Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks

Statistical language learning with and without awareness

There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.

Poster B107 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port

Soila Kuuluvainen1, Saara Kaskivuo1, Martti Vainio1, Eleonore Smalle2,3, Riikka Möttönen1; 1University of Helsinki, Finland, 2Ghent University, Belgium, 3Tilburg University, The Netherlands

Babies, children and adults use statistical learning (SL), i.e., automatic tracking of transitional probabilities, in detecting boundaries between repeating words (Saffran et al., 1996; Isbilen & Christiansen, 2022). Natural speech has a melody, i.e., prosody, which can also affect SL (Toro et al., 2009; Martinez-Alvarez et al., 2023). Prosody is exaggerated in infant-directed speech compared to adult-directed speech (Ma et al., 2020). Adults typically become aware of learned words in SL experiments using speech sequences without prosodic cues (Batterink et al. 2015). It remains unresolved how prosody (e.g., pitch changes) affects acquisition of implicit or explicit linguistic knowledge from continuous speech streams in adults. We tested, in three online experiments of 30 Finnish-speaking adults each, the effect of pitch (fundamental frequency, F0) changes on SL of trisyllabic word-forms (e.g. lu-vi-ra). In Experiment 1, with familiar prosody, the first syllable was the highest and the last one the lowest in F0, which is a typical pattern in Finnish language. In Experiment 2, the prosody was reversed, the first syllable being lowest and the last one highest, resulting in a structure that was unnatural to the listeners. In Experiment 3, F0 varied randomly in each triplet. All experiments included also neutral speech streams of another set of word-forms with constant F0. After exposure to the 4-minute syllable streams, learning was tested using a two-alternative forced choice (2AFC) task. Awareness of newly learned word-forms was assessed with confidence ratings (from “I guessed” to “I remembered”; see Batterink et al., 2015). In experiment 1, 2AFC accuracy was higher in the familiar prosody than the neutral condition. Familiar prosody specifically enhanced implicit learning of statistical regularities, but had no effect on explicit learning. In Experiments 2 and 3, the unfamiliar rising prosodic structure or random prosodic cues had no effect on accuracy relative to the neutral condition. Our results suggest that familiar pitch changes improve adult SL, and it especially enhances implicit extraction of linguistic patterns from continuous speech without awareness. However, pitch changes that are unfamiliar with respect to the listener’s native language do not affect SL. Since SL has been shown to be associated with increased neural entrainment to repeating words and decreased neural entrainment to syllables (Batterink & Paller, 2017), our current EEG experiments aim to investigate how prosody affects these changes in neural entrainment during SL. In addition, we aim to investigate motor contributions to SL (Assaneo et al., 2019) with and without prosody.

Topic Areas: Language Development/Acquisition, Prosody

SNL Account Login

Forgot Password?
Create an Account