Flux and variance form the reality for vocabulary. Words and expressions come and go, and at any one time they usually hold a multitude of different meanings that can cover a seemingly disparate array of situations. Over time these meanings often extend, break up and change – sometimes spectacularly. Remarkably, communication continues and manages most of the time to be successful. An important facet of words is therefore that they have stable core ideas (also argued by Anna Wierzbicka). As Enfield nicely shows, word meanings depend on the existence of what he dubs 'effective semantic invariance'. This doesn't mean that words are like maths symbols with a fixed and constant designation. Far from it – their meanings are both multiple and layered.

So much goes on when words are used, with context and background providing all sorts of clues that go beyond core meanings. Addresses seek to extract all the interpretations possible from any message, often imputing meanings that may not match original intentions. So a word might start off having a particular meaning, but new ones are lurking in the wings, ready to hop on at any time. Let me take the simple conjunction since to illustrate. Its earliest recorded meaning is 'from the time that' as in Life has become dull since Fred left (Enfield's Stage 1). But we 'read between the lines', looking for enriched meanings that are somehow relevant to the context of discourse. For some people, a temporal meaning implicates some sort of causal link between these two clauses (why else would someone say such a thing?) – for them Fred's leaving has something to do with life being dull (Enfield's Stage 2). Implications can become so reliable and widespread that they then get conventionalised and become part of the actual meaning; since now has both temporal ('from the time that') and causal ('because that') meanings (Enfield's Stage 3), and the latter is on the way to prevailing (Enfield's Stage 4). Compare the conjunction after, whose temporal meaning in some contexts can have a causal interpretation. (I recall my surprise some years back when half my Structure of English class described after as a causal connector in the sentence Life became dull after they left – a lovely example of Enfield's 'gap between mind and community').

'Nick Enfield's is a new and different approach, one that connects a wide range of discipline areas and offers an account of linguistic meaning that is like no other'

So when meaning shifts take place they are gradual and continual with overlapping senses along the way – and 'bridging contexts' to assist in the transition to polysemy ('multiple meaning'). This is the stuff of historical semantics, but Enfield articulates the bridging contexts more explicitly and also the journeys that new meanings take as they become public.

An important quality of meanings is therefore that they are distributed. First, they are dispersed across the contributions of both speakers and hearers in any verbal exchange; as Enfield describes it, 'a word's meaning cannot be calculated without access to a person's response that reveals this meaning'. So there is a dialogic foundation for meaning. Second, meanings are socially distributed across communities of individual minds; a change starts off as an alteration in the behaviour of individual speakers but if the linguistic concept is useful it will be 'aired, and shared' and result in a change in community-wide convention.

The meanings we carry around in our heads seem to us so natural and inborn. Yet, as Enfield argues, since we aren't telepathic and can never know for sure what goes on inside other people's heads, we can only hypothesise what they mean by the words they use. True, we refine these assumptions along the way (not consciously of course) by being exposed to varied contexts and uses, but our word meanings remain hypotheses. I can imagine many might find this idea unsettling.

N.J. Enfield

It can happen that speakers in the one community hold different ideas of what words mean. Drawing on the idea of 'false friends' in second language learning (expressions in two languages that resemble each other but are wrongly assumed have the same meaning; for example, in German Fabrik means 'factory' not 'cloth'), Enfield describes words as 'tolerable friends' at best; in other words, speakers can have different semantic representations, but this mismatch never becomes apparent and doesn't cause problems. For example, most people aren't aware of the very different meanings that exist for the word instep. Many dictionaries have only the early meaning 'the upper surface of the human foot between the toes and the ankle' (Oxford English Dictionary definition). Some (usually older) speakers have this meaning, but most now understand instep as the underneath part of that section of the foot, and others take it to be the entire middle section, the top and underneath part (I remember well a discussion some years back by the Macquarie Dictionary editors on exactly this point – and the thorny issue of when to include, even privilege, a new sense). As Enfield describes, less frequently used words (such as instep) are more prone to this kind of variation, and because the contexts of use aren't varied, speakers aren't given the same opportunity to revise their hypotheses.

'We refine these assumptions along the way (not consciously of course) by being exposed to varied contexts and uses, but our word meanings remain hypotheses'

Another recurring theme in the book is that word meanings are subjective and reflect particularly human preoccupations. As Enfield describes, 'humans have a special propensity for attributing psychological motives behind actions and behaviour'. This is very evident in the way words are used and how they track over time. Let me illustrate with a group of words from English. It seems we are mistrustful of those who are skilful and demonstrate advanced thinking powers, and our language is brimming with examples where deceitful, fraudulent senses have come to dominate. Especially noticeable are words to do with technical innovation such as forge (earlier 'to make'), counterfeit (earlier 'to make in imitation'), fabricate (earlier 'to construct, manufacture'); artful, crafty, and cunning were also once favourable terms used of people with considerable skill (clever shows all signs of moving in the same direction).

Clearly, another important consideration is the contribution of cultural ideas and values. Words are not just anthropocentric but also culture-centric. The production and interpretation of meanings are shaped by the cultural preoccupations of speakers, and as Chapter Five demonstrates, this includes even the complex grammatical resources of a language. The cultural values and beliefs of speakers give rise to many of the linguistic habits of their daily discourse – and so it is that the cultural hang-ups of these speakers work to sculpt the life of words and also the structures within grammar.

Nick Enfield, who is professor of linguistics at the University of Sydney, is renowned for his cross-disciplinary approach, spanning linguistics, anthropology, cognition, and sociology, and this wonderful book combines his considerable talents in such diverse areas as Southeast Asian languages, linguistic principles of conversation, grammatical systems, language and social cognition, and of course semantics. Given the ground-breaking nature of Enfield's account and the complexity of his explanations, I suspect The Utility of Meaning won't make it onto everybody's list of books for bedtime reading, but those who love language and who love playing with words and their meanings will surely relish this book.