the Word Reading subtest. The GS subtest uses a picture-sentence matching task and includes 16 items, each consisting of one sentence (active, passive, with a relative clause or the focalized complement) and four pictures (one corresponding to the sentence meaning and three others that are given as foils). The scores on the GS and NWR subtests are included as indicators of sentence comprehension and decoding ability, respectively. 2.2.1. Semantic and syntactic strategies detection test (DES/S) The specific DES/S test was developed ad hoc for this study from one originally designed by Soriano et al. (2006). The test consists of a multiple choice task and is composed of a set of incomplete sentences with the final word omitted. The sentence must be completed by selecting, from among four alternatives, the one that fits morphosyntactically and coherently completes the meaning of the sentence (i.e., target word). The three remaining words are foils. To build the test used in this study, 24 of the 64 sentences from the original test were selected and freely adapted for the purposes of this research. The number of items was decreased so as to reduce the time needed to complete the test, as well as to minimize potential effects of fatigue (recall that this test was embedded within a larger assessment protocol). The sentences selected and rebuilt were either short (five or fewer words) or long (nine or more words). All of them were formulated with understandable and common words for the age range of the participants. The target words and foils varied with respect to frequency of use, employing high- and low-frequency words. The lexical frequency was determined using the frequency dictionary by Alameda and Cuetos (1995). A word was regarded as having a low lexical frequency when its frequency was lower than or equal to 50 occurrences per two million, and high when its frequency was equal to or higher than 100 occurrences per two million. Thus, the test comprises six items in each of the following categories: short sentences with a frequent target word (SF); short sentences with an infrequent target word (SI); long sentences with a frequent target word (LF); and long sentences with an infrequent target word (LI). The target words selected were adjectives (11), nouns (9) and, to a lesser extent, verbs (4). The following are examples for each item category. SF: El balo´n es de _____ (The ball is made of _____) hielo (ice) / cuero (leather) / portero (goalkeeper) / partido (match) SI: Las focas del parque son _____ (Seals in the park are _____) aros (rings) / juguetonas (playful) / honestas (honest) / entrenador (trainer) LF: El fuego se extendio´ ra´pidamente porque el campo estaba _____ (The fire spread quickly because the field was _____) feliz (happy) / incendio (fire) / llamas (flames) / seco (dry) LI: Fotocopiaron el libro porque no tenı´amos suficiente dinero para _____ (They photocopied the book because we did not have enough money _____) marcharlo ([to] go out) / estudiaron ([they] studied) / leı´an ([they] were reading) / adquirirlo ([to] buy it) However, the most important variation was the introduction of a syntactic foil among the incorrect alternatives. A syntactic foil is a word that fits morphosyntactically in a sentence but that is incongruent or semantically unrelated. Thus, instead of three words semantically related with the sentence as alternatives to the target word, we only used two semantic foils and a third word that fitted morphosyntactically but was semantically incongruent, namely a syntactic foil. For example, the sentence Aquella travesura le hizo ______ (That prank made him _______) offered the alternatives: enfadarse (get angry) as the target word, castigo (punishment) and nene (child) as the semantic foils, and peinarse (comb his hair) as the syntactic foil. A pilot version of the test was submitted to the judgment of nine experts, who reviewed the quality of the items. Once the recommended improvements were implemented, the final list of stimuli was created consisting of 24 items and three trials of practice at the beginning. The items were arranged at random in the list so that fatigue and learning effects did not impact the longer and more complex items, and thus the order of presentation could not be misinterpreted as sentence length or word frequency effects. All 24 items are listed in the Appendix. 2.3. Procedure The various tests used in this study were administered in the context of a more general assessment over the course of different sessions and in random order. A quiet room shielded from any potential interruptions was used to test the participants individually as a means of keeping the children focused on the tasks. All of the children completed the tests

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

159

during the planned sessions. They exhibited a positive and interested attitude toward the assessment and they cooperated at all times during the conduct of the tests. The assessment protocol was administered by speech and language therapists or psychologists with extensive clinical experience treating children, who had been trained to assist in applying it. The standardized tests were administered and scored in accordance with the instructions provided by the test publisher. For the application of the DES/S test, the children received a booklet with the three sample sentences, followed by the list with 24 numbered sentences. After each sentence, its four alternatives were shown along with a square box to be used to select the answer choice (see Appendix). Only one alternative could be selected. The target word was placed randomly among the alternatives. The participants, who took the test in silence, had to read the sentences and mark the chosen answer. The instructions were given to all of them orally. Before starting the test, three practice sentences were provided to confirm that participants understood the instructions. No feedback was given to the participants during the test concerning their performance. They were told there was a time limit, though they were all allowed to complete the test. The time spent by each participant to complete the test was recorded. The examiners also annotated any pertinent observations on the record sheet. 3. Results The SPSS 22.0 software package was used for the statistical analysis with a value of p < .05 as the significance cut-off. For the DES/S test, the number and proportion of correct responses as well as of errors when selecting an answer to complete the sentence were obtained. Due to the size of the groups in the study, and to the violation of normality and homogeneity of variance assumptions, the data were analyzed using nonparametric statistics. Kruskal–Wallis tests followed by pairwise Mann–Whitney U post hoc tests were used for between-group comparisons. Friedman tests were also carried out to establish within-group differences, and Wilcoxon signed-ranks tests were used for pairwise post hoc comparisons. Bonferroni corrections were applied to adjust the p-values for multiple post hoc comparisons. Cliff’s Delta (d) statistic (in absolute value) was chosen as the effect size estimator because it is more appropriate when the homogeneity of variance or normality assumptions are violated. Cronbach’s alpha was estimated to determinate the reliability level of the DES/S test. In order to check its validity, the Spearman’s rho correlation between the score on the GS subtest of PROLEC-R and that obtained in the DES/S test was calculated. Nonparametric correlations were also computed in order to assess the relationship between performance on the DES/S test and decoding ability, receptive vocabulary and short-term memory span in each group. 3.1. Psychometric analysis of the DES/S test The specific DES/S test showed a high degree of reliability (a = .92), as well as a mean difficulty index adjusted for chance guessing of .70, a value that could be considered optimal because it indicates that the test was neither too easy nor too difficult for the sample as a whole. The scores on the DES/S test correlated significantly with the scores on the GS subtest (rho = .79, p < .001). The close relationship between these measures can be seen in the dispersion diagram (see Fig. 1). 3.2. Descriptive As shown in Table 3, the performance of the early-CI group on the DES/S test was between those exhibited by the NH and late-CI groups, being closer to the former than the latter, as reflected by the absolute distances between the means, which are much greater among both CI groups (6.21) than among the early-CI group and the NH group (2.54). Based on the range of scores for each group, the children in the NH group achieved the highest scores (from 20 to 24 correct responses) on the DES/S test. With respect to the CI groups, the DES/S scores of children with early-CI ranged from medium to high (11–24), while those of children with late-CI were distributed throughout the spectrum of scores (4–23). The mean percentage of errors made by children in the NH group (6.79%) was lower than that observed in the early-CI group (17.33%), and especially than that obtained by children with late-CI (43.21%), who failed to respond to almost half of the items. 3.3. Between-group comparisons A Kruskal–Wallis test revealed that there were significant differences in the number of errors made by the groups (x2 (2) = 20.52, p < .001). Pairwise post hoc comparisons showed that these differences were mainly due to a significantly poorer performance on the DES/S test by children with late-CI relative to children in the NH group (U = 40.00, Z = 4.13, p < .001, Cliff’s d = .78) and to children with early-CI (U = 87.00, Z = 2.73, p = .006, Cliff’s d = .52). But the groups also differed, although to a lesser extent, because the early-CI group scored significantly lower on the test than the NH group (U = 94.00, Z = 2.56, p = .011, Cliff’s d = .45). Statistically significant differences between groups were also found with regard to the time taken to perform the DES/S test (x2 (2) = 6.81, p = .033). Post hoc tests revealed that children with late-CI needed significantly more time to complete the test than children with early-CI (U = 31.00, Z = 2.52, p = .012, Cliff’s d = .73), while children in this latter group did not require longer than their control group peers to finish it (p = .354).

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

160

Fig. 1. Percentage of correct responses at the Semantic/syntactic strategies detection test (DES/S) as a function of the percentage of correct responses in the PROLEC-R Grammatical Structures subtest for the groups of participants.

3.4. Effect of item type Table 4 shows the mean scores of each group for the different item types (i.e., frequent target word & short sentence, infrequent target word & short sentence, frequent target word & long sentence, infrequent target word & long sentence). Nonparametric variance analyses for repeated measures were used to study the effects of length and frequency in each group. The results showed an item type effect only for the CI groups, which was greater in the early-CI group (x2 (3) = 17.50, p = .001) than in the late-CI group (x2 (3) = 9.85, p = .020). Post hoc tests revealed that this effect was associated with the differential influence of target word frequency in the context of long sentences for both the early-CI group (Z = 3.40, p = .001, Cliff’s d = .47) and the late-CI group (Z = 2.69, p = .007, Cliff’s d = .25). This can be seen in Fig. 2, which shows the mean percentages of correct responses by item type for each group. Table 3 Mean number of correct responses (out of 24 items) and standard deviation by group for DES/S test. NH

early-CI

late-CI

M (SD)

M (SD)

M (SD)

22.37 (1.34)

19.84 (3.63)

13.63 (6.99)

Table 4 Mean, standard deviation and range of correct responses by group and item type for the Semantic/syntactic strategies detection test (DES/S). Sentence

Short Short Long Long

Word

Frequent Infrequent Frequent Infrequent

NH

early-CI

late-CI

M

SD

Range

M

SD

Range

M

SD

Range

5.74 5.53 5.74 5.37

.56 .61 .45 .60

4–6 4–6 5–6 4–6

5.16 5.05 5.26 4.37

1.26 1.22 .93 1.12

1–6 2–6 3–6 2–6

3.42 3.26 3.89 3.05

1.90 2.00 1.88 1.84

0–6 0–6 1–6 0–6

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

161

100 90 80

Short, Frequent Short, Infrequent

70

Long, Frequent

60

Long, Infrequent

50 40 NH

early-CI

late-CI

Fig. 2. Mean percentage of correct responses by group and item type.

3.5. Differences between groups by item type, sentence length and target word frequency A series of analyses confirmed that there were significant differences between the groups in the number of errors made in each of the four item types of the DES/S test: frequent target words and short sentences (x2 (2) = 18.73, p < .001), infrequent target words and short sentences (x2 (2) = 16.67, p < .001), infrequent target words and long sentences (x2 (2) = 21.51, p < .001) and, to a less significant extent, frequent target words and long sentences (x2 (2) = 11.02, p = .004). Pairwise post hoc comparisons for each item type yielded the following results. The differences between groups in the short sentences with a frequent target word were due to the poor performance of children with late-CI, which was significantly lower than that of children in the NH group (U = 52.50, Z = 4.01, p < .001, Cliff’s d = .71), and even than that of children with early-CI (U = 84.50, Z = 2.89, p = .004, Cliff’s d = .53). Exactly the same was observed when comparing the performance of the groups on short sentences with an infrequent target word, in which the late-CI group differed significantly from both the NH group (U = 54.50, Z = 3.84, p < .001, Cliff’s d = .70) and the early-CI group (U = 82.00, Z = 2.95, p = .003, Cliff’s d = .55). However, the differences in performance on long sentences with a frequent target word only reached statistical significance when comparing the late-CI group to the NH group (U = 86.50, Z = 3.02, p = .003, Cliff’s d = .52). Finally, the differences in long sentences with an infrequent target word were due to the scores obtained by children with cochlear implants for this type of item, which were significantly lower than those of children in the NH group, both for the early-CI group (U = 81.00, Z = 3.12, p = .002, Cliff’s d = .55) and the late-CI group (U = 41.00, Z = 4.22, p < .001, Cliff’s d = .77). When only the influence of sentence length was considered, statistically significant differences between groups were observed for both short sentences (x2 (2) = 19.32, p < .001) and long sentences (x2 (2) = 18.87, p < .001). Pairwise post hoc comparisons revealed that the late-CI group differed significantly from the NH group, both in short sentences (U = 43.00, Z = 4.13, p < .001, Cliff’s d = .76) and long sentences (U = 47.00, Z = 4.00, p < .001, Cliff’s d = .74), and from the early-CI group as well, although only significantly in short sentences (U = 76.00, Z = 3.09, p = .002, Cliff’s d = .58). In turn, children in this latter group performed significantly below children in the NH group only in long sentences (U = 92.50, Z = 2.71, p = .007, Cliff’s d = .49). Statistically significant differences were also found when considering only the influence of target word frequency, with groups differing in their performance on both frequent words (x2 (2) = 14.51, p = .001) and infrequent words (x2 (2) = 21.87, p < .001). As with the results obtained for the sentence length, children with late-CI differed significantly from children in the NH group, both when the target word was frequent (U = 65.00, Z = 3.50, p < .001, Cliff’s d = .64) and infrequent (U = 36.50, Z = 4.25, p < .001, Cliff’s d = .80). However, unlike what we observed when considering length, the late-CI group differed significantly from the early-CI group for both target word frequency conditions: frequent (U = 95.00, Z = 2.54, p = .011, Cliff’s d = .47) and infrequent (U = 85.00, Z = 2.81, p = .005, Cliff’s d = .53). Again, children with early-CI differed significantly from children in the NH group in the most difficult condition: sentences with an infrequent target word (U = 92.00, Z = 2.66, p = .008, Cliff’s d = .49). 3.6. Relationship between performance on the DES/S test and decoding ability, receptive vocabulary and short-term memory span The scores on the DES/S test correlated significantly with the scores on NWR (rho = .68, p = .003), PPVT-III (rho = .67, p = .002) and FDS (rho = .74, p < .001) in the late-CI group. In the early-CI group, significant moderate correlations were also detected with the scores on NWR (rho = .55, p = .017) and PPVT-III (rho = .56, p = .013). In the contrast, no significant correlation was found between these measures and performance on the DES/S test (all ps  .190) in the NH group. In an attempt to further explore the relationship between vocabulary level and target word frequency (i.e., frequent vs infrequent), as well as between short-term memory span and sentence length (i.e., short vs long), nonparametric correlations were performed by group using these partial scores. Again, no significant association between short-term memory span and performance on the DES/S test was observed in the NH group regardless of item length (both ps  .417); however, the

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

162

PPVT-III equivalent age scores of the control group children correlated significantly with their accuracy scores on the DES/S test when the target word was frequent (rho = .55, p < .014). In the early and late implanted groups, significant correlations were found when examining the relationship between receptive vocabulary and the scores on the DES/S test regardless of target word frequency, although children with late-CI exhibited slightly more significant and higher correlations (frequent target word: rho = .60, p = .006; infrequent target word: rho = .77, p < .001) than children with early-CI (frequent target word: rho = .56, p = .012; infrequent target word: rho = .50, p = .031). Regarding the relationship between short-term memory span and sentence length, it was found that the scores on the FDS subtest were highly and significantly correlated with accuracy in items with both long sentences (rho = .73, p < .001) and short sentences (rho = .73, p < .001) in the late-CI group, but moderately and only with long sentence items (rho = .47, p = .045) in the early-CI group. 3.7. Error analysis The participants could make two types of errors when selecting the word to complete the sentences: (a) They could choose a semantic foil; that is, a word related to the semantic context of the sentence that does not fit syntactically. (b) They could choose a syntactic foil; that is, a word that fits morphosyntactically with the sentence, but which has an unrelated or opposite meaning. It should be note that some semantic foils could fit morphosyntactically if they were given a metaphorical or non-literal meaning or a different meaning from their primary meaning, which is unlikely at this age (i.e., children ranging from 8 and 12 years old). For example, for the sentence ‘‘El edificio nuevo se __________’’ (The new building was ___________), the correct option was ‘‘termino´’’ ([it was] completed), the syntactic foil was ‘‘marcho´’’ ([he/she/it] left), and the semantic foils were ‘‘viven’’ ([they] live) and ‘‘casa’’ (house or [she/he] marries). In Spanish, the latter could fit morphosyntactically with the sentence if it was interpreted as a verb instead of a noun. Despite the unlikelihood of such a situation, the data were analyzed with this possibility in mind and no influence attributable to this factor was found. These analyses considered the fact that the probability of randomly choosing a semantic foil was twice that of selecting the syntactic option. Therefore, inferential statistics was performed on the mean proportion of errors in each category (see Fig. 3). 3.7.1. Within-group differences by type of error Error type analysis by group revealed statistically significant differences in the proportion at which syntactic and semantic foils were chosen when the answer was wrong in the NH (x2 (1) = 15.00, p < .001) and early-CI (x2 (1) = 10.89, p = .001) groups, but not in the late-CI group (p = .819). Specifically, the proportion of errors made by selecting a syntactic foil was higher than that produced by choosing a semantic foil both in the NH group (Z = 3.45, p = .001, Cliff’s d = .72) and in the early-CI group (Z = 2.86, p = .004, Cliff’s d = .57). 3.7.2. Between-group differences by type of error Statistically significant differences between the groups were also observed in error rates, both when the error consisted of choosing a semantic foil (x2 (2) = 20.74, p < .001) and when it involved the selection of a syntactic foil (x2 (2) = 12.62, p = .002). Post hoc tests revealed that significance was due to differences between the three groups when the foil selected was semantic, since the NH group made a significantly lower number of semantic errors than the late-CI group (U = 48.50, Z = 4.12, p < .001, Cliff’s d = .73) and the early-CI group (U = 103.00, Z = 2.58, p = .010, Cliff’s d = .43), while this latter group also selected this foil type to a lesser extent than the late-CI group (U = 83.50, Z = 2.88, p = .004, Cliff’s d = .54). In contrast,

.16 .14 .12 .10 .08

Syntacc Foil

.06

Semanc Foil

.04 .02 .00 NH

early-CI

late-CI

Fig. 3. Mean proportions of wrong answers by group and foil type.

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

163

the differences were only significant between the NH and late-CI groups when analyzing the selection of syntactic foils (U = 62.00, Z = 3.54, p < .001, Cliff’s d = .66). 4. Discussion The aim of this study was to examine the reading comprehension skills of implanted deaf students at the sentence level in according to the age of implantation. For this purpose, two groups of early and late implanted deaf children were compared with a group of normally hearing children with the same age and educational level using a test developed ad hoc to detect the use of semantic/syntactic strategies in sentence reading comprehension. As the results repeatedly and systematically showed for every aspect studied here, the performance of the groups falls into three categories, with NH children obtaining the best results, followed by children with early-CI and then, with the lowest performance, by children with late-CI. The differences between NH and late-CI children were always significant, while those between NH and earlyCI children were not, except when the processing requirements were increased. Effectively, only in situations with high processing demands, deaf children with early-CI continued to experience difficulties similar to those of children with lateCI. Specifically, early implanted children performed worse than their NH peers in long sentences, and in particular in those in which the target word was infrequent. In contrast, children with late-CI performed below the norm in every item type, regardless of sentence length or target word frequency. However, they did not perform significantly differently than children with early-CI on long sentences, which indicates that the CI groups tend to match their performance under certain conditions. These performance profiles can be clearly seen in Fig. 2, which also shows that the three groups benefitted from information provided by a longer sentence context when the word that completed the statement was frequent. That is, the best performance on the test was seen in long sentences with a frequent target word in the CI groups, and at a level similar to that of short sentences with a frequent target word in the NH group. However, this advantage disappeared when the target word was infrequent, where the three groups exhibited their worst performance on the test; although this difference in the performance on longer sentences based on target word frequency only reached statistical significance in the CI groups. This result might be related to the lower level of vocabulary observed in both implanted groups. In fact, PPVT-III equivalent age scores correlated significantly with overall performance on the DES/S test in these groups. However, this explanation does not seem sufficient, especially for children in the early-CI group, whose performance experienced a sharp decline in the long and infrequent condition with respect to the other item types, even when considering short sentences with an infrequent target word. The relationship found in the early-CI group between short-term memory, as measured by the FDS subtest, and performance on longer sentences might provide some explanation for this observation. The additional complexity involving longer sentences for children in this group when the difficulty in selecting the correct target word is increased as a result of its low frequency of use might be due, at least in part, to problems maintaining information in an active or easily accessible state; which is essential for several different operations that must be performed to determine the appropriate word when its meaning is unknown. This is an extremely interesting finding; especially considering the absence of significant differences in short-term memory span observed between children with early-CI and their control group peers. It definitely highlights the need to deepen our understanding of the role of working memory in sentence reading comprehension. Specifically, future research should study the executive working memory processes required to maintain the information needed for on-line processing when this exceeds the short-term memory span. In summary, although the three groups presented a similar pattern of results across the various testing conditions, only the early-CI children’s performance level tended to resemble that of NH children, except under conditions of greater difficulty, where the DES/S test revealed some weaknesses in their sentence reading comprehension. The error analysis merits its own discussion. Recall that this work incorporates a novelty with respect to the previous studies already mentioned. This innovation basically consisted of replacing one of the three semantic foils by a syntactic one, namely a word that is semantically incongruent but morphosyntactically congruent with the end of the sentence. Earlier studies using the original test concluded that deaf children and adults rely on the so-called key word strategy to perform this task, which means that deaf individuals select a word that fits into the overall meaning of the sentence that is not understood (i.e., a semantic foil) when they have to choose a word to complete it (Domı´nguez & Alegrı´a, 2010; Domı´nguez et al., 2014, 2012). The results of this study showed that children with late-CI made significantly more errors than children with early-CI, who in turn made more errors than NH children. This was particularly true in the case of semantic errors, with the late-CI group making considerably more semantic errors than the other two groups, and the early-CI group more than the NH group. However, when considering specifically syntactic errors, significant differences were only found between NH and late-CI children (see Fig. 3). The rate at which children with early-CI made syntactic errors in the DES/S test was also at an intermediate point between those in the control (NH) and late-CI groups, and though their overall performance was close to that of NH children, they made significantly more semantic errors. In general, the children in the NH and early-CI groups made few errors, especially the former, and when they gave a wrong answer it was preferentially associated with the choice of a syntactic foil. The fact that NH and early-CI children, unlike children with late-CI, made a significantly higher number of errors due to choosing a syntactic instead of a semantic foil

164

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

seems to indicate that both groups preferentially use syntactic cues when trying to understand and complete sentences. This could mean that children with early-CI have incorporated syntactic processing in sentence comprehension, being able to guide their comprehension through syntactic cues. Note that this in no case necessarily implies syntactic sentence comprehension in children with early-CI; rather, it means that in the absence of proper understanding, they tend to choose a word that fits morphosyntactically in the sentence, which indicates that they rely on this type of information. In contrast, the mean number of errors in the late-CI group was very high, though if the likelihood of making one type of error or another is considered, no difference is observed in the ratio of errors of either type made by this group (see Fig. 3). This could mean that children with late-CI in fact occasionally use general semantic cues, in keeping with the key word strategy, but also that they sometimes employ syntactic cues, thus using both strategies. However, a more plausible interpretation for the fact that errors resulting from selecting a syntactic foil coexist with such a high number of semantic errors is that, in the absence of comprehension, this group of children generally responds inconsistently or haphazardly, without consistently using cues of any kind. The reason for these random errors in the late-CI group is unclear. Reading comprehension is a high-order complex process that requires the integrated use of lexical-semantic, syntactic and pragmatic knowledge, as well as the coordinated use of a set of cognitive skills. Thus, the error pattern obtained during the performance of the DES/S test by children with lateCI in contrast to the other two study groups might also be a consequence of their differences in vocabulary and short-term memory. It could even be due to other factors. For example, one tentative explanation, and an alternative to that based on a cognitive or linguistic deficit, could be related to a greater use of visual strategies during reading in hearing-impaired children compared with normally hearing children. It has been observed that families with deaf children, especially those with deaf parents, often tend to focus their children’s attention on visual attributes rather than auditory cues in shared reading (Berke, 2013). In addition, it has been suggested that readers with difficulty accessing phonology might use a visual access to morphology as an alternative strategy for a skilled reading (Gaustad, 2000). It is possible that children with late-CI rely on this type of visual aid to a greater extent than other children in this study, and the fact that there was limited time to perform the test could have prevented them from reading at their own pace, resulting in errors with no apparent pattern. However, the children in the late-CI group used additional time to complete the test, so any comprehension errors observed cannot be explained by reading too fast. In this respect, we should also note that, although there was a small percentage of signers in the late-CI group, they used sign language in combination with spoken language, the latter being the family’s preferred mode of communication. In conclusion, our results are only partially congruent with those of previous studies (Domı´nguez et al., 2012; Miller et al., 2012; Soriano et al., 2006), as they do not agree on the sole use of the strategy of key words by deaf individuals. When the deaf children in the late-CI group failed to correctly complete the sentence, they chose the syntactic and semantic foils in the same proportion. In the original test, there was no other possibility but to choose a semantic foil, since all of the alternatives to the correct answer were semantic in nature, forcing a semantic error every time. Therefore, it cannot be concluded from those studies that deaf individuals in fact rely solely on the key word strategy. Our findings indicate that in a large number of cases, they choose a syntactic foil. To ascertain whether they are in fact relying on syntactic cues in some cases and semantic ones in others, or whether they are simply answering randomly when they do not understand a sentence, the task would have to be modified by introducing an unrelated foil that was equally likely to be selected, along with the semantic and syntactic foils. Under conditions of equiprobable choice, it could be conclusively proven whether responses are guided by cues of one type or another, or by chance. In any case, this study confirms that early-CI certainly has a positive impact on language development, specifically on sentence reading comprehension, and provides direct evidence that deaf children with early-CI do use syntactic strategies. 5. Summary and conclusion The results of our study indicate that the use of a cochlear implant, when installed during early development, significantly improves the reading comprehension of prelingually deaf children, enabling them to use syntactic cues to understand written sentences. This improvement depends critically on the age of implantation. Obviously, an earlier implantation means earlier exposure to spoken language and to reliable listening, which is associated in turn with other factors of a technical and social nature that enhance the development of cognitive and language skills related to reading comprehension, such as a larger vocabulary and a longer short-term memory span. Thus, the pattern of results that arises when comparing NH and CI groups in the use of syntactic strategies in sentence reading comprehension might also be a consequence of their differences in these skills. The results also show that in spite of these improvements, children that are implanted even at an early age continue to have difficulties with grammatical comprehension, especially in situations that impose high temporal processing demands. These difficulties keep them from reaching a normalized level of reading comprehension in most cases. Consequently, regardless of the improvements that the cochlear implant may offer as a result of a greater and better exposure to spoken language, a specific intervention that is aimed at understanding words, especially at understanding the relationships between them, is needed to improve the reading comprehension of these children. According to our findings, any conclusion concerning the use of the so-called key word strategy by deaf individuals to enhance reading comprehension could have been derived from the procedures used to measure it, and in any case, it should not be extended to early implanted children.

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

165

Conflict of interest The authors report no conflicts of interest. The authors alone are solely responsible for the content and writing of this paper. Acknowledgements This research project (UCM Art. 83, 417/2010, 40/2014) received financial support from CLAVE Caring for Hearing Impairment, Spain. The authors would like to thank the CLAVE professionals for their help in collecting data as well. We are truly grateful to the children, their families and teachers, and to the clinical specialists in cochlear implants who participated in this study. We also thank to Professor John Grinstead from The Ohio State University (Columbus, USA) for his help with the English edition of the final version of this manuscript. Appendix Semantic and syntactic strategies detection test Below you will see some sentences with the last word missing. Your task is to choose the word that best completes the sentence. First, read the incomplete statement carefully and then read each of the four alternative words given to complete it. Once this is done, check what you think is the best option with your pencil and then quickly move to the next statement. Let’s see an example. Firstly, we will read the statement ‘‘Yesterday we saw a movie. . .’’ and then we will continue reading all the words proposed in order to select the one that best completes the sentence. Since the correct answer is ‘‘nice’’, we will put an (X) in the square preceding it.

166

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

167

168

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

Note. We have added here capital letters in subscript in order to convey some of the morphosyntactic features of the possible answers. This information consists of the number (S: singular; P: plural), gender (F: feminine; M: masculine; 0: neutral) and syntactic category (N: noun; A: adjective; V: verb) of the content words used.

References Alameda, J. R., & Cuetos, F. (1995). Diccionario de las unidades lingu¨ı´sticas del castellano. Vol. I: Orden Alfabe´tico/Vol. II: Orden por Frecuencias. [Dictionary of Castilian linguistic units. Vol. I: Alphabetical Order/Vol. II: Frequency Order]. Oviedo, Spain: Servicio de Publicaciones de la Universidad de Oviedo. Alegrı´a, J., Domı´nguez, A. B., & van der Straten, P. (2009). Co´mo leen los sordos adultos? La estrategia de palabras clave [How do deaf adults read? The key word strategy]. Revista de Logopedia, Foniatrı´a y Audiologı´a, 29(3), 195–206. http://dx.doi.org/10.1016/S0214-4603(09)70028-2 Archbold, S., Harris, M., O’Donoghue, G., Nikolopoulos, T., White, A., & Richmond, H. L. (2008). Reading abilities after cochlear implantation: The effect of age at implantation on outcomes at 5 and 7 years after implantation. International Journal of Pediatric Otorhinolaryngology, 72(10), 1471–1478. http:// dx.doi.org/10.1016/j.ijporl.2008.06.016 ?

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

169

Arfe, B., Rossi, C., & Sicoli, S. (2015). The contribution of verbal working memory to deaf children’s oral and written production. Journal of Deaf Studies and Deaf Education, 20(3), 203–214. http://dx.doi.org/10.1093/deafed/env005 Asker-A´rnason, L., Wass, M., Gustafsson, F., & Sahle´n, B. (2015). Reading comprehension and working memory capacity in children with hearing loss and cochlear implants or hearing aids. Volta Review, 115(1), 35–65. Retrieved from http://agbell.publisher.ingentaconnect.com/content/agbell/vr Asker-A´rnason, L., Wass, M., Ibertsson, T., Lyxell, B., & Sahle´n, B. (2007). The relationship between reading comprehension, working memory and language in children with cochlear implants. Acta Neuropsychologica, 5(4), 163–186. Retrieved from http://1035.indexcopernicus.com/index.php?p=1 Berke, M. (2013). Reading books with young deaf children: Strategies for mediating between American Sign Language and English. Journal of Deaf Studies and Deaf Education, 18(3), 299–311. http://dx.doi.org/10.1093/deafed/ent001 Boons, T., Brokx, J. P. L., Frijns, J. H. M., Peeraer, L., Philips, B., & Vermeulen, A. (2012). Effect of pediatric bilateral cochlear implantation on language development. Archives of Pediatrics & Adolescent Medicine, 166(1), 28–34. http://dx.doi.org/10.1001/archpediatrics.2011.748 Boons, T., De Raeve, L., Langereis, M., Peeraer, L., Wouters, J., & van Wieringen, A. (2013a). Expressive vocabulary, morphology, syntax and narrative skills in profoundly deaf children after early cochlear implantation. Research in Developmental Disabilities, 34(6), 2008–2022. http://dx.doi.org/10.1016/ j.ridd.2013.03.003 Boons, T., De Raeve, L., Langereis, M., Peeraer, L., Wouters, J., & van Wieringen, A. (2013b). Narrative spoken language skills in severely hearing impaired school-aged children with cochlear implants. Research in Developmental Disabilities, 34(11), 3833–3846. http://dx.doi.org/10.1016/j.ridd.2013.07.033 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(1), 63–88. http://dx.doi.org/10.1016/S0022-0965(03)00033-X Burkholder, R. A., & Pisoni, D. B. (2004). Digit span recall error analysis in pediatric cochlear implant users. International Congress Series, 1273, 312–315. http://dx.doi.org/10.1016/j.ics.2004.08.030 Caselli, M. C., Rinaldi, P., Varuzza, C., Giuliani, A., & Burdo, S. (2012). Cochlear implant in the second year of life: Lexical and grammatical outcomes. Journal of Speech Language, and Hearing Research, 55(2), 382–394. http://dx.doi.org/10.1044/1092-4388(2011/10-0248) Cleary, M., Pisoni, D. B., & Geers, A. E. (2001). Some measures of verbal and spatial working memory in eight- and nine-year-old hearing impaired children with cochlear implants. Ear and Hearing, 22(5), 395–411. Retrieved from http://journals.lww.com/ear-hearing/toc/2001/10000 Connor, C. M. (2006). Examining the communication skills of a young cochlear implant pioneer. Journal of Deaf Studies and Deaf Education, 11(4), 449–460. http://dx.doi.org/10.1093/deafed/enl001 Connor, C. M., & Zwolan, T. A. (2004). Examining multiple sources of influence on the reading comprehension skills of children who use cochlear implants. Journal of Speech Language, and Hearing Research, 47(3), 509–526. http://dx.doi.org/10.1044/1092-4388(2004/040) Conrad, R. (1979). The deaf schoolchild. London, England: Harper and Row. Cuetos, F., Rodrı´guez, B., Ruano, E., & Arribas, D. (2007). PROLEC-R. Baterı´a de Evaluacio´n de los Procesos Lectores Revisada. [PROLEC-R. Evaluation of Reading Processes for Children, Revised Edition]. Madrid, Spain: TEA Ediciones. Dillon, C., & Pisoni, D. B. (2004). Nonword repetition and reading in deaf children with cochlear implants. International Congress Series, 1273, 304–307. http://dx.doi.org/10.1016/j.ics.2004.07.042 Domı´nguez, A. B. (2003). Co´mo acceden los alumnos sordos al lenguaje escrito? [How deaf children access to written language?]. Ensen˜anza. Anuario Interuniversitario de Dida´ctica, 21, 201–218 Retrieved from http://dx.doi.org/10.14201/et Domı´nguez, A. B., & Alegrı´a, J. (2010). Reading mechanisms in orally educated deaf adults. Journal of Deaf Studies and Deaf Education, 15(2), 136–148. http:// dx.doi.org/10.1093/deafed/enp033 Domı´nguez, A. B., Carrillo, M. S., Pe´rez, M. M., & Alegrı´a, J. (2014). Analysis of reading strategies in deaf adults as a function of their language and metaphonological skills. Research in Developmental Disabilities, 35(7), 1439–1456. http://dx.doi.org/10.1016/j.ridd.2014.03.039 Domı´nguez, A. B., Pe´rez, I., & Alegrı´a, J. (2012). La lectura en los alumnos sordos: Aportacio´n del implante coclear [Reading in deaf students: The role of the cochlear implant]. Infancia y Aprendizaje: Journal for the Study of Education and Development, 35(3), 327–341. http://dx.doi.org/10.1174/ 021037012802238993 Domı´nguez, A. B., & Soriano, J. (2009). Mecanismos de lectura empleados por personas sordas adultas consideradas como buenas lectoras [Reading mechanisms used by deaf adult people considered good readers]. Bordo´n, 61(4), 9–20. Retrieved from http://recyt.fecyt.es/index.php/BORDON/index Dunn, L., Dunn, M., & Arribas, L. M. D. (2006). TVIP. Test de vocabulario en ima´genes Peabody. Adaptacio´n espan˜ola del PPVT-III. [Peabody Picture Vocabulary Test]. Madrid, Spain: TEA Ediciones Spanish version. Ferna´ndez, M., & Anula, A. (2002). Sintaxis y cognicio´n. Introduccio´n al conocimiento el procesamiento y los de´ficits sinta´cticos. [Syntax and cognition. An introduction to knowledge, processing, and syntactic deficits]. Madrid, Spain: Sı´ntesis. Gaustad, M. (2000). Morphographic analysis as a word identification strategy for deaf readers. Journal of Deaf Studies and Deaf Education, 5(1), 60–80. http:// dx.doi.org/10.1093/deafed/5.1.60 Geers, A. E. (2003). Predictors of reading skill development in children with early cochlear implantation. Ear and Hearing, 24(Suppl. 1), 59S–68S. http:// dx.doi.org/10.1097/01.AUD.0000051690.43989.5D Geers, A. E. (2004). Speech, language and reading skills after early cochlear implantation. Archives of Otolaryngology – Head and Neck Surgery, 130(5), 634– 638. http://dx.doi.org/10.1001/archotol.130.5.634 Geers, A. E., Tobey, E., Moog, J., & Brenner, C. (2008). Long-term outcomes of cochlear implantation in the preschool years: From elementary grades to high school. International Journal of Audiology, 47(Suppl. 2), S21–S30. http://dx.doi.org/10.1080/14992020802339167 Harris, M. S., Kronenberger, W. G., Gao, S., Hoen, H. M., Miyamoto, R. T., & Pisoni, D. B. (2013). Verbal short-term memory development and spoken language outcomes in deaf children with cochlear implants. Ear and Hearing, 34(2), 179–192. http://dx.doi.org/10.1097/AUD.0b013e318269ce50 Johnson, C., & Goswami, U. (2010). Phonological awareness, vocabulary, and reading in deaf children with cochlear implants. Journal of Speech, Language, and Hearing Research, 53(2), 237–261. http://dx.doi.org/10.1044/1092-4388(2009/08-0139) Jua´rez, A., Monfort, M., & Monfort, I. (2005). Rehabilitacio´n auditiva postaudı´fono y postimplante en hipoacusias severas y profundas. In E. Salesa, E. Perello´, & A. Bonavida (Eds.), Tratado de audiologı´a. Barcelona, Spain: Elsevier Masson. Kuhl, P. K., Conboy, B. T., Padden, D., Nelson, T., & Early, J. P. (2005). Early speech perception and later language development: Implications for the ‘‘critical period’’. Language Learning and Development, 1(3-4), 237–264. http://dx.doi.org/10.1080/15475441.2005.9671948 Lo´pez-Higes, R., Gallego, C., Martı´n-Aragoneses, M. T., & Melle, N. (2015). Morpho-syntactic reading comprehension in children with early and late cochlear implants. Journal of Deaf Studies and Deaf Education, 20, 136–146. http://dx.doi.org/10.1093/deafed/env004 Marı´n, J., & Carrillo, M. S. (1999). Test Colectivo de Eficacia Lectora (TECLE). Murcia, Spain: Departamento de Psicologı´a Ba´sica y Metodologı´a, Universidad de Murcia (unpublished manuscript). Retrieved from http://www.um.es/langpsy/Publicaciones/TECLE_Marin%20y%20Carrillo%20(1999).pdf Markman, T. M., Quittner, A. L., Eisenberg, L. S., Tobey, E. A., Thal, D., Niparko, J. K., & Wang, N. Y. (2011). Language development after cochlear implantation: An epigenetic model. Journal of Neurodevelopmental Disorders, 3(4), 388–404. http://dx.doi.org/10.1007/s11689-011-9098-z Marschark, M., Rhoten, C., & Fabich, M. (2007). Effects of cochlear implants on children’s reading and academic achievement. Journal of Deaf Studies and Deaf Education, 12(3), 269–282. http://dx.doi.org/10.1093/deafed/enm013 Marschark, M., Sarchet, T., Rothen, C., & Zupan, M. (2010). Will cochlear implants close the reading achievement gap for deaf students? In P. Nathan, M. Marschark, & P. E. Spencer (Eds.), The Oxford handbook of deaf studies, language, and education (Vol. 2). Oxford, England: Oxford University Press. Miller, P. (2000). Syntactic and semantic processing in Hebrew readers with prelingual deafness. American Annals of the Deaf, 145(5), 436–451. http:// dx.doi.org/10.1353/aad.2012.0116 Miller, P. (2005). Reading comprehension and its relation to the quality of functional hearing: Evidence from readers with different functional hearing abilities. American Annals of the Deaf, 150(3), 305–323. http://dx.doi.org/10.1353/aad.2005.0031 Miller, P. (2006). What the visual word recognition skills of prelingually deafened readers tell about their reading comprehension problems. Journal of Developmental and Physical Disabilities, 18(2), 91–121. http://dx.doi.org/10.1007/s10882-006-9002-z ?

170

C. Gallego et al. / Research in Developmental Disabilities 49–50 (2016) 153–170

Miller, P. (2010). Phonological, orthographic, and syntactic awareness and their relation to reading comprehension in prelingually deaf individuals: What can we learn from skilled readers? Journal of Development and Physical Disabilities, 22(6), 549–561. http://dx.doi.org/10.1007/s10882-010-9195-z Miller, P., Kargin, T., Guldenoglu, B., Rathmann, C., Kubus, O., Hauser, P., & Spurgeon, E. (2012). Factors distinguishing skilled and less skilled deaf readers: Evidence from four orthographies. Journal of Deaf Studies and Deaf Education, 17(4), 439–462. http://dx.doi.org/10.1093/deafed/ens022 Miyamoto, R. T., Hay-McCutcheon, M. J., Kirk, K. I., Houston, D. M., & Bergeson-Dana, T. (2008). Language skills of profoundly deaf children who received cochlear implants under 12 months of age: A preliminary study. Acta Oto-Laryngologica, 128(4), 373–377. http://dx.doi.org/10.1080/ 00016480701785012 Nicholas, J. G., & Geers, A. E. (2008). Expected test scores for preschoolers with a cochlear implant who use spoken language. American Journal of Speech Language Pathology, 17(2), 121–138. http://dx.doi.org/10.1044/1058-0360(2008/013) Perfetti, C. A., Landi, N., & Oakhill, J. (2005). The acquisition of reading comprehension skill. In M. J. Snowling & C. Hulme (Eds.), The science of reading: A handbook (pp. 227–247). Oxford, England: Blackwell. Perfetti, C. A., & Sandak, R. (2000). Reading optimally builds on spoken language: Implications for deaf readers. Journal of Deaf Studies and Deaf Education, 5(1), 32–50. http://dx.doi.org/10.1093/deafed/5.1.32 Prezbindowski, A. K., & Lederberg, A. R. (2003). Vocabulary assessment of deaf and hard of hearing children from infancy through to the preschool years. Journal of Deaf Studies and Deaf Education, 8(4), 383–400. http://dx.doi.org/10.1093/deafed/eng031 Sainz, M., & de la Torre, A. (2005). Implante coclear. In E. Salesa, E. Perello´, & A. Bonavida (Eds.), Tratado de Audiologı´a. Madrid, Spain: Elsevier Masson. Sharma, A., Dorman, M. F., & Kral, A. (2005). The influence of a sensitive period on central auditory development in children with unilateral and bilateral cochlear implants. Hearing Research, 203(1–2), 134–143. http://dx.doi.org/10.1016/j.heares.2004.12.010 Sharma, A., Nash, A., & Dorman, M. (2009). Cortical development, plasticity and reorganization in children with cochlear implants. Journal of Communication Disorders, 42(4), 272–279. http://dx.doi.org/10.1016/j.jcomdis.2009.03.003 Soriano, J., Pe´rez, I., & Domı´nguez, A. B. (2006). Evaluacio´n del uso de estrategias sinta´cticas en lectura por alumnos sordos con y sin implante coclear [Assessment of syntactic strategies used by deaf students, with and without cochlear implants, while reading]. Revista de Logopedia, Foniatrı´a y Audiologı´a, 26(2), 72–83. http://dx.doi.org/10.1016/S0214-4603(06)70105-X Sparreboom, M., Langereis, M. C., Snik, A. F. M., & Mylanus, E. A. M. (2015). Long-term outcomes on spatial hearing, speech recognition and receptive vocabulary after sequential bilateral cochlear implantation in children. Research in Developmental Disabilities, 36(1), 328–337. http://dx.doi.org/10.1016/ j.ridd.2014.10.030 Spencer, L. J., Gantz, B. J., & Knutson, J. F. (2004). Outcomes and achievement of students who grew up with access to cochlear implants. Laryngoscope, 114(9), 1576–1581. http://dx.doi.org/10.1097/00005537-200409000-00014 Spencer, P. E., & Marschark, M. (2010). Evidence-based practice in educating deaf and hard-of-hearing students. New York: Oxford University Press. Svirsky, M. A., Stallings, L. M., Lento, C. L., Ying, E., & Leonard, L. B. (2002). Grammatical morphologic development in pediatric cochlear implant users may be affected by the perceptual prominence of the relevant markers. Annals of Otology, Rhinology & Laryngology, 111(Suppl.), 109–112. Retrieved from http://aor.sagepub.com/ Wechsler, D. (2004). WISC-IV. Escala de Inteligencia Wechsler para Nin˜os – IV. [Wechsler Intelligence Scale for Children – Fourth edition]. Madrid, Spain: TEA Ediciones.