Mel’cuk (2012) represents the meaning of utterances, and potentially the meaning of whole texts, as a single network. Diagram 11 represents Mel’cuk’s illustration (2012, p. 181) of the network for the English sentence, “David is shaving (himself) with a new razor.” (It excludes the tense and aspect of the verb, and number and definiteness of the nouns.)
In the diagram, word nodes are represented by small circles, and are labelled;
relations are represented by lines, with arrowheads representing the direction of de- pendency; numbers indicate types of dependency relation.
I believe Mel’cuk’s network account to be seriously misleading in implying that the dependency does not create a hierarchy (as discussed in Chapters 8 and 9); but
it is valuable in showing that utterances are fundamentally semantic structures, that syntax creates them, and that they are relational. (It is also limited by its dealing ex- clusively with conceptual meanings.)
5.6.2 Networks in imaginative English
Imaginative English sets up extra links between senses, creating often complex net- works. Consider example (20), which is part of a description of Cleopatra, the en- chanting queen of ancient Egypt. Shakespeare’s sentence (from Antony and Cleopatra, act I scene ii.) has been set out as if it were a specimen in typological linguistics, to highlight the structure.
(20) “Age cannot wither her, nor custom stale her infinite variety.”
Subj. Pred. Comp. Subj. Pred. Comp.
The parallelism of the two simple transitive clauses helps create semantic links be- tween them. It relates ‘age’ and ‘custom’ in the two Subjects, ‘wither’ and ‘stale’ in the two Predicators, and her in the two Complements. The concepts of DECLINE and ADVERSE EFFECT link ‘age’, ‘custom’, ‘wither’ and ‘stale’; ‘infinite variety’ stands out against them all.
That analysis deals only with lexis and basic syntax. The personification of age and custom extends the network of sense elements much further, especially in the situational context of the play.
5.6.3 Relations among one word’s various senses: Polysemy Introduction
Polysemy, as a property of words, is not relevant here directly, since we are studying the structure of meaning (and how it is expressed), not expressions (and what they mean). But the meanings of a polysemous word constitute a network of inter-related senses, and the conclusions reached so far have implications for polysemy that are worth noting.
‘shave’
1 2
3 1
‘new’
‘razor’
‘David’
Diagram 11. Mel’cuk’s structure of “David is shaving (himself) with a new razor”
Linguistic argument
SOED records the two senses, ‘A soft wool prepared from… certain plants’ and ‘Any substance prepared for burning next to the skin [as therapy]’ as senses of the same word: moxa is polysemous. The reason for linking such disparate senses is histori- cal – they derive from the same Japanese word. Conversely, SOED gives “Of water etc.: clear, translucent” and “Especially of water: clear…; translucent” as the senses of different words – because the Old English word scir has produced both shire, with the first of those senses, and sheer, with the second. For the study of present-day English semantics, we should take “word” with a semantic definition, not a historical one. A
“polysemous word” like moxa will then be two homonyms.
If we accept that sublexical senses are not specified on any dimension, but be- come vague or specific, bounded and bounded, and so on, according to context in use (as argued above in §5.3.4), then it is correct for SOED to give the adjective good the indefinite (linguistic) sense, “Having (enough of) the appropriate qualities” (branch I, sense <1>). In “Give me a good knife – this one’s blunt,” good comes to mean ‘sharp’
(in cognitive semantics, when the linguistic sense is applied to the context); ‘sharp’ is not a meaning of the lexical item good; the word in branch I is monosemous. However, SOED goes on to give a series of putatively distinct senses for different circumstances, such as <1b> “Of food and drink:… fresh” and <1c> “Of soil: fertile”, although it does not give “Of knives: sharp”. As Fretheim (2011) points out, the difference between
“senses” such as <1b> and <1c> is outside language, residing in our practical knowl- edge of food and soil. From the analysis of meaning given in this book, it follows that words are much less polysemous than dictionaries such as SOED allow.
A further argument concerns the distinctness of sense boundaries The belief in polysemy entails accepting that senses are countable; that relies on regarding them as distinct. That comes with the philological and lexicographical traditions in linguistics, which make the assumption that semantics, like syntax, consists of constituents built up into a larger constituent – all with a clear boundary. But we have established, I believe, that meaning in English is fundamentally a network, as in the semantic maps used in this chapter; we have seen (§5.2.1.2 above) that the boundaries between senses realised by the same word are only weakly distinct. We have seen also (§5.3.4 above) that even established senses vary in context, reconstrual being a fundamental process in treatment of meaning; specific words and uses of words “draw boundaries” on the semantic maps, as in §5.2.3 and §5.3.4; compare Haspelmath (2003). All that argues that, to an important extent, senses are not distinct and countable, and that words are not very polysemous.
Psycholinguistic view
Psycholinguistic research gives us an insight of a different kind. It has shown that peo- ple vary in whether a particular precise sense is “stored” in a “polysemous” structure, or generated afresh in each use from a single stored sense – a “monosemous” process.
It has also shown that individual speakers vary in this respect from time to time; they
call on the ready-made meaning on some occasions, but construct it from a base meaning on other occasions.
More profoundly, the psycholinguistic understanding that senses are networks falsifies the assumption that there are integral stored senses: each sense is a pattern of activation, so when it is not being activated in use, it is merely a potential, without ex- istence as an entity. As patterns of activity, then, senses may be compared to a horse’s trotting, cantering and galloping: where does a horse store its trotting when it is not using that pattern of activity?
Conclusion
We conclude that English has some degree of polysemy, in principle; that it is less polysemous than most linguists have thought, and much less so than is assumed by dictionaries such as SOED.