This essay is going to analyse the role of s(emantic)-selection in a generative theory of syntax, passing through some divergent linguistic points of view.
First of all, we need to clarify what a generative theory of syntax is. A generative theory of syntax states that it is possible to generate an infinite number of different sentences from a finite set of syntactic rules. These rules correspond to certain constraints on how to combine words together, which need to be respected in order to have grammaticality. Grammaticality presupposes a selection of lexical items to match with other lexical items. In the generative theory of syntax there are mainly two selections: c(ategorial) – selection ; s (emantic) – selection. The former is based on categorial selectional features and the latter on semantic selectional features. C-selection is considered syntactic and s-selection semantic. This essay is going to focus on s-selection. In order to understand the importance of s-selection to a generative theory of syntax it is important to first clarify the relationship between the syntactic and the semantic components of the grammar. It has been extensively discussed whether the semantic component derives from the syntactic component or viceversa.
Chomsky, for example, wants to defend the independence of syntax. In Syntactic Structures (1957) he is mainly concerned with problems of syntactic description rather than semantic description and he considers the input of semantics only in terms of the output of syntax. However, in Aspects of the Theory of Syntax (1965) the attention is moved to the interralation between syntax and semantics and the conclusion is that it is very difficult to delimit the boundaries between the two domains, eventhough syntax seems, for him, still more important than semantics.
In Syntactic Structures (1957) Chomsky starts from Katz and Fodor (1963) to then drift apart from them. According to them the semantic component of a linguistic theory is a ‘projection device’ wich interprets abstract syntactic objects (pag 103) and which is composed by two components: a dictionary and projection rules. The dictionary assigns a meaning to every lexical item in the language and the projection rules then assigns a semantic interpretation to the strings in deep structures which have been generated by the syntactic base component. However, Chomsky thinks that semantic interpretation should not be assigned only to deep structures, because otherwise it would not be possible to attribute semantic interpretation to the transformational sub-component of the syntactic component. Chomsky uses the example of the present perfect aspect in English and the passive transformation to demonstrate that the surface structure contributes to the meaning.
He thinks that
- it is not possible to determine the semantics features of an utterence independently of the grammar.
- It is not possible to derive syntax from semantics.
Therefore syntax must precede semantics.
However, there are many objections to his argument, which try to show the priority of semantics over syntax and to which he replies with:
Syntax: semantics = linguistic form: use of language.
The formal study of the structure – i.e. syntax- is an instrument to understand the use of language – i.e. semantics. This means that we understand semantics from syntax. This confirms once again the priority of syntax over semantics.
On the other hand J. Grimshaw focuses more on the importance of semantics. After analyzing exclamatory & interrogative complements, null complement anaphora and the theory of the lexicon, she presents an account of the incompatibility between complement and predicate. This incompatibility derives from three rules:
- the predicate and its complement must be semantically compatible
- the complement must meet the idiosyncratic selectional conditions encoded in the semantic frame of the predicate
- the complement must meet the (also idiosyncratic) syntactic conditions encoded in the subcategorizations frame of the predicate.
(Grimshaw, J. (1979) Complement Selection and the Lexicon, Linguistic Inquiry 10.2 pp 279-326
The key for this account is the concept of ‘idiosyncratic’. This means that every predicate has a ‘peculiar’ semantic and subcategorization frame that has to be satisfied by the complement in order to preserve grammaticality.
Another point in favour of the selectional restrictions is made by Riemsdijk, H. & E. Williams. They think that the study of selectional restrictions permit to discover some empirical basis for the distinction between arguments and non arguments. Therefore, syntax seems to derive empirically from semantics.
Another account which strongly supports Semantic Selection over Categorial Selection is made by the the linguist Pustejovsky who says that there are two ways to arrive to the priority of semantics over syntax:
- ‘ There is no one-to-one mapping from underlying semantic types to syntactic representations; rather, a syntactic phrase is only fully interpretable within the specific semantic context within which it is embedded’;
(Pustejovsky, The Generative Lexicon, Cambridge: MIT Press).
Syntax and semantics are not on the same level. Semantics is more important and enable the interpretation of the syntactic phrase.
- ‘Because the representation of the semantic information in the qualia structure, argument structure, and extended event structure is richer than what conventional models associate with a word, a more complex model of filtering and checking is necessary for restricting the output to actual syntactic form’
(Pustejovsky, The Generative Lexicon, Cambridge: MIT Press).
Since semantics is richer than syntax, the lexical item needs to be more accurately filtered and checked before it can be considered in syntactic terms.
Moreover, if we adopt the semantic theory according to which semantics is more important than syntax and determines it, it is still possible to make another important distinction which Lehrer points out:
- ‘One level or aspect of the meaning of a word is the set of other words that it collocates with’ (The Lexical position)
- Co-occurrence restrictions are the result of the meaning of the lexical items and that collocations are reflections of this fact. (The semantic position)
(Lehrer, Semantic Fields and Lexical, North Holland Publishing Company (1974))
In conclusion, as the essay has demonstrated it is always possible to make other new distinctions wich apparently make the theory more complete and detailed but which, at the same time, arise new other issues.
Chomsky, N. (1971) Syntax and Semantics in J.Allen & P. van Buren (eds)
Chomsky: Selected Readings, Oxford: Oxford University Press. 101-126
Grimshaw, J. (1979) Complement Selection and the Lexicon, Linguistic Inquiry 10.2 pp 279-326
Lehrer, Semantic Fields and Lexical, North Holland Publishing Company (1974)
Pustejovsky, The Generative Lexicon, Cambridge: MIT Press.
Van Riemskijk, H. & E. Williams (1986) Introduction to the theory of Grammar, Cambridge: MIT Press.