UCLA

Languages use combinations of sounds to form words, and some words are only distinguished by a one sound difference (e.g. [d]ank “thank” and [t]ank “tank” in German). Sounds that can induce meaning differences are said to contrast. However, some contrasting sounds can be neutralized in specific positions in the word (in German /d/ becomes [t] at the end of a word, e.g. ra/t/ and ra/d/ both become ra[t]). Within a language, the functional load hypothesis (Jakobson, 1931; Martinet, 1952) states that sounds distinguishing more words (high functional load) are less likely to disappear or merge with other sounds during sound change.

The aim of this dissertation was to examine whether this tendency to preserve highly informative contrasts carries over to the learning of new sound patterns. Additionally, we tested whether learning alternations involving a specific sound pair impacts the perceptual distinctiveness of its members. We used artificial language learning experiments to assess the learnability of alternations involving contrasts with different functional loads in English, and a perceptual discrimination task to assess the perceptual distinctiveness of said contrasts pre- and post- learning.

There are three main results. First, functional load in English predicts the perceptual distinctiveness of contrasts before learning an artificial language. Second, alternations involving high functional load contrasts are learned better, regardless of whether the contrast is neutralized in the artificial language. Finally, learning an artificial language where a contrast is neutralized reduces the perceptual distinctiveness of that contrast post-learning. In summary, we show that phonological contrastiveness in the native language, as well as learning an artificial language, alters perception.