Semantic components

What kinds of meaning, then, are encoded in the word? We can begin by referring to the same principle of constituent assembly that has served us so well so far. When considering inflectional affixes in the last chapter, it was pointed out that it is common to find two morphemes fused into one form, as in ‘come’ + past tense = ‘came’. When considering derivational affixes we noted that un- and re- can combine with various lexical items to yield others like ‘unfix’, ‘undo’, ‘unscrew’, ‘refix’, ‘retell’, and ‘review’, and so on. We have already established the semantic character of these morphological forms. We can say, then, that a lexical item like ‘unfix’ has two semantic elements or components, each given separate expression in the word form ‘ un + fix’.

Now it happens that many such derived forms have semantic equivalents which are single morphemes: ‘unwell’ = ‘sick’, for example, ‘unhappy’ = ‘sad’. Furthermore, there are many equivalents which can take the form not of single words but of phrases where the bound morpheme separates itself from bondage and becomes free. So ‘unwell’ = ‘not well’, ‘unhappy’ = ‘not happy’, ‘reborn’ = ‘born again’, ‘replant’ = ‘plant again’, and so on. In George Orwell’s novel 1984, this principle of decomposition pro­vides the basis for the reformed English of Newspeak: in Newspeak, for example, ‘excellent’ becomes ‘plusgood’, ‘bad’ becomes ‘ungood’, ‘terrible’ becomes ‘plusungood’, and so on.

Now (without commitment to the principles of Newspeak) it seems reasonable to suggest that a lexical item like ‘sick’ is a ver­sion of ‘unwell’: it is just that the two morphological elements have become fused into one. It would follow that if ‘unwell’ has two elements of meaning or semantic components, then so does ‘sick’. And if these lexical items can be said to be encodings of different semantic components, then it would also seem logical to suppose that the same can be said of all lexical items, the only difference being that such components are explicitly signalled in some cases, but not in others.

The signalling is not straightforward, however. When a free lexical form becomes bound as an affix, its meaning is not just added, but acts upon the host lexical item in various ways. Thus ‘careful’ can be analysed as ‘full of care’, but ‘careless’ does not mean ‘with less care’ but ‘with no care’. Some affixes activate grammatical relations. The suffix -able for example, contracts a passive relation with its stem. So ‘eatable’, for example, meant not ‘able to eat’ but ‘able to be eaten ’. With -less and -able, the semantic effect of affixation is predictable. In other cases things are not so simple. The suffix -er derives a noun from a verb, and so denotes an actor of the action. Thus, words like ‘baker’ and ‘keeper’ can be taken apart and glossed as ‘a person who bakes’ and ‘a person who keeps (something)’. Here the actor is a human agent. But it can also be an inanimate instrument. A ‘cooker’ is not ‘a person who cooks’ but ‘a device for cooking’, and in words like ‘printer’, ‘cleaner’, and ‘speaker’ the suffix can denote either agent or instrument. And with words like ‘creeper’ (meaning ‘plant’) and ‘breaker’ (meaning ‘wave’) the original significance of the suffix has now, in part at least, disappeared.

And it is commonly the case, of course, that the distinctive meaning of the lexical host disappears and blends in with the affix in the historical process of etymological change. So it is with words like ‘reckless’ or ‘feckless’, which cannot mean ‘with no reck’ or ‘with no feck’ since there are no such lexemes in English. Conversely, when an affix attaches itself to an existing form, it may blend with its host, and again the lexical whole is not a sum of its parts. The prefix re- is even more unreliable in this respect j than the suffix -er referred to above. ‘Return’ does not normally mean ‘turn again’ or ‘recall’ ‘call again’. When they do signal such meanings, they are generally given a hyphen in writing and marked stress in speech to indicate that the prefix retains its semantic identity. Thus, you have a ‘re-call’ ('rı: kɔ: l) button on the telephone, but you may not ‘recall’ (rı: 'kɔ: l) how to use it.

The general point is, then, that we can conceive of all lexical items as encodings of one or more semantic elements or com­ponents, whether these are overtly signalled or not, and in identi­fying them we can establish the denotation of words. Thus, one denotation of the verb ‘return’ can be specified as [come + back], another as [give + back]. ‘Come / go’ and ‘give / take’ in turn can be said to consist of components: something along the lines of [move + self + towards / away] on the one hand, and [move + something + towards / away] on the other.

These components of meaning can be seen as analogous with segments of sound. The same principle of combination is at work. In our previous discussion, we were able to establish contrasts between phonological words by invok­ing minimal differences in the sequence of sound segments. Thus, ‘come’ contrasts with ‘gum’ with respect to the one feature of voice on the initial consonant — i. e. the sounds [k] and [g] are formed in exactly the same way, except that in [g] the vocal cords vibrate and in [k] they do not. The same principle applies here: we can establish similar minimal pairs of lexical items with respect to their semantic components. Thus, ‘come’ contrasts with ‘go’ in respect to the one feature of directionality: [movement + here ]as opposed to [movement + there ].

This approach, known as componential analysis thus provides an inventory of the semantic features encoded in lexical forms. It can, of course, become immensely complicated and unwieldy, and as in all analysis, as the details proliferate they can lose their point and create confusion. The essential purpose of componential analysis is to identify certain general conceptual categories or semantic principles which find expression in the particular com­ponents. Among such categories are state, process, causality, class membership, possession, dimension, location, and, as we have seen with ‘come’ and ‘go’, directionality. By invoking them, we can move on from the denotation of particular lexemes to the sense relations that exist between them.