The Russian verb pobedit’ ‘conquer’ shows what Sims calls canonical defectiveness, “the complete lack of any word-form filling a given paradigm cell [...] in the context of a maximal expectation that there should be some form corresponding to that cell” (250). That cell is the first-person singular nonpast, in which *pobežu is bad and so are *pobedju and *pobeždu. In this wide-ranging study she cites data from two dozen languages and employs a variety of tools like statistical analysis and information theory in order to provide a context for understanding the defectiveness of pobedit’. Introductory chapter 1 poses the question: Are paradigm gaps random anomalies, epiphenomena, or normal morphological objects? They are anomalies when they are generated by the regular rules of inflection but then must be specified [–lexical insertion] to prevent their occurring in a sentence. They are epiphenomenal when they reflect morphological rule competition, such as the competition between the Russian reflex of /dj/ (in *pobežu) and the Church Slavic reflex (in *pobeždu). The epiphenomena explanation could have been pursued further. The same competition between Russian ž and Church Slavic žd is seen in the nonoccurring imperfective *pobeživat’ and the standard imperfective pobeždat’, which shows that the Church Slavic reflex, although acceptable in derivation, is not acceptable in inflection (or no longer acceptable: Pushkin had straždut as the 3pl. of stradat’ ‘suffer’, but it has been replaced by stradajut). Sims rejects these two options and throughout the book repeatedly argues that such gaps are “normal morphological objects” (209) and that inflectional defectiveness is “a systemic variant of normal inflectional structure” (11). In chapter 2 Sims defines inflectional defectiveness and evaluates candidates for it. In the Yimas sentence taŋatpul ‘You didn’t hit me’, the absence of ma ‘you’ is not a gap because the sentence is well formed and interpreted as having a second-person singular subject. (“This is thus an example of zero expression of the nominative, which is not to be confused with lack of expression” [32]) “Inasmuch as [taŋatpul] is a well-formed sentence and the ineffability requirement of the definition is thus not met, this does” [surely the author

中文翻译：

---

Andrea D. Sims 的屈折缺陷

俄语动词 pobedit''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''''' 显示了西姆斯所说的规范缺陷，“在最大期望的背景下，完全没有任何词形式填充给定的范式单元 [...]细胞”（250）。该单元格是第一人称单数非过去时，其中 *pobežu 是坏的，*pobedju 和 *pobeždu 也是坏的。在这项广泛的研究中，她引用了两种语言的数据，并使用了各种工具，如统计分析和信息理论，以便为理解 pobedit 的缺陷提供一个背景。介绍性第 1 章提出了一个问题：范式差距是否是随机异常、附带现象、还是正常的形态对象？当它们由规则的屈折规则生成但必须指定 [-词法插入] 以防止它们出现在句子中时，它们是异常的。当它们反映形态规则竞争时，它们是附带现象，例如 /dj/ 的俄罗斯反射（在 *pobežu 中）和教会斯拉夫反射（在 *pobeždu 中）之间的竞争。可以进一步研究附带现象的解释。俄语 ž 和 Church Slavic žd 之间的相同竞争可见于未发生的未完成式 *pobeživat' 和标准的未完成式 pobeždat'，这表明 Church Slavic 反射虽然在推导中是可接受的，但在屈折中是不可接受的（或不再可接受：普希金曾将 straždut 作为 stradat'''suffer' 的 3pl.，但它已被 stradajut 取代）。西姆斯拒绝这两种选择，并在整本书中反复论证这种间隙是“正常的形态对象”（209），而屈折缺陷是“正常屈折结构的系统变体”（11）。在第 2 章 Sims 定义了屈折缺陷并评估了它的候选者。在 Yimas 的句子 taŋatpul '你没有打我'中，没有 ma 'you' 不是一个空白，因为句子结构良好并且被解释为具有第二人称单数主语。（“因此这是主格零表达的一个例子，不要与缺乏表达混淆”[32]）“因为 [taŋatpul] 是一个结构良好的句子，因此定义的不可表达性要求是不满足，这确实“[当然是作者 ma 'you' 的缺席不是一个空白，因为句子结构良好并且被解释为具有第二人称单数主语。（“因此这是主格零表达的一个例子，不要与缺乏表达混淆”[32]）“因为 [taŋatpul] 是一个结构良好的句子，因此定义的不可表达性要求是不满足，这确实“[当然是作者 ma 'you' 的缺席不是一个空白，因为句子结构良好并且被解释为具有第二人称单数主语。（“因此这是主格零表达的一个例子，不要与缺乏表达混淆”[32]）“因为 [taŋatpul] 是一个结构良好的句子，因此定义的不可表达性要求是不满足，这确实“[当然是作者