Does lexical stress or metrical stress better predict word boundaries in Dutch?

Abstract

For both human and automatic speech recognizers, it is difficult to segment continuous speech into discrete units such as words. Word segmentation is so hard because there seem to be no self-evident cues for word boundaries in the speech stream. However, it has been suggested that English listeners can profit from the occurrence of full vowels (i.e. vowels with metrical stress) in the speech stream to make a first good guess about the location of word boundaries. The CELEX database study described in this paper investigates whether such a strategy is also feasible for Dutch, and whether the occurrence of full vowels or the occurrence of vowels with primary word stress (i.e. vowels with lexical stress) is a better cue for word boundaries. The CELEX counts suggest that, for Dutch, metrical stress seems to be a better predictor of word boundaries than lexical stress.