The recognition of foreign words at first exposure in early language development: The role of phonological similarity

• In Event: Poster Session 02
  In Poster Session: PS 02 Topic - Language, Communication

Abstract

When learning new words in their native language, children take advantage of their existing knowledge about semantic (Tomasello, Mannle & Werdenschlag, 1988) and phonological properties (Newman, Samuelson & Gupta, 2008). Children may also use similarities between known words in their native language to help learn and recognize similar sounding words in a foreign language (e.g. English - baby, German - Baby; Von Holzen, Fennell, & Mani, 2019). We investigate how much phonological similarity is needed between the native and foreign words for recognition to occur (Experiment 1) and whether this recognition is constrained by the word’s location within an utterance (Endress & Hauser, 2010; Experiment 2).

Using the Intermodal Preferential Looking Paradigm (Golinkoff et al., 1987), we tested 3- to 6-year-old children (Experiment 1 n = 29/32; Experiment 2 n = 23/32) on their word recognition in English (their native language) and German (an unfamiliar foreign language). The final dataset will also include 14- and 24-month-old infants. 24 English words and their German translation equivalents were selected; half of the words were cognates. Word-pair phonological similarity was quantified using an adaptation of Levenshtein distance (Levenshtein, 1966) that takes into account sub-segmental features (Phonological Similarity). Target words were presented in the final position of the carrier phrase in Experiment 1 and in the phrase-medial position in Experiment 2 and were recorded by a female American English-German bilingual speaker. An increase in the mean proportion of target looks (logit transformed) from the period before presentation of the target word (pre-naming) to the period after (post-naming) is interpreted as recognition. For each experiment, target looks were modeled using a linear mixed-effects model with the fixed effects of Naming (pre-, post-naming), Language (English, German), and Phonological Similarity of the word pairs. Estimated marginal means were used to determine at which value of Phonological Similarity recognition occurred in German.

A summary of the results can be found in Figure 1. The models revealed that the critical interaction between Naming, Phonological Similarity, and Language was significant in both Experiment 1 (β = -1.19, SE = 0.47, p = 0.01) and Experiment 2 (β = -1.81, SE = 0.6, p = 0). Figure 1 shows that for both Experiments, the recognition effect was greater in English, although in German the size of this effect increased as Phonological Similarity increased. Recognition was only found for words deemed cognates (Phonological Similarity of .50 or greater) in both Experiment 1 (β = -0.96, SE = 0.15, p < .001) and Experiment 2 (β = -0.9, SE = 0.19, p < .001), but not for non-cognates (Experiment 1: β = 0.22, SE = 0.15, p = 0.45; Experiment 2: β = 0.05, SE = 0.2, p = 0.99). Children are able to recognize words from a foreign language when those words are phonologically similar to words they know in their native language and this is not constrained by the word’s location within an utterance. The final dataset will include younger children (14 and 24 months) to reveal whether this ability is modulated by the child’s age.

Authors

• Katie Von Holzen, TU Braunschweig
  Presenting Author

• Rochelle Newman, University of Maryland - College Park
  Non-Presenting Author