Semantic criteria
One particularly obvious way of delimiting the categories is on the basis of semantic criteria; in other words, on the basis of commonalities in the meanings of words in any given class. This may well have been the way you yourself were first taught to think of major parts of speech like noun, verb and adjective/adverb. The following set of traditional definitions are rep resentative of semantic interpretations of the parts of speech (adapted from Huddleston 1983, following Curme 1935):
Noun (substantive): word used as the name of a living being or lifeless thing.
Verb: word which denotes action or a state of being. Adjective: word which denotes a property or characteristic of some object, person or thing.
QUESTION What might some problems be with these definitions?
Semantic criteria for the parts of speech are very entrenched in our gram matical thought. One of the earliest English grammarians to write in the vernacular, William Lily (c. 1468–1522), defined noun in his Latin gram mar as ‘the name of a thing that is and may be seen, felt, heard, or under stood’; the connection with Curme’s definition is obvious. In (7) above, one of the reasons that we’re likely to reject why/if, then, after/before and try as nouns is that we feel they’re not ‘thingy’ enough: their referents are not abstract or concrete objects, unlike the referents of prototypical nouns
. Semantic definitions, however, prove to be hopelessly inadequate as defi nitions of the parts of speech. On the semantic definition, verbs are supposed to ‘denote action or a state of being’ or, in an alternative formulation (COBUILD grammar, p. 137), ‘indicate what sort of action, process or state you are talking about’. If that definition is to be accurate, it must mean that anything which does that is a verb; otherwise, the definition won’t work. But we find that there are countless nouns which ‘denote action or a state of being’ or ‘indicate what sort of action, process or state you are talking about’:
You’ll be able to think of many more examples for yourself (see Hopper 1997 for interesting discussion). Some scholars have developed more sophisticated descriptions of the claimed underlying semantics of nouns and verbs. Thus, Givón (1979) claims that noun meanings and verb meanings occur at oppo site ends of a ‘time-stability’ continuum: nouns prototypically denote percepts which possess ‘time-stability’ – they refer, in other words, to things or objects, which persist over time. Verbs, by contrast, prototypically denote percepts which lack time-stability – actions and events, which evolve through time and cannot be fixed (on prototypicality, see 7.1.3). This is an attractive idea, but it does not prove to be a viable definition of noun and verb meanings. For every noun referring to a time-stable thing or object, we can produce one which refers to a fleeting ‘object’ which lacks time-stability: think of the nouns spark, glint, flash, splash, wince, cry, blink, shiver. Many verbs, by contrast, denote stable situations: to exist, remain, soak, rest, belong and many others. These seem just as ‘prototypical’ as any others. These facts challenge Givón’s claim of prototypicality. The grammatical difference between nouns and verbs simply does not seem to be reducible to a semantic one.
Attempts to discover a semantic commonality to the class of adjectives fare no better. The traditional semantic definition of adjective, ‘word which denotes a property or characteristic of some object, person or thing’, is far from an adequate way of delimiting the category. There are many nouns and, to a lesser extent, verbs which perform exactly that function:
For instance, if I say that her sadness increased, sadness refers quite unam biguously to a ‘property or characteristic’ of the person, just as much as it does to ‘a lifeless thing’, the definition of noun. Similarly, to say that something differs from something else, is certainly not to say anything about an action or event in which it’s involved; we could say that it denotes a state of the referent, but there’s no reason not to also say that it denotes a property or characteristic. Indeed, whenever something is in a particular state it can be described as possessing the property characteristic of that state: if, for example, I know French, then I am in a particular state of being (knowing French), and I possess the property of knowing French. Or again, if you live in Sydney, the living can be described both as a state of being you’re in, and as a property you possess – the property of living in a certain place. That crossover between states and properties introduces a total indeterminacy into the semantic definitions of adjective and verb, rendering them useless as definitions of the categories.
But these sorts of indeterminacy aren’t even the main problem with a meaning-based classification. The real problem is that such classifications are circular. As pointed out by Lyons (1968: 318), the ‘only reason we have for saying that truth, beauty and electricity are things is that the words which refer to them in English are nouns’. In other words, what counts as a ‘thing’ in English seems not to be independently established, but depends on whether the word for it is a noun: possibility, fraud and implication are nouns and therefore can all be thought of as the names of things, whereas if, then, try are not nouns and cannot. Lexical category, in other words, seems to determine thingness, not the other way around. If we had some independent way of saying what counts as a thing in English, we would be in a better position. As it is, however, it seems that virtually the only evidence we have on the thingness of a concept is, precisely, what part of speech it belongs to. Why is to operate an action, but an operation a thing? For Lyons, the only reason is that the former is a verb, the latter a noun. This means that we cannot appeal to thingness, eventhood and so on as criteria for grammatical category, since they are not known independently of the very grammatical features which they are supposed to establish.
But perhaps there are some more general definitions we can find. What about the function of verbs and nouns? Could we say that verbs are the part of speech which predicate (i.e. attribute a property to a referent), and nouns are the part of speech which refer? In a sentence like Sarkozy has resigned, for instance, the proper noun Sarkozy refers to an individual, while the verb has resigned attributes the property of having resigned to him. Unfortunately, neither predication nor reference will work as definitions of ‘verb’ and ‘noun’. Predication can’t be an adequate definition of verb, since the notion of ‘predicate’ is usually understood in terms of the notion ‘verb’. Any attempt to define verbs as predicates would therefore be circular. (See the box ‘Verbs and predication’ for discussion.) As for defining nouns through reference, there are two problems. First, nouns don’t always refer. In a sentence like Charles will not be king, the subject noun Charles refers to an individual, but the complement noun king is non referring (see 3.2.2.2 for discussion). Thus, we cannot identify the function of the noun as always being referential: many nouns and noun phrases will be non-referential.
Maybe we could get around this by saying that nouns are that part of speech which can refer, but which need not do so. This might prove to be an effective strategy, if it weren’t for our second problem: once we look closely at the question, we find that it isn’t nouns that refer, but noun phrases. Consider the following sentences:
The underlined phrases in (8a–c) are noun phrases, not just nouns, for well-known syntactic reasons which I’m assuming are familiar. (If they’re not, you’ll find clear explanations in standard syntax texts like Carnie 2007, Radford 2004, or Bloor and Bloor 2004.) Sentences (8a–c) show that noun phrases can sometimes consist of a sole noun (ice, civilization and books respectively). Often, however, there needs to be more than just this single ‘head’ noun in order to convey the intended reference. In (9a–c), the underlined noun phrases contain many other parts of speech than the head noun. Yet only the full underlined noun phrase contains the information we need in order to identify the referent. Clearly, then, reference is achieved at the level of the entire phrase, not at that of the part of speech itself (see 3.2.2.2). So our suggestion that nouns might be defined as the part of speech that refers won’t turn out to be right: noun phrases, not nouns, are the bearers of referential force.
Maybe we can save the situation. Maybe if we start from a different point we can still use the connection between noun phrases and reference as a way to define nouns. What if, instead of defining nouns through reference, we defined nouns through noun phrases, and then NPs through reference? In other words, maybe we could say that nouns are the heads of NPs, and that NPs are constituents which can refer. If we were able to do that, we’d have discovered a way to account for the membership of the noun category syntactically (by defining it as the head of an NP), and we’d have defined NP functionally (as a potentially referring constituent). Unfortunately, there are two problems with this. First, the very definition of NP makes reference to the category noun. Let’s assume the rule NP S (Det) (Adj) N. Since this contains the category Noun in its definition, we need to know what the nouns are in a sentence before we can identify what the noun phrase is. That’s obviously going to be a problem if we want to use the NP category as a way of defining nouns. Second, not just Ns are the heads of NPs. Any pronoun can constitute an NP on its own:
Furthermore, if sentences like There is no after and the others in (7) contain NPs, then these are arguably headed by parts of speech other than Ns. As a result, the suggestion that we could define noun as the head of an NP will not isolate those parts of speech which we want to count as nouns in the first place. It looks as though we just won’t be able to ground the cat egory of noun in reference after all.
