Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Prosodic cues support acquisition of adjacent and nonadjacent regularities from continuous speech
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 B75 in Poster Session B, Tuesday, October 24, 3:30 - 5:15 pm CEST, Espace Vieux-Port
Saara Kaskivuo1, Soila Kuuluvainen1, Martti Vainio1, Eleonore Smalle2,3, Riikka Möttönen1; 1University of Helsinki, Finland, 2Ghent University, Belgium, 3Tilburg University, The Netherlands
In language acquisition, learners have to extract salient units from a continuous stream of speech. These units may be adjacent (e.g., in word-level segmentation) or nonadjacent (e.g., in syntax, the present continuous tense is _ing forms a nonadjacent regularity). Previous research suggests that regularities are tracked at both adjacent and nonadjacent levels (“statistical learning”), and familiar words also show word-level neural entrainment (Batterink & Paller, 2017; Smalle et al., 2022). Perceptual cues, such as pauses, prosodic cues or phonological similarities, facilitate learning; with nonadjacencies, they may be a prerequisite (Isbilen & Christiansen, 2022; Martinez-Alvarez et al., 2023). Here, we investigated how natural prosodic cues (i.e., pitch changes) in speech streams affect statistical learning of adjacent and nonadjacent regularities. In a series of online experiments, Finnish native participants (n=147) passively listened to structured speech streams containing meaningless CVCVCV triplets with both adjacent (“ABC”) and nonadjacent (“AxC”) regularities. The triplets matched Finnish phonotactics and had a phonologically balanced syllable distribution, corresponding to naturalistic speech. The experiments differed in prosody, which was either flat or had a “high-neutral-low” pitch structure for each triplet, typical for pragmatic emphasis in Finnish speech. Thus, without pitch cues, learning the triplets was only possible based on transitional probabilities between syllables, whereas the prosody structure also included pitch cues. A two-alternative forced-choice (2AFC) task probed learning of adjacent and nonadjacent regularities. To tease apart the learning of adjacent and nonadjacent regularities, nonadjacent regularities were presented in a triplet containing a novel middle syllable (“AyC”) in the 2AFC task. They were pitted against foils with two familiar syllables and one novel syllable with a syllable transitional probability of 0 in the stream. Without prosody, participants’ performance was above chance with adjacent regularities (p < .001) but not with nonadjacent regularities. With perceptual cues, i.e., prosody, participants’ accuracy was above chance on both adjacent and nonadjacent regularities (p < .001). Prosody facilitated learning with both types of adjacencies (p < .001). Our results indicate that natural pitch changes in continuous speech support learning of statistical regularities. Our future studies will investigate how prosodic cues affect neural tracking of speech streams and brain mechanisms that support the acquisition of adjacent and nonadjacent regularities.
Topic Areas: Speech Perception, Language Development/Acquisition