Prosodic cues for morphological complexity in Dutch and English

Abstract

Previous work has shown that Dutch listeners use prosodic information in the speech signal to optimise morphological processing: Listeners are sensitive to prosodic differences between a noun stem realised in isolation and a noun stem realised as part of a plural form (in which the stem is followed by an unstressed syllable). The present study, employing a lexical decision task, provides an additional demonstration of listeners’ sensitivity to prosodic cues in the stem. This sensitivity is shown for two languages that differ in morphological productivity: Dutch and English. The degree of morphological productivity does not correlate with listeners’ sensitivity to prosodic cues in the stem, but it is reflected in differential sensitivities to the word-specific log odds ratio of encountering an unshortened stem (i.e., a stem in isolation) versus encountering a shortened stem (i.e., a stem followed by a suffix consisting of one or more unstressed syllables). In addition to being sensitive to the prosodic cues themselves, listeners are also sensitive to the probabilities of occurrence of these prosodic cues.

Acknowledgments

Part of this work has been made possible by the support of a Major Collaborative Research Initiative (MCRI) grant of the Social Sciences and Humanities Research Council of Canada awarded to Gary Libben.

Notes

¹This word-pseudoword matching in our materials was not perfect: We failed to match for the status of the coda for two Dutch items, we failed to match for the status of the onset for one Dutch item, and we failed to match for the length of the vowel for one Dutch item. For one English item, we failed to match for the status of the coda.

²In American English, a stem-final /t/ typically becomes flapped in intervocalic position. Our speaker retained the non-flapped pronunciation in intervocalic position, which may be considered overly careful speech. Note, however, that the presence of unflapped stimuli in our experiment should work against our effect, as the unflapped /t/ in the constructed stem might be considered a strong cue for the monosyllabic form.

³Here and in the following analyses, reaction times were logarithmically transformed in order to normalise their distribution.

⁴As reaction times were measured from word offset, we expect a facilitatory effect of Duration: At word offset, the listener has been exposed to more information when the duration of the word is long than when the duration of the word is short, facilitating the response. In order to establish whether Stem Type has an effect independently of Duration (normal stems have longer durations than constructed stems), we included Duration as a covariate in our analyses.

⁵In our multi-level covariance models, subject variability is accounted for by using subject as a grouping factor. In the analyses of word data exclusively, item variability is accounted for by including item-specific covariates in the regression model. However, in all our analyses involving both word and pseudoword data, and in all analyses involving pseudoword data exclusively, item variability has not been accounted for, as no item-specific covariates are available for pseudowords. Therefore, in all analyses involving pseudowords, Stem Type has been treated as a between-items factor even though we would have liked to treat it as a within-items factor. Nevertheless, even without the extra power of the within-items analysis, we obtained very robust effects of Stem Type. Furthermore, an analysis on Dutch and English words and pseudowords with item as the grouping factor yielded largely the same pattern of results as the analysis with subject as the grouping factor: Stem Type, F(1, 203) = 117.3, p < 0001; Word Status, F(1, 203) = 130.2, p < .0001; Duration, t(203)=−5.2, p < .0001; Language, F(1, 203) = 138.4, p < .0001; Stem Type by Duration, t(203)=2.9, p < .01. The interaction of Word Status by Language was not significant in this analysis, F(1, 5149) = 4.2, p = .67.

⁶The correlation coefficients reported here are calculated over the items that remained after removing the items with high error percentages, and are therefore numerically different from the correlation coefficients reported in the Materials section (which were calculated over all items that were presented to the participants).