Within-word serial order control: Adjacent mora exchange and serial position effects in repeated single-word production

Abstract

An essential function of language processing is serial order control. Computational models of serial ordering and empirical data suggest that plan representations for ordered output of sound are governed by principles related to similarity. Among these principles, the temporal distance and edge principles at a within-word level have not been empirically demonstrated separately from other principles. Specifically, the temporal distance principle assumes that phonemes that are in the same word and thus temporally close are represented similarly. This principle would manifest as phoneme movement errors within the same word. However, such errors are rarely observed in English, likely reflecting stronger effects of syllabic constraints (i.e., phonemes in different positions within the syllable are distinctly represented). The edge principle assumes that the edges of a sequence are represented distinctly from other elements/positions. This principle has been repeatedly observed as a serial position effect in the context of phonological short-term memory. However, it has not been demonstrated in single-word production. This study provides direct evidence for the two abovementioned principles by using a speech-error induction technique to show the exchange of adjacent morae and serial position effects in Japanese four-mora words. Participants repeatedly produced a target word or nonword, immediately after hearing an aurally presented distractor word. The phonologically similar distractor words, which were created by exchanging adjacent morae in the target, induced adjacent-mora-exchange errors, demonstrating the within-word temporal distance principle. There was also a serial position effect in error rates, such that errors were mostly induced at the middle positions within a word. The results provide empirical evidence for the temporal distance and edge principles in within-word serial order control.

Across the great divide: Proximate units at the lexical‐phonological interface

“…Thus, syllables are selectable units in Chinese, but not in English. In Japanese, Kubozono (; see also Nakayama & Saito, ) showed that morae, phonological units intermediate in granularity to syllables and phonemes and comprising one or two phonemic segments (see Kureta, Fushimi, Sakuma, & Tatsumi, and Roelofs, in this issue, for additional details concerning morae), as well as phonemes slip. This is consistent with the claim that morae are primary phonological units in Japanese (Kureta, Fushimi, Sakuma, & Tatsumi, ; Kureta, Fushimi, & Tatsumi, ).…”

Long-term phonological knowledge supports serial ordering in working memory.

“…One possible explanation for the difference in effects is the difference in measurement (i.e., speech error vs. STM task performance). We nevertheless suspect that measurement is not the main cause of the difference in the effects, because accumulating evidence suggests common serial ordering mechanisms for speech production and serial order memory (e.g., Acheson & MacDonald, 2009; Nakayama & Saito, 2014). A plausible explanation may be the extent of training.…”

Language generality in phonological encoding: Moving beyond Indo‐European languages

“…While this error is akin to a syllable exchange, others involve a CV exchange with a nasal coda, as in /da.teN saN no ha.ra/ → da.teN sa.ra no ha.ra 'Hara with three hitting points' ( Terao, 2022 ), which cannot be a syllable error. Such cases have been attributed to the mora in Japanese ( Kubozono, 1989;Nakayama & Saito, 2014;Saito & Inoue, 2017 ), suggesting that it is a selectable unit in this language. Furthermore, parallelling the differential pattern found by Chen et al ( 2002 ) and O'Séaghdha et al ( 2010) with Mandarin syllables, form preparation experiments with Japanese participants found that initial shared onsets do not reduce latencies but initial CV sequences do ( Kureta et al, 2006 ).…”

“…ra no ha.ra ‘Hara with three hitting points’ (Terao, 2022), which cannot be a syllable error. Such cases have been attributed to the mora in Japanese (Kubozono, 1989; Nakayama & Saito, 2014; Saito & Inoue, 2017), suggesting that it is a selectable unit in this language. Furthermore, parallelling the differential pattern found by Chen et al.…”