Elsevier

Brain and Language

Volume 104, Issue 1, January 2008, Pages 75-88
Brain and Language

Duration of auditory sensory memory in parents of children with SLI: A mismatch negativity study

https://doi.org/10.1016/j.bandl.2007.02.006Get rights and content

Abstract

In a previous behavioral study, we showed that parents of children with SLI had a subclinical deficit in phonological short-term memory. Here, we tested the hypothesis that they also have a deficit in nonverbal auditory sensory memory. We measured auditory sensory memory using a paradigm involving an electrophysiological component called the mismatch negativity (MMN). The MMN is a measure of the brain’s ability to detect a difference between a frequent standard stimulus (1000 Hz tone) and a rare deviant one (1200 Hz tone). Memory effects were assessed by varying the inter-stimulus interval (ISI) between the standard and deviant. We predicted that parents of children with SLI would have a smaller MMN than parents of typically developing children at a long ISI (3000 ms), but not at a short one (800 ms). This was broadly confirmed. However, individual differences in MMN amplitude did not correlate with measures of phonological short-term memory. Attenuation of MMN amplitude at the longer ISI thus did not provide unambiguous support for the hypothesis of a reduced auditory sensory memory in parents of affected children. We conclude by reviewing possible explanations for the observed group effects.

Introduction

Specific Language Impairment (SLI) is defined as delayed development of language in the absence of any obvious explanation such as hearing loss, frank neurological damage or below normal nonverbal intelligence. Among other phenotypic features of the disorder, a deficit in phonological short-term memory as measured by nonword repetition has been consistently reported (for review see Gathercole, 2006). Nonword repetition is a task in which participants are required to repeat a series of nonsense words that vary in numbers of syllables and phonotactic complexity e.g., ‘ballop’ or ‘perplisteronk’. Difficulties in performing this task may stem from deficits in the storage capacity for verbal materials, poor encoding of incoming speech materials due to impaired speech perception, weak knowledge of phonotactics due to limited vocabulary, and/or problems with speech production. In the case of children with SLI, Gathercole (2006) argues that poor performance on the task is due, at least in part, to deficits in the storage capacity of verbal short-term memory, with phonological representations showing unusually rapid decay. This conclusion is based on work by Gathercole and Baddeley (1990). In a study focused on investigating which component of short-term memory was impaired in children with SLI, they found that children with SLI were more sensitive to increasing numbers of syllables i.e., increasing memory load than they were to variations in phonotactic complexity i.e., articulatory complexity. Gathercole and Baddeley acknowledged some role for perceptual deficits, but argued that any contribution of such deficits to poor performance on nonword repetition tasks was far outweighed by the deficits in the storage capacity of short-term memory.

It is a moot point whether weak short-term memory in SLI is purely associated with verbal materials. Memory for nonverbal auditory sequences has been investigated far less than memory for verbal material, in part because of difficulties in finding a behavioural paradigm that does not introduce other task demands that confound interpretation of the data. An early study by Tallal and Piercy (1973) used a task in which children were presented with sequences of high (300 Hz) or low (100 Hz) tones and trained to associate one response button with the high tone and another with the low one. They were then required to tap out presented sequences of tones of varying lengths using the corresponding buttons. Although Tallal and Piercy were primarily interested in assessing the impact of inter-tone interval on performance, they also considered the effect of tone sequence length, and found that regardless of presentation rate, children with SLI were significantly worse at longer sequences, suggesting an impairment in auditory memory. Lincoln, Dickstein, Courchesne, Elmasian, and Tallal (1992) subsequently replicated this result and further showed that the deficit was specific to SLI, since children with a diagnosis of autism showed no evidence for a similar deficit.

Auditory short-term memory deficits have also been reported in dyslexia, a reading disorder that is frequently co-morbid with SLI (Bishop and Snowling, 2004, Catts et al., 2005). France et al. (2002) found circumstantial evidence for memory impairments in a study of frequency discrimination performance in people with dyslexia. As part of this research, they included a range of inter-stimulus intervals (ISI; silent interval between offset of one stimulus and onset of the next) from 0 to 1000 ms and observed that frequency discrimination ability in adults with dyslexia decreased with increasing ISI. Analysis of the data within the framework of a signal detection model of perceptual resolution (MacMillan, Goldberg, & Braida, 1988) suggested that the differences in frequency discrimination abilities in the group with dyslexia relative to the control group were attributable both to greater sensory variance (i.e., the internal noise associated with the processing of a stimulus) and to greater trace variance (i.e., sensory memory).

In another study involving people with dyslexia, Banai and Ahissar (2004) also noted considerable variability in their ability to perform a range of psychoacoustic tasks. Observing a bimodal distribution of frequency discrimination abilities among their participants, they subdivided them into two groups according to performance—either poor or adequate. After doing this, they found that people with dyslexia who were bad at the psychoacoustic tasks in their test battery also had deficits in working memory as measured using the digits backward recall task. Banai and Ahissar envisaged sound discrimination as a two stage process involving first early sensory encoding then sound comparison. This latter stage was performed at a higher level by working memory. They suggested that participants who did poorly on the frequency discrimination task formed a subgroup who were characterised by a deficit in working memory as distinct from verbal memory. Impaired performance on the nonverbal auditory discrimination task thus arose as a consequence of this deficit rather than because of poor initial sensory encoding.

Finally, Marler, Champlin, and Gillam (2002) combined behavioural and electrophysiological techniques involving the same backward-masking stimuli to investigate whether higher order language deficits in children with SLI developed out of a neurophysiological impairment in auditory memory. They found significantly higher thresholds in the auditory backward-masking tasks for the children with SLI and further reported that the children had particular difficulty detecting the masked stimulus if it occurred earlier in the three-interval forced choice task. This suggested deficits in auditory memory. These findings were supported by the electrophysiological data. The event-related potentials (ERPs) of the children with SLI were normal, indicating normal auditory processing. However their mismatch negativities (see below) in response to a change in intensity of the masked stimulus differed significantly from those of the typically developing children in being longer in latency and reduced in amplitude. Marler et al. concluded that the children with SLI had deficits in their early memory systems for complex sounds rather than deficits in processing of the auditory stimuli.

Though differing in research focus and interpretation of the data, all these studies suggest that the short-term memory deficits observed in participants with language or literacy impairments may in fact extend to nonverbal as well as verbal auditory input.

SLI has a genetically heritable component (for review see Stromswold, 1998). In previous research, Barry, Yasin, and Bishop (2007) found that parents of children with SLI were poor at the nonword repetition task i.e., they had deficits in verbal short-term memory. In the present study, we wished to investigate whether the memory deficits for nonverbal materials that had previously been reported for children with SLI were also present in parents of children with SLI. Such a finding would implicate a broad range of heritable short-term memory deficits in the disorder.

Psychophysical tasks, such as those employed by France et al. (2002), are not ideal for this purpose because they provide rather indirect evidence of memory impairment, and may be affected by the use of encoding strategies (e.g., implicit labelling of stimuli) and variations in attention and motivation. The simpler methods adopted by Lincoln et al. (1992) may additionally be influenced by a person’s musical background (Bishop, 2001). To avoid such confounds, we tested for auditory short-term memory deficits using electrophysiological rather than behavioural techniques, adapting a method developed by Grau, Escera, Yago, and Polo (1998) for this purpose. Specifically, we used auditory event-related potentials and the mismatch negativity.

Although purely physiological in nature, auditory event-related potentials (ERPs) show close correspondences with the early stages of information processing as outlined by Cowan, 1988, Cowan, 1995. In this model, there are three interrelated systems namely: the sensory store, comprising two phases of processing; the short-term memory store; and, the long-term memory store. The first phase of sensory memory represents the sensory response to input just prior to perception. This phase lasts long enough (several hundred milliseconds) for the input to become perceptually resolved into its component features. At this point, it is drawn into the second stage of sensory memory, ready to activate information held in long-term storage so that it can brought into the short-term memory storage area. Memory traces held in this second stage of sensory memory are predicted to last for between 10 and 20 s. With respect to speech and language processing, Cowan (1995) suggests that the first phase of sensory memory may facilitate perceptual integration of each phoneme into its phonemic context, while the second phase, operating at the sentence rather than the word-level, may hold parts of sentences in memory until they can be fully integrated with succeeding phrases in an utterance.

The mismatch negativity (MMN) is thought to provide a means for measuring sensory memory. The ERP component is elicited by means of an oddball paradigm. In this paradigm, a neural representation (memory trace) of a standard stimulus is developed through frequent repetition (Näätänen, 1992). Once this memory trace has been established, a rarely heard deviant stimulus elicits a negative-going component which peaks at around 100–180 ms after stimulus onset and is typically observed at fronto-central electrodes. This component, which is referred to as an MMN, has been commonly exploited in the clinical context to test for sound discrimination abilities. However, by comparing MMN amplitudes at different ISIs, one can also obtain an index of auditory memory; in particular if the MMN is normal at short ISIs but attenuated at long ISIs i.e., normal sound discrimination but more rapid than normal decay of sensory memory. Furthermore, by progressively increasing the ISI between standard and deviant stimuli, the MMN can be used to estimate the durability of the auditory sensory memory trace. It has thus been shown that in normal adult populations, sensory memory endures from between 4 s (Mäntysalo & Näätänen, 1987) to about 10 s (Böttcher-Gandor and Ullsperger, 1992, Sams et al., 1993, Winkler et al., 2001). It is important to note, however, that though these findings were originally interpreted in terms of duration of sensory memory, they in fact provide a measure of the period for which the preceding auditory context remains relevant to the deviant (Sussman, Sheridan, Kreuzer, & Winkler, 2003). Thus, in studies of the duration of sensory memory, the MMN is typically reported as being present or not at a certain ISI (e.g., Mäntysalo and Näätänen, 1987, Sams et al., 1993) and there is little evidence for a gradual decrease in MMN amplitude corresponding to a progressive fading of the memory trace over time (Cheour, Leppänen, & Kraus, 2000).

As noted, the MMN is a purely physiological response to stimulus change; however there are correspondences between MMN amplitude and behaviour (Näätänen & Winkler, 1999). First, size of MMN has been shown to correlate with an individual’s ability to discriminate the same changes behaviourally (e.g., Aaltonen et al., 1993, Kraus et al., 1995, Kraus et al., 1996). Second, duration of the MMN component corresponds well with the duration of the second phase of auditory sensory memory as estimated on the basis of behavioural research (Cowan, 1988, Cowan, 1995). Finally, the MMN can be elicited without active attention (Näätänen, 1992), and thus conforms well to the cognitive concept of auditory sensory memory (Broadbent, 1958).

Using the MMN, it has been shown that auditory sensory memory decreases with age. Adults with a mean age of 59 years have a considerably attenuated MMN to frequency changes relative to younger adults (mean age 22 years) after an ISI of 4500 ms (Pekkonen, Rinne, Reinikainen, Kujala, & Alho, 1996). In a similar vein, patients with Alzheimer’s disease also have a significantly attenuated auditory sensory memory after an ISI of 3000 ms compared with control subjects (Pekkonen, Jousmäki, Mervi, Reinikainen, & Partanen, 1994). Finally, the MMN has also been used to demonstrate deficits in auditory sensory memory in children with CATCH syndrome—a syndrome characterised among other things by language learning deficits (Cheour et al., 1997). Though not explicitly stated, all these studies into auditory sensory memory probed the second, not the first phase of sensory memory.

The first phase of auditory sensory memory has been studied by Winkler, Reinikainen, and Näätänen (1993) using a range of ISIs from 20 to 400 ms and brief tones in a backward-masking paradigm. Winkler et al. reported good correspondences between MMN amplitude and behaviourally determined thresholds on backward-masking tasks. They further reported that a minimum ISI of 150ms was required to elicit an MMN, however the MMN amplitude was not fully developed until after an ISI of 470 ms.

Marler et al. (2002) employing longer stimulus onset asynchrony (SOAs) intervals of 500 and 1000 ms which spanned the period corresponding to the end of the first phase of sensory memory and leading into the second phase, also reported an increase in MMN amplitude with the longer ISI. This suggests that the memory trace may extend longer than 470 ms and it further suggests that the fleeting first phase of sensory memory (half-life 50–75 ms), may not be entirely amenable to study by the MMN.

In summary, the MMN provides a useful means for probing for differences in the second phase of auditory sensory memory between groups of participants. Furthermore, electrophysiologically measured differences in sensory memory have been shown to correlate with deficits in language learning (Cheour et al., 1997, Marler et al., 2002). However, measurement of any ERP component requires the averaging of a sufficient number of responses to maximise the signal-to-noise ratio (SNR). This can make experiments quite time-consuming especially where long ISIs form part of the paradigm. For example, Gomes et al. (1999) used the MMN to investigate maturational effects on duration of the second phase of auditory sensory memory using two ISI conditions. To minimise the time required to run the experiment, they designed their paradigm around trains of eight stimuli separated by a short ISI of 250 ms with variable inter-train intervals (ITI) of either 1 or 8 s. Each train began with either a standard (1000 Hz) or a deviant (1200 Hz) tone. The paradigm successfully demonstrated developmental changes in sensory memory; however the data acquisition phase lasted between 2 and 2.5 h and the total time for testing was around 3–4 h making the method impractical for many clinical contexts.

Aside from the time required to run the paradigm used by Gomes et al. (1999), there is also an issue associated with the use of regular trains of rapidly presented tones separated by long ITIs. The aim of the train of rapid tones is to minimise the time required to establish a strong memory trace for the standard stimulus. However, as noted above, elicitation of the MMN depends on the standard and deviant tones being automatically coded as belonging to the same perceptual group. The combination of long ITIs separating groups of closely spaced stimuli allows for the standard and deviant stimuli to organise into separate temporally defined groups (Winkler et al., 2001). This means that with increasing ISI, the deviant stimulus may be pre-perceptually encoded into a different group to the standard stimulus train and no MMN will be elicited even though a memory trace to the standard may still be present. Like Gomes et al. (1999) we focused on probing the second phase of auditory sensory memory. To avoid the problems inherent in the original paradigm employed by Gomes et al., we used a paradigm that combined features from those described by Grau et al., 1998, Gomes et al., 1999. We further randomised the lengths of the trains of standard stimuli and jittered the ISIs between the stimuli in the trains to minimise the possibility of pre-perceptual grouping of stimuli. Our paradigm consisted of two parts. A memory trace to a standard 1000 Hz tone was first established by means of a rapid train of 15 standard stimuli at an ISI of 250 ms. Then a series of 73 blocks of 8 stimuli comprising seven standard stimuli and one deviant (1200 Hz) were presented. Within each block, the ISIs between stimuli alternated between 250 ms and another interval, which could be 350, 450, and either 800 ms (Condition 1) or 3000 ms (Condition 2). The deviant stimulus was only ever preceded by the ISI of interest (either 800 ms, Condition 1; or 3000 ms, Condition 2) and could occur at a quasi-random point within each block of eight stimuli. A single block from the paradigm used is illustrated in Fig. 1.

The longer ISI of 3000 ms (Condition 2) was chosen because it was greater than 2000 ms—the extent of phonological short-term memory predicted by Baddeley and Hitch (1974)—and was in the range of the estimated half-life for the second phase of sensory memory (Cowan, 1995). Furthermore, data from Winkler et al. (2001) together with our own early pilot data suggested that MMNs would be reliably obtained for typical adults at this ISI. The standard and deviant stimuli were chosen to have a large frequency difference between them which was readily discriminable for all participants. They were also the same frequencies as those used by Gomes et al. (1999) in their study probing developmental changes in the duration of auditory sensory memory.

The main predictions for this study were that: (a) MMN amplitude would be reduced for deviants at the longer ISI of 3000 ms relative to the shorter one of 800 ms, reflecting decay of the auditory memory trace for the preceding standards; (b) this effect would be amplified in parents of children with SLI compared with parents of typically developing children; (c) though we did not directly probe duration of the first phase of sensory memory (associated perceptual integration of sequences of phonemes, Cowan, 1995), we further predicted that deficits in sensory memory as reflected by reductions in the MMN amplitude in Condition 2 (ISI 3000 ms) would correlate with scores on the nonword repetition task. This is based on the assumption that deficits in one phase of sensory memory will reflect more pervasive deficits extending across both phases of sensory memory.

Section snippets

Participants

The participants for this study were recruited at the time that they gave consent for their child to participate in a larger study investigating the causes and correlates of SLI. The children (n = 27) were defined as having SLI if they had been clinically referred, had normal hearing, no neurological impairments, a nonverbal IQ (NVIQ) of 80 or above on the Wechsler Abbreviated Scale of Intelligence (WASI) (Wechsler & Chen, 1999), and performed below the 10th percentile on at least two

Auditory event-related potentials

As illustrated (Fig. 2, Fig. 3), clear P1-N1-P2 components were observed in both conditions for both groups of parents. The size of the N1 component was considerably larger in both groups of parents for Condition 2 (ISI = 3000 ms). This is to be expected because N1 had more time to recover back to its initial excitatory response level (Budd et al., 1998, Picton and Hillyard, 1988). The increase in the size of the components resulted in a larger signal-to-noise ratio (SNR) in Condition 2 across the

Discussion

The primary aim of this study was to use electrophysiological measures to test for rapid decay of auditory sensory memory traces in two groups of parents: those with children affected by SLI and those with typically developing children. Neither group of parents was homogeneous in terms of history for language learning difficulties, though there was a higher prevalence of impairment among the parents related to children with SLI. Previous research with these two groups of parents had

Acknowledgments

This research was funded by a Wellcome Trust program grant. We particularly thank both the schools who helped us to contact suitable children for the research, and families who agreed to participate. Without their support this research would not have been possible. Dr. Carles Escera is gratefully acknowledged for valuable feedback at the outset of the study regarding the design of paradigms for probing sensory memory. Finally, we thank members of OSCCI, including Faith Ayre, Joy Rosenberg, and

References (67)

  • T. McGee et al.

    Is it really a mismatch negativity? An assessment of methods for determining response validity in individual subjects

    Electroencephalography and Clinical Neurophysiology

    (1997)
  • E. Pekkonen et al.

    Variability and replicability of the mismatch negativity

    Electroencephalography and Clinical Neurophysiology

    (1995)
  • E. Plante

    MRI findings in the parents and siblings of specifically language-impaired boys

    Brain & Language

    (1991)
  • P. Tallal et al.

    Developmental aphasia: Impaired rate of non-verbal processing as a function of sensory modality

    Neuropsychologia

    (1973)
  • R. Uwer et al.

    Stability of mismatch negativities in children

    Clinical Neurophysiology

    (2000)
  • A.D. Baddeley

    The concept of working memory

  • T. Baldeweg et al.

    Impaired auditory frequency discrimination in dyslexia detected with mismatch evoked potentials

    Annals of Neurology

    (1999)
  • K. Banai et al.

    Poor frequency discrimination probes dyslexics with particularly impaired working memory

    Audiology and Neuro-Otology

    (2004)
  • J.G. Barry et al.

    Heritable risk factors associated with language impairments

    Genes, Brain and Behavior

    (2007)
  • D.V.M. Bishop

    Genetic and environmental risks for specific language impairment in children

    Philosophical Transactions of the Royal Society London. Series B

    (2001)
  • Bishop, D. V. M. (2003). Test for Reception of Grammar (version 2). London: Psychological...
  • D.V.M. Bishop et al.

    Different origin of auditory and phonological processing problems in children with language impairment: Evidence from a twin study

    Journal of Speech, Language and Hearing Research

    (1999)
  • Bishop, D. V. M. (in press). Using mismatch negativity to study the neurobiological basis of language and literacy...
  • D.V.M. Bishop et al.

    Developmental dyslexia and specific language impairment: Same or different?

    Psychological Bulletin

    (2004)
  • C. Böttcher-Gandor et al.

    Mismatch negativity in event-related potentials to auditory stimuli as a function of varying inter-stimulus interval

    Psychophysiology

    (1992)
  • A.R. Bradlow et al.

    Effects of lengthened formant transition duration on discrimination and neural representation of synthetic CV syllables by normal and learning-disabled children

    Journal of the Acoustical Society of America

    (1999)
  • D.E. Broadbent

    Perception and communication

    (1958)
  • H.W. Catts et al.

    Are specific language impairment and dyslexia distinct disorders?

    Journal of Speech, Language and Hearing Research

    (2005)
  • M. Cheour et al.

    The first neurophysiological evidence for cognitive brain dysfunctions in children with CATCH

    NeuroReport

    (1997)
  • N. Cowan

    Evolving conceptions of memory storage, selective attention, and their mutual constraints within the human information-processing system

    Psychological Bulletin

    (1988)
  • N. Cowan
    (1995)
  • S.D. Dalebout et al.

    Reliability of the mismatch negativity in the responses of individual listeners

    Journal of the American Academy of Audiology

    (2001)
  • S.D. Dalebout et al.

    Mismatch negativity to acoustic differences not differentiated behaviorally

    Journal of the American Academy of Audiology

    (1999)
  • Cited by (17)

    • Low-level neural auditory discrimination dysfunctions in specific language impairment—A review on mismatch negativity findings

      2017, Developmental Cognitive Neuroscience
      Citation Excerpt :

      Group differences have been found with a relatively rapid SOA (200–600 ms; Datta et al., 2010; Korpilahti and Lang, 1994), as discussed above, which should not pose high demands on memory-trace formation or maintenance. Studies comparing MMNs obtained in various SOA conditions support this notion, showing no MMN group differences to large frequency changes in short SOA conditions but diminished MMNs in participants with or at risk for language impairment in long SOA conditions (e.g., Barry et al., 2008; Grossheinrich et al., 2010), suggesting difficulties in maintaining the memory trace of the stimulus. This kind of faster than normal sensory memory trace decay might impair sound representation formation, which could result, e.g., in poor phonological representations in infancy.

    • Co-localisation of abnormal brain structure and function in specific language impairment

      2012, Brain and Language
      Citation Excerpt :

      The aims of the study were to characterise the brain abnormalities associated with SLI and to determine whether previously described functional and structural abnormalities were related. Participants were recruited through a research participant database of families with at least one child with SLI and families with typically-developing children who had participated in previous studies (Barry, Yasin, & Bishop, 2007; Barry et al., 2008). Participants were required to have normal hearing (a bilateral pure tone audiometric screening test at 25 db HL ISO for 500, 1000, and 2000 Hz), a non-verbal IQ (NVIQ) score of 80 or above on the Wechsler Abbreviated Scale of Intelligence (Wechsler & Chen, 1999), English as their first language, and no reported neurological impairments.

    • The mismatch negativity (MMN) - A unique window to disturbed central auditory processing in ageing and different clinical conditions

      2012, Clinical Neurophysiology
      Citation Excerpt :

      Late talkers had a shorter sensory-memory duration than that of controls, as judged from MMN data for tone-frequency change as a function of the ISI recorded at the age of 4 years 7 months, even though the language development was normal by that age (Grossheinrich et al., 2010). Furthermore, the Barry et al. (2008) recent MMN data for tone-frequency change as a function of the ISI suggested a shorter auditory sensory-memory duration in parents of children with SLI than that in parents of normally developing children. In the foregoing, it was already mentioned that in patients with schizophrenia, no normal-size MMN can be obtained irrespective of experimental manipulations, not even with very short SOAs (Javitt et al., 1998).

    • Relation between language, audio-vocal psycholinguistic abilities and P300 in children having specific language impairment

      2011, International Journal of Pediatric Otorhinolaryngology
      Citation Excerpt :

      The results also indicate that with increase of the speed of processing there is improvement of the expressive language age (Table 3), grammatical closure, auditory sequential memory, sound blending and auditory closure and speech discrimination (Table 4). These findings were consistent with other findings in suggesting impaired short-term auditory memory in those children [36]. The insignificant correlation of P300 amplitude and latency with auditory reception and association (Table 4) results disagreed with results of another study which concluded that SLI children have auditory processing deficit, not only in integration, but also in associative ability [37].

    View all citing articles on Scopus
    View full text