Facilitatory Effects of Vowel Epenthesis on Word Processing in Dutch

Abstract

We report a series of experiments examining the effects on word processing of insertion of an optional epenthetic vowel in word-final consonant clusters in Dutch. Such epenthesis turns film, for instance, into filəm. In a word-reversal task listeners treated words with and without epenthesis alike, as monosyllables, suggesting that the variant forms both activate the same canonical representation, that of a monosyllabic word without epenthesis. In both lexical decision and word spotting, response times to recognize words were significantly faster when epenthesis was present than when the word was presented in its canonical form without epenthesis. It is argued that addition of the epenthetic vowel makes the liquid consonants constituting the first member of a cluster more perceptible; a final phoneme-detection experiment confirmed that this was the case. These findings show that a transformed variant of a word, although it contacts the lexicon via the representation of the canonical form, can be more easily perceptible than that canonical form. © 1999 Academic Press Key Words: word recognition; vowels; epenthesis; Dutch.No one act of spoken-word recognition is exactly like another. Even the same word spoken by the same person to the same listener in the same room may occur in a different context or against different ambient sound. Variability arising from talker differences and environmental conditions has prompted an enormous volume of research in speech perception and has formed one of the principal issues along which models of speech perception and spoken-word recognition divide: whether at some level of processing invariant cues to sounds and words may be abstracted and represented.As if the infinite variability offered by talker and environmental factors were not enough, however, the listener's lot is further complicated by variation as a function of the speech context. The precise form of speech sounds differs as a function of the other sounds that surround them; phonemes are not articulated separately, but are coarticulated with other sounds and vary in accord with the characteristics of their neighbours in the speech signal. This process can alter with speech rate, as well, such that at faster rates of speech coarticulation may result in greater contextual effects or effects which extend across a wider neighborhood; thus it adds another way in which variability complicates spoken-word recognition. And finally, as almost the coup de grace, yet further variation is permitted as a function of certain phonological processes. Thus sounds may assimilate to their phonological context, so that in English, for example, a phrase like hot cakes may be pronounced with either a /t/ or a /k/ at the end of the first syllable.One such form of phonologically determined variation is the epenthesis, or insertion, of isolated sounds where no such sound exists in the underlying form of the word. Epenthesis is a mechanism of historical change in word forms; thus the English word oven, Dutch oven, and German Ofen all have an...

Rule-governed missing information in spoken word recognition: Schwa vowel deletion

“…Van Donselaar, Kuijpers, and Cutler (1999) found that recognition of variant words produced with an optional vowel (vowel epenthesis) was faster than the canonical, schwa-less, pronunciations. Van Donselaar et al suggest that the inclusion of the schwa enhanced perceptibility of the cluster's initial consonant.This could accommodate the present f inding that recognition of deleted tokens was more difficult.…”

“…As our and other recent data make clear, however, this is not always the case. We have discussed two of these studies above (i.e., Utman et al, 2000;van Donselaar et al, 1999). Since there is no longer a ubiquitous advantage to repetition cases per se, the current state of research simply does not permit a clear statement concerning the benefit of repetition, relative to other priming conditions.Clearly, additionalresearch is required to place theoretical constraints on the locus of repetition priming effects.…”

Eight questions about spoken word recognition

“…the realization of intervocalic stops in American English as flaps; Connine, 2004;McLennan et al, 2003), reduction (the deletion of single or multiple segments from words; Ernestus et al, 2002;LoCasto and Connine, 2002;Mitterer and Ernestus, 2006;Utman et al, 2000), and epenthesis (e.g. the insertion of the vowel schwa into the canonical form of a Dutch word; Donselaar et al, 1999). Space restrictions prevent detailed discussion of these phenomena.…”

“…It is worth noting, however, that, as with assimilation, word recognition in the context of such processes entails a combination of phonological knowledge (e.g. knowledge about where epenthesis is legal; Donselaar et al, 1999), and the use of fine-grained phonetic detail to help recover the speaker's intentions (e.g. in the resolution of ambiguities between dernier oignon 'last onion' (with liaison) and dernier rognon 'last kidney' (without liaison); Spinelli et al, 2003).…”

Vowel devoicing and the perception of spoken Japanese words

“…In line with this, the deletion of vowels, even where it occurs regularly in casual speech, makes words harder to recognize: for example, lexical decision responses can be slower for words with deleted vowels ͑e.g., s'maine for semaine; Racine and Grosjean, 2000͒. In contrast, insertion of a vowel into a consonant cluster ͑e.g., fillum for film͒ makes recognition easier, in part because the consonants in the cluster indeed become easier to identify if separated ͑van Donselaar et al, 1999͒. Vowel devoicing in Japanese could thus be perceptually disadvantageous. Further, it could complicate the parsing of continuous speech into its component words.…”