Semantics - Typology, Diachrony and Processing

Klaus Heusinger, Claudia Maienborn, Paul Portner, Klaus Heusinger, Claudia Maienborn, Paul Portner

• Publisher: De Gruyter Mouton
• Year: 2019
• ISBN:
  9783110587326
• Edition: 1
• Topic: Lingue e linguistica
• Subtopic: Semantica linguistica

Now available in paperback for the first time since its original publication, the material in this book provides a broad, accessible guide to semantic typology, crosslinguistic semantics and diachronic semantics. Coming from a world-leading team of authors, the book also deals with the concept of meaning in psycholinguistics and neurolinguistics, and the understanding of semantics in computer science. It is packed with highly cited, expert guidance on the key topics in the field, making it a bookshelf essential for linguists, cognitive scientists, philosophers, and computer scientists working on natural language.

Emmon Bach and Wynn Chao

1 Semantic types across languages

Emmon Bach and Wynn Chao, London, United Kingdom

1. Introduction
2. Universals and universals
3. Model-theoretic semantics
4. Universal types: Model Structures for natural languages
5. Types and categories
6. From types to categories
7. From categories to types
8. Lexical semantics
9. Conclusions
10. References

Abstract: A semantic perspective on language universals deals with absolute universals of meaning and the semantic import of syntactic and inflectional categories, and lexical items. This paper takes a model-theoretical basis as primary and looks both at the independent semantics and at the mappings from syntax to semantics and from semantics to syntax. Denotational semantics is contrasted with conceptual semantics. The grammatical side of the discussion works mainly with categorial and cartographic approaches. The general approach is hypothetico-deductive and at every point in a description it is necessary to ask whether we are dealing with a universal property of human language or with a property or distinction that is limited to one or several languages. In the latter case we need to ask about the space of variation and about dependencies across such spaces. Looking from grammar to meaning, the paper takes up nominal, verbal, and clausal constructions as well as sub-topics: mass and count nouns, numbers and number, adjectival modification, kinds of verbs and verbal complexes.

1 Introduction

Languages can be compared as to the semantic types they invoke per se and as to the reflections of these types in other aspects of language, for example, syntactic categories, syntactic and morphological features, word-building, and so on. We are interested in universals in both perspectives.

There are two main approaches to semantics, sometimes placed in opposition: (i) Denotational semantics and (ii) Conceptual semantics. The two may, however, be considered complementary: Denotational semantics assigns denotations “in the world” to the expressions of the language being interpreted. Conceptual semantics links expressions of the language to concepts considered as psychological entities “in the head”. In practice, people who follow the two modes or traditions mostly do so independently of each other. In this sketch, we will mainly follow the first approach. In both approaches questions about universals loom large. At every point when we investigate a particular language or build and test a general theory about human languages we face this question: Is this piece of our description something that is limited to a single language, or is it something that is common to many or all languages?

Two moves are possible and both have been and still are popular:

• A. Assume parochial until licensed to say otherwise!
• B. Assume universal unless forced to say otherwise!

Here we find in practice a parting of the ways on methodological grounds. On the one side we have the empiricist stance epitomized by Leonard Bloomfield’s dictum: “The only useful generalizations about language are inductive generalizations” (Bloomfield 1933: 20); on the other, we have the hypothetico-deductive procedure of most generative linguists. With these two approaches, the notion of a universal takes on quite different meanings. For one side, saying that something is a universal means that you have investigated all languages and found that every language instantiates the universal. Of course no one has done or ever will investigate all languages, so we must be content with looking at representative samples. On the other side, a universal is just an element in a theoretical Structure that can be supported by looking at the consequences of the general theory. (Lewis 1972 is a classic defense of the denotational view. Ray Jackendoff (in many writings, for example 1972 is a prominent advocate of the conceptual position. Zwarts & Verkuyl 1994 is a valuable study, which in effect provides a model-theoretic interpretation for one version of conceptual semantics.)

The first approach was characterized like this by William Croft:

One of the features that distinguishes the typological method of discovering constraints on possible language type is the empirical method applied to the problem. If a typologist wants to find restrictions on possible relative clause Structures, for example, he or she gathers a large sample of languages and simply observes which of the possible relative clause types are present and which are absent. That is, the restrictions on logically possible language types are motivated by the actually attested language types. If there is a gap in the attested language types, then it is provisionally assumed that the gap represents a constraint on what is a possible language, and explanations are sought for the gap. This is the inductive method, which must be used in constructing generalizations from empirical data. (Croft 2003: 49)

Here is a representative defense of the second position by Guglielmo Cinque, writing about the cartographic approach to syntactic Structures (see section 6.2.2 below):

What makes the enterprise all the more interesting is the mounting evidence of the last several years that the distinct hierarchies of functional projections may be universal in the inventory of the heads they involve, in their number, and in their relative order (despite certain appearances). This is, at any rate, the strongest position to take, as it is compatible with only one state of affairs. It is the most exposed to refutation, and, hence, more likely to be correct, if unrefuted.

In front of recalcitrant facts we might be led to a weaker position—one that allows languages to vary either in the inventory, or the number, or the order of the functional heads that they admit (or any combination thereof). Even if this position should eventually turn out to be right, methodologically it would be wrong to start with it, discarding the stronger position. That would only make us less demanding with respect to the facts and could lead us to miss more subtle evidence supporting the stronger position (a risk not present under the other option). (Cinque 2002: 3–4)

Typological questions arise immediately when you try to advance along either route.

In our research, we follow generally the second mode. Why? Perhaps surprisingly, in part because it leads to better empirical coverage. The main reason is that recording what is can tell you what is possible but never tell you for sure what isn’t possible. Of course, there are dangers, but you have to do good linguistics no matter how you operate. Elicitation has its pitfalls but so also does the online search!

Take the NP-Quantifier universal of Barwise & Cooper (1981):

NP-QUA...