Abstract
This study investigates the hypothesis that tone alternation directionality becomes a basis of structural bias for tone alternation learning, where “structural bias” refers to a tendency to prefer uni-directional tone deletions to bi-directional ones. Two experiments were conducted. In the first, Mandarin speakers learned three artificial languages, with bi-directional tone deletions, uni-directional, left-dominant deletions, and uni-directional, right-dominant deletions, respectively. The results showed a learning bias toward uni-directional, right-dominant patterns. As Mandarin tone sandhi is right-dominant while Cantonese tone change is lexically restricted and does not have directionality asymmetry, a follow-up experiment trained Cantonese speakers either on left- or right-dominant deletions to see whether the right-dominant preference was due to L1 transfer from Mandarin. The results of the experiment also showed a learning bias toward right-dominant patterns. We argue that structural simplicity affects tone deletion learning but the simplicity should be grounded on phonetics factors, such as syllables’ contour-tone bearing ability. The experimental results are consistent with the findings of a survey on other types of tone alternation’s directionality, i.e., tone sandhi across 17 Chinese varieties. This suggests that the directionality asymmetry found across different tone alternations reflects a phonetically grounded structural learning bias.
Highlights
Experimental work has focused on two types of learning biases in phonological pattern learning: structural bias, a bias that favors patterns involving simple featural specifications, and substantive bias, a bias that favors phonetically natural patterns (Moreton and Pater, 2012a,b)
The results suggest that a structural bias plays a role in tone alternation learning, as evidenced by the better performance in learning uni-directional, rightdominant tone alternations than bi-directional alternations
As mentioned in section “Directionality of tone alternations in Chinese,” either throughout a tone sandhi system or within each grammatical category, uni-directional tone alternations were more common than bi-directional ones, and rightward patterns were more common than leftward ones within uni-directional patterns
Summary
Experimental work has focused on two types of learning biases in phonological pattern learning: structural bias, a bias that favors patterns involving simple featural specifications, and substantive bias, a bias that favors phonetically natural patterns (Moreton and Pater, 2012a,b). In terms of structural bias, experimental work using artificial language learning paradigms shows that patterns involving more phonological features are harder to learn than patterns involving fewer phonological features (Moreton and Pater, 2012a). We show that little attention has been paid to structural bias in learning phonological patterns defined by suprasegmental features, while some work reported substantively biased learning of stress and tone patterns. The following section explains how we define structural complexity for the directionality of tone alternations
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
