Learning to speak by listening: Transfer of phonotactics from perception to production

Abstract

The language production and perception systems rapidly learn novel phonotactic constraints. In production, for example, producing syllables in which /f/ is restricted to onset position (e.g. as /h/ is in English) causes one's speech errors to mirror that restriction. We asked whether or not perceptual experience of a novel phonotactic distribution transfers to production. In three experiments, participants alternated hearing and producing strings of syllables. In the same condition, the production and perception trials followed identical phonotactics (e.g. /f/ is onset). In the opposite condition, they followed reverse constraints (e.g. /f/ is onset for production, but /f/ is coda for perception). The tendency for speech errors to follow the production constraint was diluted when the opposite pattern was present on perception trials, thus demonstrating transfer of learning from perception to production. Transfer only occurred for perceptual tasks that may involve internal production, including an error monitoring task, which we argue engages production via prediction. Keywordsphonotactics; transfer of learning; speech errors; language perception; language production; implicit learning Speakers learn to speak by listening. But how do acts of speech perception lead to change within the production system? Our ability to speak depends on the acquisition of general patterns such as the fact that, in English, adjectives precede nouns or that one says "an" before words beginning with vowels. This paper is concerned with perception-to-production transfer of a specific kind of generalization, phonotactic constraints. Phonotactics are constraints about the ordering of segments, typically within syllables. They are language specific and hence must be learned. For example, in English, /h/ must be a syllable onset Correspondence to: Audrey K. Kittredge, audreyk@andrew.cmu.edu. Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. (occur at the beginning of a syllable, e.g. /hum/) and /ng/ must be a syllable coda (occur at the end of a syllable, e.g. /song/). In Persian, though, /h/ can be a coda (e.g. /dah/ "ten"), and in Vietnamese, /ng/ can be an onset (e.g. /ngei/ "day"). HHS Public AccessKnowledge of native-language phonotactics emerges in infancy (e.g. Jusczyk et al., 1993) and, throughout life, constrains language perception and production (e.g. Pitt, 1988). Production models assume that phonotactic constraints are consulted during the encoding of word forms, particularly during the construction of syllables (e.g. Levelt, Roelofs, & Meyer, 1999). Evidence for this assumption c...

The role of consolidation in learning context-dependent phonotactic patterns in speech and digital sequence production

“…The selected first-order studies tested, in a single-session, the learning of rules involving similar consonants and trial structure as the second-order studies, but lacking the vowel-contingency. Eight second-order and 11 first-order studies met these criteria (Tables S1 and S2) (11,(17)(18)(19)(26)(27)(28)(29)(30). The extent to which slips of restricted consonants maintained their syllable positions (%restricted) was compared with the extent to which the unrestricted control consonants maintained their syllable positions (%unrestricted); the difference between these values served as the index of learning in each study.…”

Same, different, or closely related: What is the relationship between language production and comprehension?

“…Four papers concern language learning in adults, specifically the way comprehension affects production or vice versa. Kittredge and Dell (2016) report three experiments on the acquisition of novel phonotactic constraints (e.g., ''/f/ appears only as a syllable onset, not as a coda") in a speech production task. They examined whether experiencing the same or an opposing constraint (e.g., ''/f/ appears only as a coda") in another person's speech affected the participants' speech output (specifically the types of errors committed).…”

Occluding the face diminishes the conceptual accessibility of an animate agent

“…But the fact that we found similar results in Study 3, a comprehension task where lexical access was not a factor, as the sentences were given to the participants, supports the interpretation that people construe Body-Agent and Hand-Agent events in different ways. Although the relation between representations underlying production vs. comprehension is a topic of long-standing debate, few researchers argue that these representations are strictly separate (see Kittredge & Dell, 2016;Meyer, Huettig, & Levelt, 2016, for review). More importantly for the current study, conceptual representations of events serve as input to both language production and language comprehension, and production findings (here, the likelihood of mentioning the agent vs. patient as Subject) and comprehension findings (here, different interpretations for agent vs. patient Subjects) both reflect how people conceptualise events.…”