Predictive processing during novel word learning: ERP measures of vowel harmony

Abstract

This project investigated the effect of Turkish vowel harmony (TVH) on the acquisition of multi-syllabic words via cross-situational word learning (CSWL). CSWL is an implicit learning paradigm requiring learners to statistically track word-referent pairs across ambiguous conditions. Although there are many studies have examined certain lexical conditions that span word boundaries, within-lexical unit characteristics, such as TVH, have not been examined. We conducted 2 experiments; Experiment 1 is an internet-based experiment with Native Turkish speakers and speakers of other languages, and Experiment 2 is an EEG experiment focusing on N400 and LPC ERPs with Native Turkish speakers. In experiment 1, participants received training and recognition test phases. During training, participants were presented with novel pseudowords, both harmonious and disharmonious for TVH, with multiple referent pictures. In the recognition test, participants heard a pseudoword and were asked to select the correct referent for it. The results of the first experiment did not show a significant difference. In experiment 2, learning was additionally measured by a semantic judgment task (SJT). In the SJT, participants were presented with either a learned pseudoword or a real Turkish word, then with another real Turkish word. Their task was to indicate whether the two presented words were from the same semantic category. The recognition test results showed high accuracy, implying successful learning, with no differences among TVH conditions. The SJT results were modulated by both the novelty of the words (pseudowords vs. real words) and the harmony conditions in both behavioral and ERP measures. Specifically, harmonious words lead to N400 responses similar to real words, with no such effect on the disharmonious words. These results suggest that the phonological rules of participants’ native language had an impact on semantic consolidation.