Phrasal idioms have been used as evidence in syntactic theorizing for decades. A common assumption, occasionally made explicit (e.g., Larson 2017 ), is that non-literal phrasal idioms differ significantly from completely literal collocations in the kinds of syntactic structures they can be built from. I show with a detailed empirical study that this is false. In fact, the syntactic constraints on idioms and collocations are identical . In particular, patterns that are missing from one are missing from the other, most strikingly ditransitives with a fixed first object and open second object (* throw the wolves X ). Idioms and collocations should therefore be treated the same, as a broad class of conventionalized expressions . I propose a new analysis of the syntactic forms that conventionalized expressions can take. Unlike most previous analyses, I take the open slots to be part of the expression, and the task then becomes to explain the distribution of the open slots. A structural constraint on open slots accounts for the missing ditransitive pattern. It also explains why expressions with a fixed subject but open object are rare, but also why certain examples of this pattern do exist in English and in other languages.

中文翻译：

---

习语、搭配和结构

几十年来，短语习语一直被用作句法理论的证据。一个常见的假设，偶尔会明确（例如，Larson 2017），即非文字短语习语与完全文字搭配的显着不同在于它们可以构建的句法结构类型。我用详细的实证研究表明这是错误的。事实上，习语和搭配的句法限制是相同的。特别是，一个缺少的模式从另一个中丢失，最引人注目的是具有固定的第一个对象和开放的第二个对象（* 扔狼 X ）的双及物。因此，习语和搭配应该被视为一大类约定俗成的表达方式。我建议对约定化表达式可以采用的句法形式进行新的分析。与之前的大多数分析不同，我将空位作为表达式的一部分，然后任务变成解释空位的分布。开放槽的结构约束解释了缺失的双传递模式。这也解释了为什么带有固定主语但开放宾语的表达很少见，也解释了为什么这种模式的某些例子确实存在于英语和其他语言中。