A Quantitative and Qualitative Assessment of Verbal Short-Term Memory and Phonological

A QUANTITATIVE AND QUALITATIVE ASSESSMENT OF VERBAL SHORT-TERM MEMORY AND PHONOLOGICAL PROCESSING IN 8-YEAR-OLDS WITH A HISTORY OF REPETITIVE OTITIS MEDIA

Steve Majerus12, Pierre Amand1, Vincent Boniver3, Jean-Pierre Demanez3,

Laurent Demanez3Martial Van der Linden14

1Department of Cognitive Sciences, University of Liege, Belgium

2National Fund of Scientific Research, Belgium

3Department of Oto-Rhino-Laryngology, University Hospital of Liege, Belgium

4Cognitive Psychopathology and Cognitive Neuropsychology Unit, University of Geneva, Switzerland

Accepted for publication in the Journal of Communication Disorders

Address for correspondence

Steve Majerus

Department of Cognitive Sciences / Cognitive Psychopathology Sector

University of Liege

Boulevard du Rectorat, B33

4000 Liège

Tel: 0032 43664656

Fax: 0032 43662808

E-mail:

Abstract

Language outcome in children experiencing fluctuant hearing loss due to otitis media (OME) remains highly equivocal. In the current study, we assessed performance on highly sensitive verbal short-term memory (STM), new word learning and phonological processingtasks in 8-year-old children who had suffered from recurrent OME before the age of 3. Relative to a control group with no history of OME, we observed strictly normal performance for different STM and new word learning tasks. Performance on these tasks was also normally influenced by phonotactic, lexical and semantic variables. However, at the level of phonological processing, a small but significant decrease of performance was found in a speeded nonword identification task and a rhyme judgment task. The results of this study suggest that outcome of OME is characterized by subtle impairments at the level of perceptual-phonological analysis, but there is no significant impact on verbal STM and new word learning abilities.

Learning outcomes: As a result of this activity, the participant will be able to (1) Explain the outcome of recurrent OME before age 3 on later language and verbal STM development. (2) Be aware of the complex relationships that link language development and verbal STM. (3) Explain how fluctuant hearing loss during infancy and early childhood could affect verbal STM development and learning capacity for new phonological information. (4) Describe different verbal STM measures that distinguish retention capacities for phonological and lexico-semantic information. (5) Explain the influence of phonotactic frequency on nonword processing in language and verbal STM tasks.

Keywords: Otitis media; Verbal short-term memory; Central auditory processing; Phonological processing; Phonological learning

Introduction

The current study exploresthe impact of recurrent otitis media with effusion (OME) during infancy and early childhood and its related hearing loss onverbal short-term memory (STM)outcome at the age of 8 years. Although language outcome has been frequently studied in children with a history of OME, potential adverse effects on verbal STM functioning have not yet received systematic investigation. This is surprising, given that verbal STM is a critical function involved in language development itself as well as in other cognitive skills such as arithmetic and foreign language learning.Before discussing more specifically the possible existence of verbal STM limitations in children with a history of OME, we will briefly review the existing literature on outcome of language processing measures.

The impact of recurrent OME before age 3 on later language outcome

After more than 30 years of research and more than 100 reports or meta-analyses, the long-term outcome of language development in children with recurrent OME is still highly equivocal. A number of recent studies have reported increased risk for delays in measures of phonological processing (articulation; sound discrimination), vocabulary knowledge, verbal comprehension and readingin children aged 3-10 years (e.g., Bennett & Haggard, 1999; Lindsay, Tomazic, Whitman & Accardo, 1999; Kindig & Richards, 2000; Klausen, Moller, Holmefjord, Reisaeter, & Asbjornsen, 2000; Maw, Wilks, Harvey, Peters, & Golding, 1999; Nittrouer & Burton, 2005; Rosenfeld et al., 2000; Shriberg, Flipsen et al., 2000; Shriberg, Flipsen, Kwiatkowski, & McSweeny, 2003;Shriberg, Friel-Pratti, Flipsen, & Brown, 2000; Shriberg, Kent, Karlsson, McSweeny, Nadler & Brown, 2003). However, other studies did not reveal any significant associationsbetween OME and language outcome (Feldman et al., 1999; Paradise et al., 2000, 2001; Roberts et al., 1998, 2000, 2002; Rovers et al., 2000). A recent meta-analysis by Roberts, Rosenfeld and Zeisel (2004) found no consistent association between a number of standardized language measures at preschool age and OME related hearing loss. Although some occasional significant associations were observed, they were very small and the authors questioned their clinical relevance.It is important to note that the studies observingsignificant associations between a positive history of OME and language impairment are often those that use carefully designed psycholinguistic tasks while the studies obtaining null effects more often use standardized test batteries or parental checklists (e.g., Paradise et al., 2001; Roberts et al, 2002; see also Briscoe, Bishop and Frazier-Norbury, 2002, for related findings in children with sensori-neural hearing loss). Furthermore, studies investigating qualitative aspects of language development also observed subtleabnormalities in speech production. For example, Shriberg, Kent et al. (2003) showed qualitative changes in speech samples of preschool children with a positive history of early OME, as reflected by an increased rate of backing (abnormal posterior lingual positioning)of a number of consonants whose correct pronunciation requires an anterior lingual positioning (e.g., /k/, /z/, and /G/). In sum, both quantitative and qualitative effects of OME on outcome of receptive and productive language development have been reported, even though the effects remain small and inconsistent. As the possible language impairments related to OME seem to be subtle, their detection requires theadministration of tasks that are carefully designed with respect to their sensitivity and the underlying psycholinguistic components they are assumed to measure.

The impact of recurrent OME on later verbal STM outcome

Paralleling the extensive use of rather general language measures for exploring long-term effects of OME-related fluctuant hearing loss on language outcome, there are currently very few studies that have provided a detailed and qualitative exploration of verbal STM outcome in children with a history of OME. Indeed, many studies have included digit span measures in their more general language or auditory perception test batteries, showing no differences between children with a history of OME and control groups (e.g., Brandes & Ehinger, 1981; Hoffman-Lawless, Keith, & Cotton, 1981; Schilder, Snik, Straatman, & van den Broek, 1994). However, digit span is not a very sensitive measure. Studiesfrom other research fields show that patients withdevelopmental STM deficits might have relatively preserved digit span, while displaying significantly reduced performance for more sensitive STM tasks, such as immediate serial recall of word or nonword lists (e.g., patient TG in Majerus, Van der Linden, Poncelet & Metz-Lutz, 2004; patient BS in Hanten & Martin, 2001). One study,which used other test material than digit span for exploring verbal STM outcome in children with a positive history of OME, is the work described by Mody, Schwartz, Gravel and Ruben (1999). These authors compared identification and immediate serial recall for phonetically similar and dissimilar nonsense syllables in 9-year-old children with a history of OME during infancy and observed poor performance only for the phonetically similar condition, and this for both identification and immediate serial recall, suggesting residual impairments at the level of perceptual-phonetic analysis rather than genuine verbal STM storage capacities. On the other hand, in 5-year-old children with a positive history of OME, Nittrouer and Burton (2005) observed poor performance in a verbal STM task for both phonologically similar and dissimilar word conditions, but no deficits in a speech perception task requiring fine-grained acoustical analysis. It should be noted that the verbal STM task used by Nittrouer and Burton was characterized by task demands which were relatively complex for 5-year-old children (the children had to listen to a sequence of three or four words, and then had to rearrange pictures in the order in which the words were heard); thus,difficultieswith comprehension of the orally presented task requirements might have affected the results. In support of this possibility, the OME group in the study by Nittrouer and Burtonwas alsoimpaired for the comprehension of complex sentences.

Overall, evidence for residual STM impairments in children with a history of OME remains conflicting. However, it is important to establish a comprehensive picture of verbal STM outcome in these children as verbal STM capacities have a major impact on the development of other cognitive abilities, such as language development, foreign language learning and reading.In typically developing children, a substantial body of research has shown that verbal STM capacity predicts many aspects of language development, such as productive and receptive vocabulary knowledge, speed of acquisition of new lexico-semantic information and sentence production, for both native and foreign languages (Adams & Gathercole, 1995, 1996, 2000; Avons, Wragg, Cupples, & Lovegrove, 1998; Bowey, 1996; Gathercole, 1995; Gathercole & Adams, 1993, 1994; Gathercole & Baddeley 1989, 1990, 1993; Gathercole, Willis & Baddeley, 1991; Gathercole, Frankish, Pickering, & Peaker, 1999; Michas & Henry, 1994; Service 1992; Service & Kohonen, 1995). Moreover, children with developmental verbal short-term memory impairments display severe difficulties in acquiring a foreign language vocabulary (e.g., Hanten & Martin 2001). These empirical data have led to the development of a number of STM models assuming the existence of strong interactions between verbal STM and language processing (e.g., Baddeley, Gathercole & Papagno,1998; Gupta & MacWhinney, 1997; Hartley & Houghton, 1996). Certain models even consider verbal STM to be a necessary gateway through which new verbal information entering long-term memory must pass (e.g., Baddeley et al., 1998). In older children with a history of OME, limitations in verbal STM capacity might have no serious impact on native language functioninganymore as initially delayed native language development might eventually have become normal due to intensive experience and possible speech and language therapy. However, reductions in verbal STM capacity might still make more difficult the acquisition of a foreign language phonology and vocabulary, which will negatively affect the educational and career prospects of these children.

The aim of the current study was to realize a detailed investigation of both quantitative and qualitative aspects of verbal STM functioning in 8-year old children with and without a history of recurrent OME. In a first set of tasks, we determined verbal STM performance for immediate serial recall of word and nonword lists of increasing length. We were not only interested in overall levels of performance relative to an age-matched control group, but also in the impact that lexical, semantic and phonological knowledge can have on verbal STM performance, by comparing recall for lists of words of high or low lexical frequency (frequency effect), for lists of words of high and low imageability ratings (imageability effect) and for lists of words and nonwords (lexicality). Indeed,verbal STM performance is not only determined by intrinsic short-term storage capacity, but also by the availability of language knowledge about the items that have to be stored. For example, recall of word lists yields consistently better performance than recall of nonword lists in both children and adults (e.g., Hulme, Maughan & Brown, 1991; Gathercole et al., 1999; Majerus & Van der Linden, 2003; Roodenrys, Hulme & Brown, 1993). Similarly, word frequency and word imageability have also been reported to affect immediate serial recall of word lists at any ages (e.g., Majerus & Van der Linden, 2003). Given these considerations, OME related fluctuant hearing loss could affect verbal STM in two different ways: (1) residual perceptual deficits might hinder the construction of appropriate phonological codes in verbal STM leading to poorer storage capacity for any type of verbal items (e.g., Mody et al., 1999; Nittrouer & Burton, 2001); (2) impoverished perceptual input during infancy and early childhood might lead to a more poorly structured network of permanent phonological, lexical and semantic representations, affecting also verbal STM as performance in these tasks very directly depends on the integrity of language representations; in that case, we should expect not only poorer performance in verbal STM tasks, but effects of word frequency, lexical and semantic knowledge might also be abnormal.

The hypothesis of a poorly structured language network as underlying weak verbal STM performance was further explored through nonword processing tasks investigating more precisely the degree of segmentation of phonological representations. Indeed, processing of nonwords, in both simple identification and verbal STM tasks, depends on the availability of finely segmented phonological representations and phonological knowledge. For example, Gathercole et al. (1999) showed that nonwords containing phoneme combinations which are frequent in the native language of the participating children yielded higher recall performance than nonwords made of low frequency phoneme associations. These data suggest that nonword recall is influenced by some form of statistical knowledge about phoneme combinations in a given language (the so-called nonword phonotactic frequency effect). In a recent study, we further showed that nonword repetition in children was influenced by new sublexical phonological knowledge that had been learned incidentally just before the nonword recall task (Majerus, Van der Linden, Mulders, Meulemans, & Peters, 2004). Vitevitch and Luce (1998, 1999) also demonstrated that speeded single nonword repetition is faster and more accurate for nonwords composed of frequent phoneme associations as compared to nonwords with less frequent phoneme groupings. In the present study, we explored the influence of phonotactic knowledge on a single nonword repetition task, by investigating the presence of normal-sized phonotactic frequency effects. The degree of segmentation of the phonological network was further explored through a set of phonological awareness tasks, for which we also compared performance for nonwords with high or low phonotactic frequency patterns, as well as for words.

Furthermore, two additional STM tasks assessed recognition for phonological or lexico-semantic information, via a rhyme probe and a semantic category probe task. Martin, Shelton and Yaffee (1994) have shown dissociations between these two tasks in brain damaged patients, suggesting that STM capacities for the retention of phonological and lexico-semantic information might be distinct. In the present context, we wanted to explore whether hearing loss during infancy has a differential impact on phonological and lexico-semantic retention abilities.

Finally, in order to investigate whether a positive history of OME is associated with difficulties in later foreign language learning, as discussed earlier, we determined learning capacities for new verbal information via a word-word and a word-nonword paired associate learning task.If children with OME have limited capacitiesfor learning new phonological information, then they should also present very poor performance on the word-nonword paired associate learning task. Moreover, we also investigated the influence of phonotactic knowledge on the word-nonword paired associate learning task, by comparing performance for learning nonwords containing high or low phonotactic frequency patterns.

Lastly, we presented a set of tasks controlling for the presence of residual perceptual deficits as well as for generalvocabulary knowledge. A central auditory processing battery which assessed dichotic listening and speech-in-noise identification abilities, was administered. Measures of productive and receptive vocabulary knowledge were also obtained. In order to really assess long-term effects of OME on verbal STM and language outcome and given the complexity of some of our tasks (e.g., category probe task), we chose to study 8-year-old children rather than younger preschool age children.

Methods and Material

Participants

The experimental group was comprised of 20 children who had presented a history of repetitive and severe otitis media during infancy and early childhood. The children were chosenfrom the medical records of the Department of Oto-Rhino-Laryngology of the University Hospital of Liege. In order to be included in the study, they had to meet several criteria: (1) recurrent and serious episodes of otitis media with significant hearing loss during at least 3 months before the age of 3; (2) at least one perforation of the tympanic membrane with seromucosal secretions, evacuatedviathe insertion of a trans-tympanicventilation tube (see Table 1); (3) normal hearing at the time of the present study, as assessed by puretone audiometry for the frequencies.5, 1.0, 2.0, and 4.0 kHz; (4) estimates in the normal range for receptive vocabulary (EVIP; Dunn, Thériault-Whalen, & Dunn, 1993) and non-verbal reasoning abilities (Colored Progressive Matrices, Raven, Court & Raven, 1998); (5) no other signs of neurological or neurodevelopmental disturbances.Due to the retrospective design of the present study, it was not possible to get precise estimates of hearing loss for all episodes of OME in each child, especially for those episodes where the child was not treated by our medical department. Many children were also preverbal when first episodes of OME occurred, and hence standard behavioral audiometry requiring avolitional response by the child was not always feasible at these early periods.Also note that 6 children benefited fromspeech and language pathologyafterinsertion of trans-tympanic ventilation tubes.

The control group included 20 typically developing children with no documented episodes of otitis media and no other medical condition that could have affected auditory processing. Formal hearing screening was conducted at the time of study, using puretone audiometry as for the experimental groupand confirming normal hearing levels for each child (Table 1). Otherwise, the control children had to show normal estimates of verbal and non-verbal intellectual abilities and no signs of neurological or neurodevelopmental disorders following the same criteria as for the experimental group. They were selected from two schools of the province of Liege. Both groups had a mean chronological age of 8 years 2 months (age range: 7 years 7 months – 9 years 2 months) and were matched for gender; all the children were from middle class socio-economic backgrounds and their native language was French.The children from both groups were all from third or fourth grade classes in normal primary school settings and they showed chronological age-appropriate levels of instruction.

< INSERT TABLE 1 ABOUT HERE >