The Efects of Lexical Stress in Visual and Auditory Word Recognition

The aim of the present study was to compare the processing of lexical stress across modalities in native speakers of English. We utilised the same disyllabic stimuli in both visual and auditory grammatical classification. Participants were asked to classify each individually presented word as being either a noun or a verb as quickly and accurately as possible. In line with previous findings (Arciuli & Cupples, 2002; Kelly & Bock, 1988; Kelly, 1988)), the results of our visual task showed that typically stressed words (trochaic nouns and iambic verbs) elicited advantaged processing over atypically stressed words (iambic nouns and trochaic verbs). In contrast, the results of our auditory task showed no overall effect of stress typicality. This finding is also in line with previous research (Davis & Kelly, 1997, Cutler & Clifton, 1984). Interestingly, however, a median split analysis showed that participants with high error rates did exhibit a significant effect of stress typicality during auditory grammatical classification. INTRODUCTION In English, the majority of disyllabic nouns exhibit trochaic stress e.g., “apple”), whereas most disyllabic verbs exhibit iambic stress (e.g., “accept”). This pattern has been investigated by both Sereno (1986) and Kelly and Bock (1988). Sereno sampled 1425 disyllabic nouns and 523 disyllabic verbs and found that 93% of nouns exhibited trochaic stress while 76% of verbs exhibited iambic stress. Kelly and Bock (1988) examined 3,000 disyllabic nouns and 1,000 disyllabic verbs and found a similar pattern. Their analysis showed that 94% of nouns exhibited trochaic stress while 69% of verbs exhibited iambic stress. Thus, trochaic nouns and iambic verbs could be considered to be typically stressed while iambic nouns and trochaic verbs could be considered to be atypically stressed. With regard to visual word recognition, numerous studies carried out by Kelly and colleagues suggest that effects of stress typicality emerge when certain types of ‘off-line’ (unspeeded) tasks are used (e.g., Kelly & Bock, 1988; Kelly, 1988). Recently, we obtained stress typicality effects using ‘online’ naming and lexical decision (Arciuli & Cupples, 2002). Our results showed that, in both tasks, typically stressed words (trochaic nouns and iambic verbs) elicited significantly fewer errors than atypically stressed words (iambic nouns and trochaic verbs). Interestingly, we noted orthographic differences between our typically and atypically stressed words that were not associated with length, neighbourhood or spelling-sound consistency. We carried out a corpus analysis of 340 word-endings and found support for our hypothesis that there are orthographic correlates of both grammatical class and lexical stress in word-endings. In additional experiments we showed that readers are sensitive to these correlates in their processing of nonwords. Our findings led us to wonder whether stress typicality effects could be obtained during spoken word recognition. To our knowledge, there has been no direct comparison of the effects of lexical stress (where stress typicality is defined on the basis of grammatical category) across visual and spoken word recognition using the same stimuli and the same task. There has, however, been some previous research that has specifically addressed the processing of lexical stress in spoken word recognition. Cutler and Clifton (1984) employed an auditory grammatical judgement task where participants heard experimental items in the context of “to...” or “the...” and were asked to decide Proceedings of the 9th Australian International Conference on Speech Science & Technology Melbourne, December 2 to 5, 2002.  Australian Speech Science & Technology Association Inc. Accepted after abstract review page 58 Arciuli et al. The Effects of Lexical Stress whether what they heard was a grammatical phrase. The results did not indicate advantaged processing for typically stressed words. Davis and Kelly (1997) used an auditory grammatical classification task with native and nonnative speakers of English. Their task was different to the one used by Cutler and Clifton in that participants were presented with individual words and asked to classify each one as being either a noun or a verb. The results indicated an effect of stress typicality in non-native speakers. In contrast, there was no significant effect of stress typicality found in the native speaker group. Davis and Kelly went on to suggest that non-native speakers showed an effect where native speakers did not because of task difficulty (an average error rate of 15% for non-native speakers vs. 5% for native speakers). Some of the above mentioned studies provide support for the hypothesis that stress typicality is an orthographic effect in that native speakers have not shown advantaged processing for typically stressed words in the auditory modality. However, Davis and Kelly did find an effect of stress typicality using auditorily presented stimuli with a group of non-native speakers. This finding raises interesting questions about the nature of the stress typicality effect as an orthographic effect and about the possibility of individual differences in the processing of lexical stress (even, perhaps, in native speakers). Our main aim here was to examine the processing of lexical stress in native speakers of English using the same stimuli across both auditory and visual modes of presentation. In addition, we extended the work of Davis and Kelly (1997) by investigating the possibility of individual differences in the processing of lexical stress within a group of native speakers. Specifically, we wondered whether some individuals who find the task particularly difficult might process lexical stress differently from other individuals. These issues were investigated using a grammatical classification task in the following two experiments.