標記性與詞彙典型性：漢語接受度判斷實驗

Markedness and Lexical Typicality in Mandarin Acceptability Judgments

眾所皆知，越是接近典型詞的非真詞越容易被說母語者判斷為可接受；而從結構的標記性來看，則是越少標記的結構越容易被判斷為可接受。本研究用實驗的方法蒐集了漢語母語人士對非詞彙音節的判斷，建構了巨量資料庫，並首次以此探討標記性與詞彙典型性的交互作用。我們以漢語詞彙音節所共享該實驗項目聲母的程度來定義詞彙典型 性；標記性則以該子音分布於跨語言音素系統的多寡來定義。與先前的研究一致，標記性和詞彙典型性都能增進非真詞的接受度，但兩者也確實呈現交互作用：詞彙典型性對於低標記性結構的影響高於對高標記性結構的影響。相同的交互作用也見於對英語的研究。然而這樣的交互作用優選論無法預測，但我們可以用並列結構來解釋。如果我們反向看大家所熟悉的聯合制約，我們會斷定當兩個子限制條件都滿足時，其並列的聯合限制條件才能得到滿足。

It has long been known that native speakers judge nonlexical forms as more acceptable the more lexically typical they are (i.e. similar to real words). It has also been shown that speakers judge less marked (i.e. universally more natural) structures as more acceptable than more marked structures. In this study, we investigate for the first time how markedness and lexical typicality interact, using data from a large corpus of experimentally collected Mandarin native-speaker judgments of nonlexical syllables. We defined the lexical typicality of a test item in terms of how many lexical Mandarin syllables share the item's onset consonant, and defined markedness in terms of how many phoneme inventories have this consonant cross-linguistically. Consistent with prior research, both markedness and lexical typicality improved acceptability, but the two factors also interacted positively: the lexical typicality effect was stronger for less marked forms than for more marked forms. The same interaction appears in a reanalysis of English nonword judgment data. This interaction is not predicted by standard Optimality Theory, but it can be formalized with conjunctive coordination, the inverse of the more familiar local conjunction, whereby a coordinated constraint is obeyed if and only if both of its component constraints are obeyed.