In their first year of life, infants develop from universal listeners, who are able to distinguish most human speech sound contrasts, to language-specific listeners: They form the phoneme categories relevant for their native language, getting better in distinguishing native language speech sound contrasts, but worse in distinguishing most non-native contrasts. In order to become proficient listeners of their native language, not only knowledge of phonemes, but also of phonological processes is necessary. Phonological processes are rules that alter the phonetic realization of phonemes in certain contexts. For example, “n” and “m” are different phonemes that distinguish meaning in English, as in “nap” and “map”. However, a process called place assimilation causes the “n” in “ten” to be pronounced as “m” in the context of “ten men”.

In my PhD thesis, I comparatively investigate the acquisition of such rules in two languages, Dutch and Japanese. I show that statistical cues, i.e. always hearing an “n” when the word “ten” is heard in isolation, but an “m” when “men” is following, help infants learn in which cases they can ignore the difference between “m” and “n”. My goal is to show how and when infants incorporate these statistical cues into their native language phoneme representations.

The first scene of my PhD dance depicts the world how it possibly sounds to an infant immediately after birth. Each dancer represents a different speech sound, which the infant has not weighted for their relevance or assigned any structure to yet. The second scene represents statistical learning processes in the first months of life, in which infants learn to recognize and distinguish the speech sounds of their native language during listening.
The dance then proceeds to illustrate that Dutch and Japanese infants form different phoneme categories after a few months of such exposure to native language.
The following two scenes introduce the next level infants have to master: knowledge of phonological rules. It is not sufficient to know that “m” and “n” are different phonemes, but it is also important to know that this difference has to be ignored in certain cases (otherwise the child would end up thinking “ten” and “tem” were different words, just as “nap” and “map”).
Learning this kind of regularity aided by (amongst others) statistical cues is crucial for infants in order to get highly proficient native language listeners.