Link to the University of Pittsburgh
Jesus College, University of Cambridge

What is a 'keyword'?

A ‘keyword’, in the sense in which we investigate keywords on this website, is a socially prominent word (e.g. art, industry, media or society) that is capable of bearing interlocking, yet sometimes contradictory and commonly contested contemporary meanings.

Interaction between the meanings of such a word, which commonly include technical as well as everyday meanings, remains unresolved across a range of topics and fields. Failing to grasp the complexity of the word can lead to cross purposes and confusion in public debate as well as in personal conversation. This use of ‘keyword’, it will be quickly appreciated, differs significantly from the increasingly common use of ‘keyword’ to mean a commercially important online search term (for example in online advertising).

Keywords in the ‘complex social vocabulary’ sense we are concerned with here call for analysis. But such analysis can be difficult. This is partly because the confusions that arise between different senses of a word judged to be a ‘keyword’ don’t all happen in the same way or for the same reason.

There are at least two major sources of misunderstanding (though in fact the two types are interconnected).

Historical changes of meaning

The first type of difficulty arises because of historical changes of meaning. The various meanings of a given word are like historical strata or layers. Some of a word’s earlier meanings persist into the present; others have become recessive; and others again have disappeared altogether and been replaced by new ones. (Such processes can be seen for example in the words faith, or gay.) If we wish to engage with discourse from earlier periods (as is essential, if we are to understand the accumulated beliefs and values of earlier members of what we take to be our own language community), then it is essential to identify senses that speaker or writers could have intended to convey at the time they said or wrote what they did.

Scroll to top ▲

Multiple concurrent meanings (polysemy)

The other type of difficulty reflects a general characteristic of the language system as any language user mentally models it. The difficult ‘key’ word is polysemous or vague (that is, it has multiple, concurrent senses which are historically and semantically related, or it under-specifies what it denotes to such an extent that its use calls for narrowing or modulation in different contexts).

For example, liberal can be used in a narrowly technical, party-political sense. But the word can also be used in a broader, political sense that characterises positions not all of which are ‘liberal’ in the first sense. The word ‘liberal’ is also commonly used in other, general and non-political senses. For example it can just mean ‘generous’. Each of these senses can be given either a positive or negative spin, depending on context and the speaker's beliefs or attitudes - or the assumed beliefs of the person or people the speaker is addressing. Different senses for the given keyword are therefore simultaneously available: they are alternatives within the model of the language that the speaker or hearer has built up in his or her mind.

The two kinds of difficulty that perplex keywords – which we are describing as  historical and concurrent (which linguists generally describe as ‘diachronic’ and ‘synchronic’ dimensions of language) - are inextricably linked. Since (polysemous) senses are available for a given word only during specific periods of use (some coming to prominence, some remaining stable, and others fading in currency or undergoing a process of change), identifying meanings that a speaker could plausibly have intended in a given context, and assessing what impact those senses might have on a particular audience, is far from straightforward.

Scroll to top ▲

Keywords and social thinking

Simultaneous but divergent senses associated with ‘keywords’ are significant for contemporary public debate. On the tongues of different speakers, they can confuse discussion by obscuring socially important arguments that might otherwise be advanced using the word in question.

Consider this general circumstance for possible misunderstanding. We can only understand the social world around us by reflecting on (as well as simply using) the symbolic resources available to us, since symbols are the principal tools with which we articulate thoughts, feelings and attitudes. But the symbolic resources we use are something we inherit, rather than create out of nothing for ourselves. So understanding them must have a historical dimension, for example a need to identify meanings that may have calcified earlier patterns of thought or values. The challenge in analysing such meanings is not merely a ‘history of the language’ task: that is, a task of describing how the symbolic resources available to any given language user developed in the past into what is now our contemporary language system, in which those meanings are learnt and experienced. Additional complexity is created by the fact that the history of symbolic resources is not a single, shared history of ‘use’. Rather, it is a complicated history of varying ‘uses’: by different social groups of speakers and writers, for different purposes, under given and changing conditions, and at different times.

The challenge in understanding present use of ‘keywords’ is accordingly a more difficult task than one of either historical narrative, or contemporary description, or both. It involves grasping how any given speaker, on a given occasion of use, selects among and deploys – though not always fully intentionally – the set of options available to them for conceptualising and expressing what they want to say. Those options available to the speaker or writer might appear to map directly onto meaning ‘options’ as experienced by all the other members of the same language community, and as set out in a dictionary. To some extent, however, this is not the case. If it were so, we would all share an identical, even if highly complex, understanding of what democracy means, or what public interest is, or what corporate conveys). In fact, the meanings available to any given language user - whether they are trying to get something across, or trying to understand what someone else is saying or writing - have been conditioned by the contradictory history of a language that each speaker inherits differently, as well as to different extents, depending on education, circumstances, and accumulated beliefs and values.

Scroll to top ▲

The word ‘keyword’

The Keywords Project builds on and seeks to extend an analysis of keywords developed by Raymond Williams, and outlined in his Preface to Keywords: a Vocabulary of Culture and Society (Fontana 1976/1983). Williams’ approach is well illustrated in the many word entries contained in that book. Raymond Williams did not, however, invent the word ‘keyword’. Nor did he assert any proprietary interest in that word. Other writers had used the term earlier, in overlapping but sometimes divergent ways. And others have used it since, often slightly differently (cf. Tony Bennett, Lawrence Grossberg and Meaghan Morris’s, New Keywords [Blackwell, 2005] or the journal Key Words, published by Spokesman Books).

The origin of the word ‘keyword’ is sometimes traced to Michel Bréal’s Semantics: Studies in the Science of Meaning (1900), translated by Mrs Henry Cust (Dover, New York, 1964). More recently, emphasising connections between lexical semantics and social history, Georges Matoré developed his notion of ‘mots-clés’ for socially privileged key words in La méthode en lexicologie: Domaine français (Didier, Paris, 1953). Williams was working on his ‘keywords’ study from around the same period as Matoré, though evidently without any connection (and with a different aim: that of publishing a historical glossary as an appendix to his book Culture and Society 1780-1950 [Chatto and Windus, 1958]).

If you google the word ‘keyword’ now, however, and look through your first twenty results, the chances are you will find only references to using target words in searchable databases or for finding relevant web pages. This is a different sense of ‘keyword’, illustrating the general point made about the polysemy of words above. When used in this sense, ‘keyword’ tends to collocate, or be found next to, words such as ‘smart’ and ‘good’. This use of the word also combines to form the compound ‘meta-keyword’ (which indicates a way of targeting clusters of keywords rather than individual words in a dataset). Exactly how such keywords affect patterns of thought and memory, and how they relate to proprietary interests (e.g. as regards how they function in slogans and trade marks) remains controversial.

The value of distinguishing these different meanings of the term ‘keyword’ lies in showing that the meaning of this term also changes, in the same way that the meaning of other words does. Changes in the meaning of ‘keyword’, as with changes in those other words, have implications as regards how different people will tend to think – in this case about words and their meanings. Current prevalence of a ‘search’ sense for ‘keyword’, for example, challenges any obviousness that might be presumed in the ‘Williams sense’ among readers coming to his book Keywords for the first time. Or indeed when users visit this website.

Following Williams’ practice of developing accounts of words as reflective essays based on the Oxford English Dictionary (OED), we may take the OED as a point of departure.

‘Key-word’, and ‘keyword’ as a combination, both appear under the entry for the headword ‘key’ (sense ‘key’ n1, 18 comb.). There we find the following:

  1. a word serving as a key to a cipher or the like (‘The key-word of these inscriptions’);
  2. a word or thing that is of great importance or significance; spec. in information-retrieval systems, any informative word in the title or text of a document, etc., chosen as indicating the main content of the document.

What might be called a ‘significance’ sense is given in both senses (a) and (b) here, alongside the ‘search’ sense. But the idea of what makes something significant isn’t necessarily the same in the two cases. How ‘search’ relates to ‘significance’ is not always easy to be sure of, without further context. This is partly because you generally search for things that, at least while you are searching, seem things you will attach significance to if you find them. Most searches therefore have a significance of some kind. We can speculate, however, that the balance in relative prominence of the ‘search’ and ‘significance’ senses may have altered during the recent period of online information. (This is something that may be suggested by, but isn’t provable from, the OED quotations which range from 1859 through to 1971.) There is a further difference between the sense of search and that of significance for a keyword, too, which relates to how you view what happens once you've found an instance of your keyword. In one case, your search is over. In the other, it is just beginning.

‘Keyword’ is a figurative expression. It evokes privileged ways into something (a small door into a large room, lifting the lid off a locked casket). This aspect of the word begins to explain why the term is used in the titles of reference works that list core technical vocabulary for specialized fields (e.g. in a volume called Canadian Constitutional Keywords, or Keywords in Language and Literature). In such publications, the focus is on selected words (each made the headword for an entry) that may or may not be significant in their everyday sense, if they have one; but the word in question is assumed to offer special leverage in understanding some technical field. Implicit in such collections is an idea that what you unlock with your keyword will be enriching or at least worth unlocking. In this respect, we can contrast ‘keywords’ with what might be called ‘symptom-words’. These are words that may still offer a key to analyzing a cultural milieu, but which in themselves appear trivial and are often referred to as ‘buzzwords’. Such words undoubtedly offer social insights, but of a different kind. Many collections of such ‘buzz words’ are available – particularly because the words in question and their meanings change especially fast; a good illustration both of the general phenomemon – and also of how quickly such collections become dated - is John Morrish, Frantic Semantics: Snapshots of our Changing Times (Macmillan, 2000).

Scroll to top ▲

Building on Raymond Williams’ Keywords

Raymond Williams published his book Keywords in1976. It carried the subtitle, ‘a vocabulary of culture and society’ and  was an extended wordlist that took ‘key words’ in contemporary social and political debate and used their philological history as a means of engaging with their contradictory contemporary uses and meanings.

The book had both a linguistic interest and also much wider implications in relation to social debate. For Williams the process and value of social debate was something to be explored, both in the present and historically, in all available forms of publication and communications media, as his other publications show. Particularly important in Williams’ intellectual formation, however, were relatively centralised groupings of speakers, ideas and institutions, including national media dominated by a relatively small number of newspapers, radio stations and television broadcasters, along with organised political forces such as mass-membership political parties and trades unions. The climate of social debate in which Williams viewed ideas as circulating (expressed in the complicated and changing meanings of ‘keywords’) was typically one of large, ideological blocks of language use, to which he brought his own forms of Marxist analysis.

The general approach to keywords developed by Williams still applies to the contemporary world, despite major and obvious social upheavals since the period in which Williams was working. The post-Cold War, post-9/11 world, with its deeply changed geopolitical structures and institutions, is a world of radically new communications media including the Internet, of reduced membership of political parties and trades unions, and of a reduced role in many societies for the state by comparison with other kinds of organisation (including commercial companies, NGOs, charities, and other kinds of association). The cohesion and well-being of such pluralistic societies is generally believed to be mediated by some kind of ‘public conversation’. Such a conversation, to the extent that one exists, simultaneously serves to express control and regulation, to build social legitimacy, to circulate opinion, and to give form to critical debate and interrogation of the ideas and aspirations of different membership groups. In terms of communication, these varied functions are all inherent in the celebrated notion of, to draw on two keywords, a ‘marketplace of ideas’. What is less clear, and invites research in relation to our contemporary ‘public conversation’, is which words now serve as its ‘keywords’, and how those words work in the changing context of always-on, omnipresent digital media.

Scroll to top ▲

History of words and contestation of their meanings

By comparison with work in linguistic semantics, Williams’ Keywords showed less interest in general mechanisms governing meaning change than in the pressures under which people or social factions extend and transform word meanings.

In this respect, Williams introduced a new emphasis into a tradition of work on word meaning towards which he expressed at best mixed feelings: that of earlier, American ‘General Semantics’.Williams’ work, by contrasted, accentuated adversarial uses of language within the using society, rather than changes between periods and the social causes of those changes. This altered emphasis reflected Williams’ view of society as an often conflictual interaction between different social forces (especially, for Williams, classes). Repeatedly in Keywords, to address the social contradictions inherent in meaning attribution, Williams uses the phrase, ‘there is then both controversy and complexity in the term’.

The most important instances of meaning change, in Williams’ perspective, tend to be ones in which change is motivated by valorisation of a word away from a previously available neutral descriptive sense. The change either gives the word a positive or promotional ‘spin’, in a kind of political or ideological advocacy, or the word begins to occur ‘with a derogatory implication’. Williams in this way emphasised a further element in the analysis of keywords: some words are not only significant and ‘complex’; they are also ‘controversial’ and ‘contested’. Their meaning is not (or not only) a matter of distinct, technical senses that may initially seem opaque but can be understood if you have the right glossary. Rather, Williams’ analysis of keywords explores interlocking, contemporary meanings whose interaction inevitably remains difficult and unresolved.

To understand a ‘keyword’, in this extended sense, you have to engage not only with what the word can mean on its own, but also with its relations with other, similarly complex words. Is it typically opposed to or contrasted with specific other words in a given context, for instance, in a way that accentuates or obscures some aspect of its wider meaning or alternative meanings? Because keywords are embedded in social patterns of thought and argument, and in fact serve to create specific directions of thought and argument,  they are  inescapable. They form important building blocks of cultural understanding that can be presumed, relied on, or challenged.

When encountered in earlier discourse, keywords are the material of other people’s thinking that you need to analyze and evaluate if you are to reach your own view of something. But as regards use in the present, those same words are tools available for further, new thinking you might engage in: they are the material resources out of which new concepts and developments of thought will be formed. Confuse the meanings of a keyword, on this understanding of keywords, and you may fail to grasp the historical contours of a debate you are involved in, or the state of its current arguments. This in turn will restrict your ability to contribute to discussion of whatever version of a better future you may wish to advocate. By engaging with the complex meanings of keywords rather than only with topics they are used to discourse about, on the other hand, it is possible to disentangle genuine misunderstandings, expose fallacious or deceptive rhetoric where such words are used disingenuously, and move discussion and understanding forward on a more secure footing.

Scroll to top ▲

Keyword lists: updating and changing

When the second edition of Keywords was published in 1983, notes had been added on an additional 21 words (including anthropology, development, ecology, ethnic, exploitation, jargon, racial, sex, technology , and western) to Williams’ list of 110 entries for the 1976 first edition (which had included, among other words, alienation, bourgeois, commercial, culture, equality, hegemony, humanity, idealism, liberal, media, modern, nature, radical, reform, socialist, utilitarian, wealth, welfare, and violence).

Some words added for the second edition, Williams says in the ‘Preface’ to that edition, had been re-introduced from an original list of words he had begun researching during the 1950s. Others had 'become important in the period between that original list and the present time'. With Williams' death in January 1988, the process of revision of existing words - and possible extension of the list - came to an end.

The question therefore remains open whether a further tranche of words would be needed to reflect the original ambitions of Williams’ work, or whether the already recognised body of 131 words, having been central (as Williams saw it) to the formation of ideas and values over several centuries, would remain broadly our keywords still.

From 2005, the Keywords Project group began to meet informally to discuss the possibility of an updated version of Keywords. After preliminary seminars (at the University of Exeter and University of Sheffield, UK, in 2005 and 2006), the group began compiling a new Keywords that would update Williams' entries and wordlist and would also seek to apply modern corpus search technologies in finding and collating examples of the words selected.

A new corpus Adobe Acrobat Icon of keywords to work on was identified by the group, which added some 50 currently difficult words to Williams’ list (e.g network, globalization, customer). The new list also dropped approximately 50 other words that had either settled in their contemporary meanings or lost their topicality (e.g. alienation, existential, bourgeois). With a provisional list of headwords to investigate in place, the group began developing a web presence to allow cumulative publication of entries as they were written.

The process of devising the initial list was written up in an article which also contained more general reflection on criteria that seem important in deciding to call a particular word a ‘keyword’.

Scroll to top ▲

Criteria regarding which words may be ‘keywords’

Over the period in which research has been conducted by the group into particular words, a general sense has developed of helpful criteria in distinguishing which words are likely to function as keywords. A keyword is likely, for example, to be:

1) Currently used

Keywords are typically words with both a ‘popular’ and ‘technical’ distribution. They are commonly used to express and negotiate meanings in day-to-day discourse, while often also implying a claim to authority derived from one or more discipline-specific uses. Conversely, if a technical word is not commonly used outside a specialised literature, it is unlikely to have acquired the combination of technical and general senses that make it complex and potentially confusing. Some words may be far from obsolete in the general vocabulary but nevertheless obsolete as keywords, because of narrowing in their distribution.  But caution should be exercised in applying this criterion. Any test adopted for ‘current’ use cannot be reduced to a simple frequency count. ‘Current use’ for a keyword needs to be assessed taking into account some notion of prominence or salience in a given context, and across a corpus combining ‘opinion forming’ media (such as news and documentaries, feature articles, phone-ins, blogs, tweeting) as well as relevant academic or professional literatures (e.g. in fields such as psychology, business, science, law or literary criticism).

2) Polysemous

Keywords seem typically to be words that are construed differently on different occasions of use. In clearest cases, this is a matter of polysemy, or  relatively distinct senses any one of which closes off other meanings that the word is capable of conveying in different circumstances. In other cases, the word is vague, or a ‘wide word’ (that is, a word whose meaning is attributed in use by modulating – especially narrowing – a broad meaning so that the result appears focused and relevant to the surrounding discourse and context), Under some conditions of use, however – perhaps especially where words are used in highly abstract contexts, or by users who differ substantially in social or educational background or core beliefs, or where use of the word is openly adversarial – different aspects of overall meaning potential are activated simultaneously and lead to misunderstanding or confusion.

3) Categorical

Keywords are typically words used to designate – in effect, to carve out - social or cultural concepts and practices. They are especially influential because they give recognised verbal identity to, or ‘lexicalise’, social practices, beliefs, value systems, and preferences. Often such words are nouns (though in most cases they also have other, related word forms.) Typically, keywords are less likely to be concrete terms (e.g. terms that denote particular types of meal, song form, sport or style of clothing) than relatively abstract names for general practices, theories or standards of judgment. This is because keywords define and comment on culture and society, rather than functioning as part of the fabric of day-to-day interaction and local transaction. The contrast is not simply abstract / concrete, however, with one type of word elevated or conceptual, and the other embedded in everyday social practice: both types of word are embedded in and give form to social practice of different kinds, often in different spheres and sometimes on different scales. Exactly which form of a given word is the most ‘key’ is a matter of judgement. Sometimes it is a matter simply of which form occurs most frequently in non-technical discourse; with other words there may be a specific reason why one form cuts across different kinds of discourse with greater ambiguity or complexity of meaning than another variant of the same term.

4) Actively contested

Keywords are typically words that play a role in some kind of social debate or dispute.  Such debates and disputes differ, however, from controversies between schools of academic, technical or professional doctrine in that they involve more widespread, ‘popular’ viewpoints and argument.  The test of whether a word is sufficiently contested to make it a ‘keyword’ is in part whether the disagreement it creates circulates in the public domain beyond any given professional field, for instance as part of what politicians and media programmers like to call a ‘national debate’.

5) Part of a cluster of interrelated words which typically co-occur

As well as having multiple senses, keywords typically function either as part of a group of interrelated words which are, together, the terminology used in debate surrounding a particular topic, or as the principal word in a semantic field containing other related words, or cognates. With this criterion of ‘clustering’, the interest of a proposed word turns on its relations to the other words: how it relates to those other words by which it may be substituted in a given context, or of which it represents a type or part (as a hyponym or meronym); whether it conventionally serves as a euphemism or other kind of circumlocution for a given other word; or how it relates to other words in the same semantic field either as an opposite or as part of some scale of value or importance.


Such criteria are only guidelines. Under a wide interpretation of 'keyword', it seems arguable that all words are 'keywords' at least to some degree. The history and present polysemy or vagueness of very many words reflect characteristics that are general and unavoidable. Selection of what constitutes a 'keyword', over and above this characteristic, may therefore reflect social interests (and possibly the particular interests of the analyst) as much as anything inherent in a given word. On the other hand, a narrower interpretation of what constitutes a keyword will also emphasise a word's social as much as its linguistic importance, either within a particular area of debate (e.g. how we conceive management, or what we understand by education) or in general social debate in societies typically now mediated less by national political and media institutions and more by dispersed and increasingly user-selected online communications.

In the course of the Keywords Project so far, the process of questioning and in local ways revising our 'word-list in progress' has been more or less continuous. Changes are made both in the light of experience in writing and evaluating particular entries, and as the result of feedback gained during more public presentations — at which participants are sometimes invited to put forward their own keyword lists. Some words we have investigated, for example, turn out to be extremely important words in contemporary political or cultural argument because a relatively well-understood concept they convey is vigorously contested, but not because the word itself is multi-faceted or complex; such words we reject as keywords, in that linguistic analysis is unlikely to contribute much to understanding how they are used or the opposing arguments and positions with which they are associated.

Scroll to top ▲