Acoustic reduction for repeated words could be the result of articulation and motor practice (Lam & Watson, 2014), facilitated production (Kahn & Arnold, 2015; Gahl et al., 2012), or audience design and shared common ground (Galati & Brennan, 2010). We sought to narrow down what kind of facilitation leads to repetition reduction. Repetition could, in principle, facilitate production on a conceptual, lexical, phonological, articulatory, or acoustic level (Kahn & Arnold, 2015). We compared the durations of the second utterance of a target word when the initial production was aloud or silent. The silent presentation either involved unmouthed or mouthed inner speech. Overt production, unmouthed and mouthed inner speech all led to reduction in target word onsets, but target word durations were only shortened when a word was initially said aloud. In an additional experiment, we found that prior naming of a homophone of the target word also led to duration reduction. The results suggest that repetition reduction occurs when there is a recently experienced auditory memory of the item. We propose that duration may be controlled in part by auditory feedback during production, the use of which can be primed by recent auditory experience.

No abstract

Some accounts of acoustic reduction propose that variation in word duration is a reflection of the speaker’s internal production processes, but it is unclear why lengthening within a word benefits planning. The present study examines whether variability in word length is partly attributable to phonological encoding. In an event description task, speakers produced words with longer durations when the word shared part of its phonology with a previously articulated word than when it did not. More importantly, lengthening was greater when the overlap was word-initial than when it was word-final. These differences in duration are in line with predictions of serial phonological competition models, which claim that words that overlap in onsets create more competition than words that overlap in offsets and are thus more difficult to produce. That word duration is sensitive to differences in production difficulty suggests a link between speakers’ duration choices and phonological encoding. We propose that lengthening provides the production system with the necessary processing time to produce a word’s sounds.

No abstract

It has been claimed that English has a metrical structure, or rhythm, in which stressed and unstressed syllables alternate. In previous research regular, alternating patterns have been shown to facilitate online language comprehension. Expanding these findings to downstream processing would lead to the prediction that metrical regularity enhances memory. Research from the memory literature, however, indicates that regular patterns are less salient and therefore less well remembered, and also that strings of similar sounds are harder to remember. This work suggests that, like lists of words with similar sounds, lists of words with similar metrical patterns are less accurately remembered than comparable metrically irregular patterns. The present study tests these conflicting predictions by examining the effects of metrical regularity in a recall task. We find that words are better recalled when they do not match their metrical context, suggesting that a regular metrical structure may not be beneficial in all contexts.