Semantic criteria: what words mean

There are no semantic criteria, aspects of the meaning of the different classes of words, that would enable us to decide whether any given word is a noun, adjective, verb, adverb or preposition. We must accept right here that meaning cannot be exploited in this way. The traditional definition of nouns as words denoting people, places or things does not explain why words such as anger, idea or death are classified as nouns. Race the noun and race the verb both denote an event, as do the verb transmit in They transmitted the concert live and the noun transmission in The live transmission of the concert.

On the other hand, this book is based on the view that grammar is interesting because it plays an essential role in the communication of coherent messages of all sorts. It has been demonstrated many times that humans cannot (easily) remember meaningless symbols such as random sequences of words or numbers, like telephone numbers and PIN numbers. Psycholinguists know that children cannot learn sequences of symbols without meaning. It would be surprising were there no parallels at all between patterns of grammar and semantic patterns; we abandon the traditional notion that classes of words can be established on the basis of what words denote, but careful analysis does bring out patterns. The analysis uses the ideal of central, prototypical members of word classes as opposed to peripheral members, and it focuses on what speakers and writers do with words rather than on the traditional dictionary meanings.

The key move in the investigation of word classes is to accept that word classes must be defined on the basis of formal criteria – their morphological properties, their morpho-syntactic properties and their syntactic properties. Only when these formal patterns have been established can we move on to investigate the connection between meaning and word classes.

How do formal criteria and the concept of central members of a word class help the investigation of meaning? Interestingly, the traditional description of nouns as referring to persons, places and things turns out to be adequate for central nouns. Nouns such as girl, town and car combine with the and a, take the plural suffix -s, are modified by adjectives and occur to the left or the right of the verb in [NON-COPULA, ACTIVE DECLARATIVE] clauses. They also refer to observable entities such as people, places and things. What is significant is the combination of syntactic and morpho-syntactic properties with the semantic property of referring to people, places or things.

Many analysts argue that nouns such as anger, property and event do not denote things. However, these nouns do possess all or many of the syntactic and morpho-syntactic properties possessed by girl, town and car: a property, the properties, an interesting property, invent properties, This property surprised us and so on. Anger meets some of the major criteria The anger frightened him [subject, and combination with the] but not *an anger. The fact that the major formal criteria for prototypical nouns apply to words such as property and anger is what justifies the latter being classed as nouns. On the assumption that these formal properties are not accidental, it also suggests that ‘ordinary speakers’ of English treat anger as though it denoted an entity.

A discussion of the linguistic and cognitive issues would be inappropriate here. What cannot be emphasized enough is that a word’s classification as noun, verb and so on is on the basis of formal criteria; the terms ‘noun’, ‘verb’ and so on are merely labels for classes which could be replaced by neutral labels such as ‘Class 1’, ‘Class 2’ and so on. Words apparently very diverse in meaning such as anger and dog share many major syntactic and morpho-syntactic properties, and this raises deep and interesting questions about how ‘ordinary speakers’ conceive the world. It leads to the unexpected conclusion that the traditional semantic definitions of word classes, while quite unsatisfactory as definitions, nonetheless reflect an important fact about language and how ordinary speakers understand the world around them.

The need for both formal and semantic criteria becomes quite clear in comparisons of two or more languages. Descriptions of Russian, say, contain statements about the formal properties of nouns and verbs in Russian; descriptions of English contain statements about nouns and verbs in English. But formal criteria do not allow the English words labelled ‘noun’ to be equated to the Russian words labelled ‘noun’; the formal criteria for the English word class are completely different from the formal criteria for the Russian word class. In spite of this, analysts and learners of Russian as a second language find no difficulty in talking of nouns in English and nouns in Russian and in equating the two. The basis for this behavior must be partly semantic; central nouns in Russian (according to the Russian formal criteria) denote persons, places and things, and so do central nouns in English.