Book review
Interface at a higher level: Review of Universal Grammar and Iconicity

Excerpt 节选



When Chomsky set up the generative grammar theory framework in the 1950s, he made it clear that the same set of rules and structural representations, which he called universal grammar (UG), provide the basis for the individual grammar of every language (Chomsky 1957). The ultimate goal for establishing generative grammar is to figure out what these rules are and how they work. As means to achieve his goal, Chomsky gave up the fundamental assumptions of structuralism and divided language studies into modules such as syntax, semantics, phonetics, phonology and so on. His intention is to study each module independently without referring to the others and a new term “autonomous syntax” is created to convey this idea, illustrated by the well-known sentence “colorless green ideas sleep furiously”. He admits that this sentence is semantically absurd but insists that it is syntactically well-formed and could be studied on its own.

The research conducted along the line of autonomy of syntax and other modules made tremendous progress in the past sixty and so years, generated an enormous amount of literature and brought linguistic science to an all-time height. At the same time, some hidden problems emerged gradually and one of them is related to the autonomy of syntax. To segment language into independent modules is in a sense an adaption of a common practice in scientific research and naturally inherits the shortcomings of such practice. A natural language is the means of describing the world and conveying thoughts for its speakers and it always works as a whole. Taking the language apart makes it easier to analyze its individual modules but also makes it harder to understand the interaction between modules. Chomsky’s solution to this problem is to evoke interface operations, which are supposed to involve two or more modules.


The relationship between the internal elements in these N-V and V-N compounds is apparently more diversified than what Li presents in his book. There are some more types (Shi 2011) but it is not necessary to discuss all of them here since that is beyond the scope of this paper. Suffice it to say that one reason for such diversity is the lack of overt morphological markings, which represent different semantic and syntactic status of the verb in languages like English and could help to differentiate destroying city, destroyed city city’s destruction. Another reason is that the interpretation of an unmarked Chinese V-N or N-V compound is sometimes influenced by discursive, pragmatical and other factors. The compound kao-rou ‘roast-meat’, for example, is ambiguous between an event reading and an entity reading, and which reading is chosen depends on the context. The event reading will be picked in the context of describing someone’s activity and the entity reading will be chosen for listing items on the dinner table.



It is not easy to find an account comprehensive enough for such complexity of the noun incorporation analysis or within the current framework of generative grammar. On the other hand, this is exactly what the UG-Iconicity Interface proposed in this book is designed to handle, since the order of V and N is determined by the meaning each compound is expected to produce and is therefore not random. The details are waiting to be worked out and it is my hope that Li will come up with a proposal soon.

Share on facebook
Share on google
Share on twitter