«Auditory Processing of Amplitude Envelope Rise Time in Adults Diagnosed With Developmental Dyslexia Elisabeth S. Pasquini and Kathleen H. Corriveau ...»
SCIENTIFIC STUDIES OF READING, 11(3), 259–286
Copyright © 2007, Lawrence Erlbaum Associates, Inc.
Auditory Processing of Amplitude
Envelope Rise Time in Adults Diagnosed
With Developmental Dyslexia
Elisabeth S. Pasquini and Kathleen H. Corriveau
University of Cambridge and Harvard University
University of Cambridge
Studies of basic (nonspeech) auditory processing in adults thought to have developmental dyslexia have yielded a variety of data. Yet there has been little consensus regarding the explanatory value of auditory processing in accounting for reading difficulties. Recently, however, a number of studies of basic auditory processing in children with developmental dyslexia have suggested that a reduced ability to discriminate the rate of change in amplitude envelope onsets (rise time) may be linked to phonological processing difficulties and thereby to reading difficulties. Here, we select a range of different rise-time tasks used with children, and give them to adults with developmental dyslexia, along with 2 other auditory tasks (intensity discrimination and temporal order judgment). Deficits in both rise-time perception and temporal order judgment were found to predict literacy attainment in adults with developmental dyslexia, but the data were suggestive of different causal pathways.
Developmental dyslexia is characterized by difficulties with fluent word recognition and spelling, and it is typically accompanied by a cognitive deficit in the accurate representation of phonology (e.g., Lyon, Shaywitz, & Shaywitz, 2003; Snowling, 2000). Children and adults with developmental dyslexia have problems with reading and spelling that cannot be accounted for by hearing or visual impairments, low intelligence, neurological damage, or poor educational opportunities. Studies seeking a potential sensory cause for the core phonological deficit have investigated a variety of hypotheses. Some researchers have suggested Correspondence should be sent to Usha Goswami, Center for Neuroscience, Faculty of Education, 184 Hills Road, Cambridge CB2 8PQ, United Kingdom. E-mail: email@example.com 260 PASQUINI, CORRIVEAU, GOSWAMI that magnocellular impairment in both the visual and auditory systems causes impaired phonological representation (e.g., Stein & Talcott, 1999). Others have proposed a general sensory processing deficit (Ramus, 2003), a deficit in the cerebellum (Nicolson, Fawcett, & Dean, 1995), or a deficit in discriminating signals from noise (Sperling, Lu, Manis, & Seidenberg, 2005). The hypothesis we focus on here is that the phonological impairments observed in individuals with developmental dyslexia result from lower level auditory processing deficits. Auditory processing deficit accounts of developmental dyslexia are theoretically attractive, as the primary source of language input is usually auditory.
The auditory deficit hypothesis tested in this study is derived from recent studies of basic auditory processing in children with developmental dyslexia. Studies suggested that dyslexic children are relatively insensitive to auditory cues important for processing the prosodic patterns in speech, in particular cues to speech rhythm and stress. Prosodic cues found to be impaired in children with dyslexia include the rate of change of the amplitude envelope at onset (rise time), amplitude modulation (AM) depth, duration, and pitch contour (Foxton, Talcott, & Witton, 2003; Goswami et al., 2002; Muneaux, Ziegler, Truc, Thomson, & Goswami, 2004; Richardson, Thomson, Scott, & Goswami, 2004; Rocheron, Lorenzi, Fullgrabe, & Dumont, 2002). Individual differences in sensitivity to these cues are reliably associated with reading and phonology, even when IQ is controlled for. For example, individual differences in rise-time measures predicted 25% of unique variance in reading and spelling after controlling for age and IQ in the developmental cohort studied by Goswami et al. (2002), whereas individual differences in sensitivity to duration predicted 12% of unique variance in nonword reading in the cohort studied by Richardson et al. (2004). Sensitivity to rise time is also impaired in children with dyslexia across languages. Children with dyslexia learning
both stress-timed (English: Richardson et al., 2004) and syllable-timed (French:
Muneaux et al., 2004; Hungarian: Csépe et al., 2006) languages have difficulties in auditory tasks based on rise time. These difficulties typically characterize the majority of a particular sample (e.g., 63% of children with dyslexia studied by Richardson et al., 2004) and have diagnostic value (e.g., 81% of Hungarian children with dyslexia were identified on the basis of a high threshold in a rise-time task, Csépe et al., 2006).
It is currently unclear whether the auditory processing deficits that characterize children with developmental dyslexia persist into adulthood, whether different auditory processing deficits characterize adult dyslexics, or whether auditory processing improves as a result of maturation or remediation. For example, a recent review of studies of both children and adults (Ramus, 2003, pp. 212–213) suggested that auditory processing deficits are characteristic of, at best, a “fraction” of individuals diagnosed with dyslexia, suggested to be around 39%. Ramus claimed that “[auditory] disorders … have little influence on the development of phonology and reading” (p. 213). However, as many of the studies in his review were of adults
AUDITORY PROCESSING IN DYSLEXIAwith developmental dyslexia, conclusions regarding the influence of auditory processing deficits on the development of phonology and reading are difficult to draw.
In contrast, it has been argued that auditory disorders may have a profound early effect on the development of phonology, and therefore of language and reading, with recent studies demonstrating subtle auditory processing disorders in infants at genetic risk of dyslexia (see Goswami, 2003). The role of auditory maturation also deserves study. For example, some researchers have noted that although both auditory and phonological deficits may characterize children with developmental dyslexia, auditory processing may improve with age, leading to adult developmental dyslexics who show phonological but not auditory deficits (Galaburda, LoTurco, Ramus, Fitch, & Rosen, 2006). These alternative theoretical possibilities make it timely to study sensitivity to the auditory cues that are prominent in the auditory difficulties exhibited by young children in adults with developmental dyslexia.
One of these auditory cues is the rate of change of the amplitude envelope at onset (rise time; see Goswami et al., 2002; Richardson et al., 2004). Only two prior studies of auditory processing in adults with developmental dyslexia have explored rise-time sensitivity. Hämäläinen, Leppänen, Torppa, Müller, and Lyytinen (2005) used a same–different judgment task based on two tones with either identical rise times (both 10 ms) or different rise times (10 ms vs. 30 ms, and 10 ms vs. 80 ms). None of the adult participants could detect the difference between 10-ms and 30-ms rise times (even the normal readers), but a group difference was found in detecting 10-ms versus 80-ms rise times, with significantly poorer performance by the adults with dyslexia. Hämäläinen et al. reported a significant relation between rise-time sensitivity and phonological and reading abilities in their sample, even after controlling for IQ and short-term memory (digit span). In Hämäläinen et al.’s study, rise time contributed 35% of unique variance to phonological skills (rhyme recognition), and contributed 18% of unique variance to a lexical decision task. A second study explored sensitivity to rise time, intensity, and duration cues, but not temporal order judgment (TOJ), in English adults with developmental dyslexia (Thomson, Fryer, Maltby, & Goswami, 2006). The adults with dyslexia were significantly poorer than IQ-matched controls in all the auditory tasks, and sensitivity to rise time and duration (but not intensity) was related to reading and spelling after controlling for both verbal and nonverbal IQ. The adults with developmental dyslexia were also poorer at generating an internal rhythm when asked to continue tapping to a beat determined by a metronome. This rhythm generation task was strongly correlated with performance in the auditory rise-time task only. Thomson et al. suggested that as those adults who found it most difficult to generate an internally consistent rhythm were also those who found it most difficult to detect the primary cue for rhythmic timing in speech (rise time), a supra-modal explanation might lie in P-center detection. The concept of a P-center was introduced by Morton, Marcus, and Frankish (1976) to refer to that moment in an extended auditory event (e.g., a syllable, a musical note) that is the perceptual moment of occurrence.
262 PASQUINI, CORRIVEAU, GOSWAMI When speech is rhythmic or periodically produced, or when any sequence of events is spoken or heard as rhythmically regular, then the P-centers (by definition) occur at regular intervals. In speech, P-centers depend primarily on the rise time associated with the vowel in a syllable. Long onsets before the vowel (e.g., “skate”) move the P-center temporally to the left, whereas long codas (e.g., “banks”) can move it to the right (see Port, 2003). The concept of P-centers was also extended to motor timing by Morton et al., who noted its importance in coordinating auditory and motor rhythms (e.g., in dance). A rise-time perception difficulty may thus contribute to the problems with motor and musical timing noted in some studies of dyslexic individuals (e.g., Wolff, 2002).
Most studies of auditory processing in adults with developmental dyslexia have used other auditory tasks. These include gap detection (Ahissar, Protopapas, Reid, & Merzenich, 2000; McAnally & Stein, 1996), frequency discrimination (Amitay, Ahissar, & Nelken, 2002; Hill, Bailey, Griffiths, & Snowling, 1999), backwards masking (France et al., 2002; Griffiths, Hill, Bailey, & Snowling, 2003), TOJ (Kinsbourne, Rufo, & Gamzu, 1991; Laasonen, Service, & Virsu, 2001), stream segregation (Helenius, Uutela, & Hari, 1999), and tone detection (McAnally & Stein, 1996). However, methodological weaknesses in design complicate interpretation of the group differences that have been reported in many of these auditory processing studies. Some of the studies failed to match IQ between participating dyslexics and controls (e.g., Ahissar et al., 2000; Helenius et al., 1999), and very few studies controlled for individual variation in IQ in the analyses exploring possible relations between the auditory tasks used and reading (of the studies noted previously, only Griffiths et al., 2003, and Kinsbourne et al., 1991, controlled for IQ in such analyses). Yet IQ is a critical variable in studies of auditory processing.
For example, Banai and Ahissar (2004) showed that poor auditory performance (frequency discrimination) is related to low nonverbal IQ for both good and poor readers. Clearly, group matching does not preclude individual variation within groups playing a significant role in any relations between auditory processing and literacy–phonology that may be found.
When studies failing to control for IQ are excluded from consideration, then rather few auditory variables aside from rise time are associated with developmental dyslexia in adults. The most consistent findings concern TOJ, AM, and frequency modulation (FM) at lower rates. TOJ tasks requiring participants to judge the order of sounds that follow each other closely in time are thought to be one of the best measures of the ability to process rapidly presented acoustic information (Tallal & Piercy, 1973), and TOJ has been linked to reading difficulties in a study of children (Tallal, 1980). This has led to a number of investigations of potential relations between TOJ and reading in adults. Kinsbourne et al. (1991) asked adults to judge the order of two sounds delivered to the left versus right ear (one click to each ear). They found that a group of 23 adults with dyslexia performed significantly more poorly than 21 controls matched for both verbal and nonverbal IQ.
AUDITORY PROCESSING IN DYSLEXIAKinsbourne et al. also found significant relations between TOJ and reading and spelling after verbal IQ was controlled for in their sample. Ramus et al. (2003) tested 16 adults, who had formal childhood diagnoses of dyslexia, using a TOJ task based on two sounds easily identifiable as the beeping of a car horn and the barking of a dog. Participants had to judge which sound came first. Significant group differences in TOJ threshold compared to IQ-matched controls were found; however, no relation was found between TOJ and phonology or reading once the data were
corrected for multiple comparisons. Laasonen et al. (2001) gave 16 dyslexic participants and 16 controls, matched for full-scale IQ, TOJ tasks in three modalities:
auditory, visual, and tactile. Participants were asked to decide on the order of two tones, two flashes of light, or two pressure pulses to the left index and middle fingers. A significant group difference was found in the auditory and tactile TOJ tasks. Auditory TOJ was also significantly linked to nonverbal IQ. Once IQ was controlled for, TOJ did not correlate with phonological processing. Therefore, although all of these studies found a group difference in TOJ, only Kinsbourne et al.
(1991) found a relation between TOJ and literacy.
However, Griffiths et al. (2003) failed to find group differences in temporal order detection in an extensive study of adults with dyslexia and IQ-matched controls. Their TOJ task consisted of four pairs of tones, the second or third of which reversed the standard high–low order. Participants were asked to identify the low–high tone pair. The dyslexic group did not differ significantly from the control group in either 20-ms inter–stimulus interval (ISI) or 200-ms ISI conditions, although both groups found the 20-ms condition (rapid presentation) more difficult.
Partial correlations controlling for vocabulary showed a significant relation between a composite phonology variable and the 20-ms measure for both groups, but no relation with literacy. In fact, the question of whether rapid auditory processing plays any role at all in literacy development remains a highly contentious issue (see Marshall, Snowling, & Bailey, 2001; McArthur & Bishop, 2001; Rosen, 2003;