Elsevier

Cognition

Volume 102, Issue 3, March 2007, Pages 341-360
Cognition

Reading spoken words: Orthographic effects in auditory priming

https://doi.org/10.1016/j.cognition.2006.01.001Get rights and content

Abstract

Three experiments examined the involvement of orthography in spoken word processing using a task – unimodal auditory priming with offset overlap – taken to reflect activation of prelexical representations. Two types of prime–target relationship were compared; both involved phonological overlap, but only one had a strong orthographic overlap (e.g., dream-gleam vs. scheme-gleam). In Experiment 1, which used lexical decision, phonological overlap facilitated target responses in comparison with an unrelated condition (e.g., stove-gleam). More importantly, facilitation was modulated by degree of orthographic overlap. Experiment 2 employed the same design as Experiment 1, but with a modified procedure aimed at eliciting swifter responses. Again, the phonological priming effect was sensitive to the degree of orthographic overlap between prime and target. Finally, to test whether this orthographic boost was caused by congruency between response type and valence of the prime–target overlap, Experiment 3 used a pseudoword detection task, in which participants responded “yes” to novel words and “no” to known words. Once again phonological priming was observed, with a significant boost in the orthographic overlap condition. These results indicate a surprising level of orthographic involvement in speech perception, and provide clear evidence for mandatory orthographic activation during spoken word recognition.

Introduction

Although linguistic information is often conveyed within one modality, form representations spanning multiple modalities are often implicated in the processing of a stimulus. In particular, there is a clear involvement of phonology in visual word recognition. Initial findings using the naming paradigm showed that latencies are shorter for words like MINT, which has a regular spelling-to-sound mapping, than for words like PINT, which does not (Baron & Strawson, 1976). Results like this might be explained in terms of phonological activation only at the output level, but grapheme-phoneme relations also influence tasks that do not require speech production, such as lexical decision, perceptual identification, letter detection, semantic categorization and masked priming (see Perfetti, 1999, for a review). Furthermore, both grapho-phonological and phono-orthographic inconsistencies seem to play a role in such visual tasks (e.g., Stone et al., 1997, Ziegler et al., 1997; although see Peereman, Content, & Bonin, 1998).

Amongst the theories accounting for phonological involvement in reading, many interactive models assume bi-directional connections between orthographic and phonological representations at various processing levels (e.g., Frost and Katz, 1989, Grainger and Ferrand, 1996, Plaut et al., 1996, Seidenberg and McClelland, 1989, Ziegler et al., 2003). Such a view of reading is strongly suggestive of a symmetry between orthographic and phonological systems that would extend to speech perception, implying significant involvement of orthography in spoken word recognition. Nonetheless, there may be separate reasons for thinking that the interplay between the two domains is asymmetrical. In developmental terms the asymmetry is clear: phonology is acquired prior to and independently of spelling, and may be more reliant on innate structures. There are obvious economies to be made in terms of reusing speech representations when learning to read, but the converse is not true. The architecture of the visual and spoken word recognition systems in adult literates may thus reflect the consequences of this developmental asymmetry.

There is substantially less evidence available implicating spelling in auditory word processing, particularly in situations where no orthographic information is explicitly provided. Seidenberg and Tanenhaus (1979) showed that rhyme judgments for spoken words were delayed when the rhyming stimuli were orthographically dissimilar, compared with when they were not (e.g., pie-rye vs. pie-tie). Other studies have shown similar effects, again using metaphonological tasks such as phoneme monitoring (e.g., Dijkstra et al., 1995, Hallé et al., 2000) or word “blending” (Ventura, Kolinsky, Brito-Mendes, & Morais, 2001). In these tasks, listeners must explicitly decompose spoken words into sublexical units. Given that awareness of sublexical speech units is strongly related to literacy (e.g., Morais et al., 1979, Olson, 1996), it is possible that the orthographic codes involved in these conscious judgments are not involved in normal spoken word recognition. A similar argument applies to the recent demonstration by Muneaux and Ziegler (2004) that spelling influences offline judgements in a phonological neighbour generation task.

A few studies go further, signifying an automatic and significant role of orthography in spoken word recognition. Lexical decisions to spoken words appear to be sensitive to orthographic properties. Auditory lexical decisions and shadowing are relatively slow in cases where the rimes of words can be spelled in several ways (Ventura et al., 2004, Ziegler and Ferrand, 1998). Recent results also demonstrated a facilitatory effect of dense orthographic neighborhood in auditory lexical decisions and word shadowing (Ziegler et al., 2003). As these studies did not use metaphonological tasks, their effects can be interpreted as reflecting automatic involvement of orthography in spoken word recognition.

Yet without converging evidence from other paradigms, the contribution of orthography in speech recognition processes must remain questionable. Intramodal visual priming paradigms have been widely used as an “acid test” of phonological involvement in visual word processes, but little is known about the converse situation: the role of orthographic information in intramodal auditory priming. In the auditory case, participants hear a prime followed by a target, and are usually asked to repeat the target (shadowing) or respond on the basis of its lexical status. Responses in cases where prime and target overlap phonemically are compared with an unrelated control condition. Initial overlap (e.g., musk-mulch) produces delayed responses, supposed to reflect lexical competition mechanisms (Slowiaczek & Hamburger, 1992). On the contrary, a final overlap (e.g., dream-gleam) facilitates responses (e.g., Dumay et al., 2001). The latter effect is assumed to reflect automatic activation of sublexical units because it is restricted to the auditory modality (e.g., Dumay et al., 2001), does not depend on the lexical status of the prime (e.g., Monsell & Hirsh, 1998) or the lexical status of the overlapping sequence (Dumay, 2002), and is insensitive to prime–target relative frequency (Radeau, Morais, & Segui, 1995).

Slowiaczek, Soltano, Wieting, and Bishop (2003) found facilitation when auditory prime and target had either orthographic, or phonological and orthographic initial overlap (first syllable) in lexical decision (cf. Jakimik, Cole, & Rudnicky, 1985) and shadowing tasks, suggesting that orthographic information is automatically involved in spoken word processing. However, the locus of effects for multiple-phoneme initial overlap is the subject of some debate. Such effects are often assumed to reflect lexical competition, but Goldinger, 1999, Pitt and Shoaf, 2002 argue that they reflect response biases, even when related trial proportion and interstimuli interval (ISI) are reduced. It is therefore difficult to isolate the role of orthography in speech perception using this kind of overlap.

On the other hand, Miller and Swick (2003) examined orthographic effects in final overlap phonological priming for two pure alexic patients (i.e., with severely impaired word reading, but relatively intact writing skills), along with age-matched controls. Their study compared priming of lexical decision for spoken-word pairs varying in phonological and orthographic similarity. Concentrating on the results for the normal participants, orthographic overlap in the absence of strong phonological overlap (e.g., deaf-leaf) did not facilitate responses to the targets. However, pairs that overlapped both orthographically and phonologically (e.g., barn-yarn) produced an increased priming effect, compared to pairs overlapping on form alone (e.g., cloud-crowd). This is a promising result, but again does not constitute good evidence for fast, mandatory access to orthographic information in speech perception. Miller and Swick’s study made use of a long ISI (over 800 ms) and did not include the kind of foil trials that have been demonstrated to be necessary for excluding anticipation strategies (Norris, McQueen, & Cutler, 2002). These factors, allied with a high orthographic relatedness proportion (50% of the trials with a word target, although only half of these were related both phonologically and orthographically) and relatively slow response times (over 1000 ms in the control condition) meant that participants had an excellent opportunity to develop anticipation strategies that could have caused the orthographic effect.

The primary aim of the present study was to examine whether there is a swift, automatic participation of orthography in speech perception. For this purpose, we used intramodal phonological priming, with responses based on the lexical status of the target. We focused on final overlap because priming in this case is thought to be mainly due to residual activation of prelexical representations. In order to avoid spurious results based on the strategic processing of the primes and targets (cf. Goldinger et al., 1992, Slowiaczek and Hamburger, 1992, Slowiaczek and Hamburger, 1993, Posner and Snyder, 1975), Experiment 1 made use of a short ISI (20 ms), a low prime–target relatedness proportion, and included filler items intended to counteract such strategies (Norris et al., 2002). Potential strategic accounts of the priming effects were further addressed in Experiments 2 and 3, by requiring participants to respond more quickly (Experiment 2) and by employing a reversed lexical decision in which participants are asked to respond positively to nonwords (Experiment 3). The critical issue in all three experiments was whether the orthographic similarity of the overlap between prime and target would modulate any intramodal auditory priming effect observed on lexical decision latencies.

Section snippets

Participants

Thirty-six University of York students participated for course credit. None reported any hearing or language impairment.

Materials and design

All stimuli were spoken monosyllables (see Appendix A). Primes were English words, whereas half the targets were words and half were pseudowords, matched in CV structure with their primes (24 CVVC, 17 CCVVC, 3 CVVCC, 2 CVCC, 1 CCVC and 1 CVCC). Forty-eight test target words (mean frequency 41 per million and mean duration 560 ms) were each associated with 3 primes. In the two

Participants

Forty-two University of York students participated for course credit. None reported any hearing or language impairment.

Materials, design and procedure

The stimuli, general design and procedure were essentially the same as for Experiment 1, with the exception of three noteworthy changes made to encourage faster responses. First, a mild (17%) pitch-preserving temporal compression was applied to the auditory stimuli of Experiment 1 using Adobe Audition (Adobe Systems Inc). This had no noticeable effect on the comprehensibility

Participants

Thirty-six students from the same population as in Experiment 1 participated in this experiment for course credit.

Materials, design and procedure

Materials and design were the same as in Experiment 1. Participants had to decide whether the second stimulus of each pair was a made-up word or not, and press the “yes” button if they thought it was a made-up word, or “no” if they thought the stimulus was not. This instruction was chosen to provide as positive a valence as possible for the pseudowords (e.g., they were described as

General discussion

Like previous studies, our experiments have shown that responses to a spoken word are facilitated by a phonological rime overlap with a preceding spoken prime. The key finding of our research, however, is that final-overlap auditory priming is also sensitive to orthographic similarity between prime and target. This effect was found in three experiments, using normal, speeded and reversed lexical decisions. Final-overlap auditory priming appears to arise from much more than just the residual

Acknowledgements

The work reported herein was supported by a postdoctoral research grant from the Fyssen Fondation (France) to Céline Chéreau, and a grant from the UK Medical Research Council (G0000071) to Gareth Gaskell. We thank Rebecca Larkin for her voice, and Peter Bailey for advice on speech compression algorithms.

References (51)

  • J.C. Ziegler et al.

    The feedback consistency effect in lexical decision and naming

    Journal of Memory and Language

    (1997)
  • J.C. Ziegler et al.

    Neighborhood effects in auditory word recognition: Phonological competition and orthographic facilitation

    Journal of Memory and Language

    (2003)
  • J. Baron et al.

    Use of orthographic and word-specific knowledge in reading words aloud

    Journal of Experimental Psychology: Human Perception and Performance

    (1976)
  • T. Dijkstra et al.

    Orthographic effects on phoneme monitoring

    Canadian Journal of Experimental Psychology

    (1995)
  • Dumay, N. (2002). Is the syllable an intermediate representation in French speech processing? Evidence from form...
  • N. Dumay et al.

    Behavioral and electrophysiological study of phonological priming between bisyllabic spoken words

    Journal of Cognitive Neuroscience

    (2001)
  • K.I. Forster

    Priming and the effects of sentence and lexical contexts on naming time: Evidence for autonomous lexical processing

    Quarterly Journal of Experimental Psychology

    (1981)
  • K.I. Forster et al.

    A Windows display program with millisecond accuracy

    Behavior Research Methods Instruments & Computers

    (2003)
  • R. Frost et al.

    Orthographic depth and the interaction of visual and auditory processing in word recognition

    Memory and Cognition

    (1989)
  • M.G. Gaskell

    Modeling lexical effects on semantic categorization and semantic effects on word recognition

    Behavioral and Brain Sciences

    (2000)
  • M.G. Gaskell et al.

    Integrating form and meaning: A distributed model of speech perception

    Language and Cognitive Processes

    (1997)
  • M.G. Gaskell et al.

    Representation and competition in the perception of spoken words

    Cognitive Psychology

    (2002)
  • S.D. Goldinger

    Only the shadower knows: Comments on Hamburger and Slowiaczek (1996)

    Psychonomic Bulletin and Review

    (1999)
  • S.D. Goldinger et al.

    Form-based priming in spoken-word recognition: The roles of competition and bias

    Journal of Experimental Psychology: Learning, Memory and Cognition

    (1992)
  • D. Holender

    Expectancy effects, congruity effects, and the interpretation of response latency measurement

  • Cited by (0)

    This manuscript was accepted under the editorship of Jacques Mehler.

    1

    Nicolas Dumay is now at the University of Bristol.

    View full text