Epilogue to Bilingual Language and Thinking

Vivian Cook  Multi-competence site  Other papers

Newcastle University

Unpublished: originally intended as an epilogue to Cook & Bassetti Language and Bilingual Cognition (2011) but not published for several reasons, one being that it seemed to end the book on too negative a note. The first part on language I have probably done to death in various later papers but the second devil's advocate part on the restricted scope and misunderstanding of linguistics in bilingual cognition research has not appeared in any form elsewhere I think, nor been said by anyone else.

While the prologue to this book focused on thinking, the epilogue concentrates on language. Most research into language and thinking has been conducted from a psychological rather than a linguistic perspective. The present epilogue argues the need to pin down the language side of the language/thinking relationship. Though this is central to the structure-centred approach to language and thinking (Lucy, 1997), a working concept of language is still necessary whichever approach is adopted. The epilogue tries to make two simple points: language is many things to many people; research connecting language to thinking should use an adequate analysis of language.

The discussion of language and thinking inevitably hinges on what the words language and thinking are taken to mean. There are doubtless as many linguistic models of language as there are psychological theories of cognition. Investigating the relationship between language and thinking therefore means being explicit about the nature of both language and thinking and being aware of the alternative versions of both. Any research in effect tests whether an aspect of cognition relative to a particular psychological theory correlates with an aspect of language relative to a particular linguistic theory.

But there are obvious dangers in such marriages of convenience. The psychological and linguistic theories are selected to favour the researcher’s ends, whether for or against the language/thinking link, in a sense inevitably since the partners need enough in common to form a relationship. Some psychological theories deny any distinctive role to language; some linguistic theories put barriers between language and cognition. In both cases the language/thinking relationship can be discounted in advance.

Linguists and psychologists are both primarily responsible for their own side of the fence and may deny any duty to reconcile it with the view from their neighbours’ side. Both linguists’ views of thinking and psychologists’ views of language are probably regarded as naïve by their counterparts in the other discipline. Cook (2008) criticised the reliance of psycholinguists on a small range of linguistic terms, such as phoneme and word, that many linguists have questioned or rejected in recent years, and their use of terms from linguistics; for example morphosyntax is used as a synonym for morphology (grammar below the word) plus syntax (grammar above the word), traditionally known to linguists simply as grammar, rather than the small area where syntax and morphology overlap, i.e. how inflections can be used in syntax. Any credible outcome from this debate has to be couched in terms that are sound for both language and thinking; if either partner is ineligible then the partnership is invalid.

1. meanings of language

A starting point is the word language, virtually taken for granted by the contributors to this volume. Yet language is not a word with a single unambiguous interpretation: it means what a particular theory says it means. Nor are the meanings of the English word languagenecessarily found in other languages: the three-way distinction in French between langue,  langage and parole (de Saussure, 1915/1976) has always proved a problem for English translators and linguists. In the field of SLA research, Cook (2010) has distinguished the six different meanings of language seen in the table below, which will be used to organise the discussion here. These are working definitions rather than watertight boxes and doubtless overlap and contradict each other in various ways; they are dealt with at greater length in Cook (2010).


a human representation system

‘human language’


an abstract external entity

‘the English language’


a set of actual or potential sentences

‘the language of Shakespeare’


the possession of a community

‘the language of English people’


the knowledge in the mind of an individual

‘I know English’


a form of action

‘language is doing’

Table 1. Six meanings of language (based on Cook, 2010)

– The Lang1 sense of language as a human representation systems treats it as a possession of human beings: ‘a species-unique format for cognitive representation’ (Tomasello, 2003: 13). Lang1 language is an uncountable noun; you have language or you don’t but you don’t have a Lang1 language. Here the language/thinking relationship is at its most general: does human language itself have a connection to human thinking? The semantic primitives of Wierzbicka (1996) such as part, kind (the relationship between things) or big, small (the size of things) would be prime candidates for a central inalienable core of human language and human thinking. Lucy (1997: 292) sees this as a semiotic level at which ‘speaking any natural language at all may influence thinking’. To Evans (this volume) ‘due to the nature of our bodies, including our neuro-anatomical architecture, we have a species-specific view of the world’. In this sense human beings themselves are incapable of examining the links between their language and their thinking: only a non-human intelligence not bound by human language and thinking might be able to detect them. Exploring the limitations of human thinking might involve science fiction solutions of seeing ourselves through the eyes of another intelligent species or an intelligent machine, as speculated on in Brooke’s epilogue to this volume.

The multi-competence perspective sees most of the human race as possessing languages, not language. It cannot be assumed that a typical human being has acquired only a single language with a single grammar, a single mental lexicon, and so forth: all human beings potentially, and the majority of them actually, possess more than one language (Cook, 2009). The question is, not how knowing a single language relates to our thinking, but how knowing two or more languages relates to thinking. Perhaps the general links of Lang1 to thinking occur regardless of how many languages the person knows; perhaps, however, the ordinary second language (L2) user connects language to thinking in ways that monolinguals do not.

– The Lang2 sense ‘an abstract external entity’ sees language as existing in the world of abstractions, like the rules of football; it is described in rulebooks such as the dictionary and the grammar, an entity in Popper’s third world (Popper, 1972). So the English language is captured in grammars such as the Comprehensive Grammar of the English Language (Quirk et al, 1972) and dictionaries such as the Oxford English Dictionary (OED, 1997); the French language is regulated by the French Academy and defined in Le Petit Robert (Robert, 2006); seven governments agreed in 1990 on a protocol for reforming Portuguese spelling.

Lang2 has been used to justify wars of territoriality, as in Hitler’s claims to German-speaking areas of Europe, fostered as part of the independence movement for minority groups such as Catalans, imposed upon conquered territories as in the imposition of Japanese names on Koreans in 1939, and used as a unifying lingua franca to fight against its native speakers as in the Black People’s Convention in South Africa (Biko, 1978). (Cook, 2010).

Anything that has ever occurred in the language anywhere is part of Lang2: every word since 1150 is in principle in the pages of the OED. In this sense people lament the decline of English or talk about its beauty and richness, and languages die – or are born, though birth seldom seems to be celebrated apart from Nicaraguan Sign Language (Senghas et al, 2004).

Lang2 is dangerously confusable with the individual’s knowledge of language, described below as Lang5. For Lang2 is not the same as individual knowledge in scope – who could actually know the 600,000 odd words and the 430 meanings of set included in the OED or the contents of all 1779 pages of the Comprehensive Grammar? Nor does it correspond to any individual’s actual usage, which will comprise at best a small subset of the words and grammatical rules in the Lang2 language. In addition it sweeps dialect speakers under the carpet in favour of a single standard, usually the variety spoken by a status group from one area or one class: Geordie speakers who distinguish singular you from plural yous will not figure, along with the vast majority of speakers of English in the UK  – only 10% of people in England speak with the standard RP accent (Wells, 1982). Nor does the form of the grammarian’s description have a necessary connection to the knowledge stored in the brain; it is highly unlikely that the rules of the grammar-book correspond directly to the psychological systems and processes in individuals’ minds, let alone to the way they are stored in their brains.

Lang2 is language as a countable noun; there is the English language, the French language … up to the 6,909 living languages in Ethnologue (Gordon, 2005). Hence the question can be asked whether users of Lang2 A think differently from users of Lang2 B, just as it might be asked whether people in countries with Common Law legal systems behave differently from those in countries with the Napoleonic Code. This is a matter of correspondence between an idealised language object and idealised thinking, typical of early suggestions from Whorf about Hopi and Algonkian languages (Whorf, 1941a/1956).

One strain of Lang2 analysis can be called philosopher’s grammar; it consists of someone of high intelligence, high sensitivity to language and high knowledge of linguistic theory applying these attributes to the analysis of language, essentially while sitting in their armchair and looking at a small sample of specimen sentences that come to hand. The prime source of insights is the mind of the investigator, supplemented by observations of people, texts or experiments, as in the great grammars of Jespersen (1933) and Zandvoort (1957). An outstanding modern exponent is Talmy, whose analyses of language deal with an ideal object, and usually with more than one language; the evidence in Talmy (2005) consists of thought experiments describing boards lying across streams and personal communications about Dutch and Makah rather than descriptive grammars or sentences collected in a corpus. Such analysis depends on the brilliance of its instigator; Talmy’s (1985) insight into verb-framed and satellite-framed languages has become a pillar in the language/thinking debate, as we see from papers in this volume (Czechowska & Ewert; Hendriks & Hickmann; Evans), though the neat either/or dichotomy between languages seems to have turned into a matter of more or less (Slobin, 2004): even in English it sometimes comes down to a stylistic choice based on level of formality between a verb and a verb plus preposition, enter/come in, insert/push in or land/touch down, often reflecting the French and Old English strata in English vocabulary. It is nonetheless an insight about Lang2, the abstract third world entity.

Necessarily these third world Lang2 entities have been single languages rather than in-between languages like pidgins (Romaine, 1988). Yet corpora and grammars are now starting to emerge for varieties of English such as English as Lingua Franca (ELF) that do not have native speakers (Seidlhofer, 2004). But these are descriptions of native-speaker-less languages, not authoritative statements of the Lang2 of a multilingual nation or group, and have indeed been attacked for undermining the ‘standards’ of such national communities. The Lang2 sense is equally remote from the L2 user; an individual L2 user no more knows a Lang2 in either language than a monolingual.

Lang3 is language as ‘a set of actual or potential sentences’: ‘the totality of utterances that can be made in a speech-community’ (Bloomfield, 1926/1957: 26) or ‘a set (finite or infinite) of sentences, each finite in length and constructed out of a finite set of elements’ (Chomsky, 1957: 13): a language is a corpus of sentences that have been spoken or written. This approximates to the sense of language in usage-based connectionist and emergentist studies in psychology and Conversation Analysis. This can be opposed to the Chomskyan notion that the goal of a grammar is to describe all the sentences that could be spoken or written – the creative aspect of language use, as described by Evans (this volume): ‘Formal linguistics, for instance, attempts to model language by positing explicit algorithmic procedures operating on theoretical primitives in order to produce all the possible grammatical sentences of a given language’. Lang3 descriptive grammars such as Biber et al (1999) reflect the properties of actual corpora; the COBUILD dictionary (1995) reports meanings from actual usage rather than the complete definitions found in the OED. The question of whether the set of sentences constituting language A goes with different thinking from the set constituting language B is, however, virtually unanswerable as language is being treated as a objective external object, not as an internal mental reality, and so the possibility of thinking as such does not arise.

A particular problem occurs with the Lang3 sense for research into L2 users (Cook, 2010.): how do you establish which of the L2 user’s sentences belong to language A, which to language B? Weinreich (1953: 7) said that ‘A structuralist theory of communication which distinguishes between speech and language ... necessarily assumes that every speech event belongs to a definite language’: but, in the Lang3 sense, how do you know which is which? That is to say without invoking other senses of language, such as whether the sentences use the grammar and vocabulary of language A or B, which turns it into a discussion of Lang2 or of one of the other senses. Bilingual speech therapy insists on observations of both the child’s first language and their second (Stow & Dodd, 2003). If Lang3 were important for the language/thinking relation, this issue of separating Lang3 languages would need to be treated explicitly.

– The Lang4 sense is language as ‘the possession of a community’: ‘The mental individuality of a people and the shape of its language are so intimately fused with one another, that if one were given, the other would have to be completely derivable from it’ (Humboldt, 1836/1999: 46). Lan­g4 is shared among a group of speakers such as ‘the English-speaking world’ or ‘native speakers of Chinese’. A language community is often equated with a nation – people born in Japan tend to speak Japanese. Nevertheless Sapir (1921: 179) insists ‘It is impossible to show that the form of a language has the slightest connection with national temperament’. It can also be a virtual community unconstrained by political borders (Anderson, 1983): Kurdish is spoken in Iraq, Turkey and Iran despite the lack of a country of Kurdistan (Gordon, 2005). A community may also be multilingual, using several languages for different functions in everyday life. In India for example everyone has to know Hindi and English, plus the local state language if the local state language is neither Hindi nor English, known as the ‘Three Language Formula’ 3±1 system (Laitin, 2000). And, as House (this volume) points out, there may be many sub-communities within such large groupings as nations.

The question for thinking and language research is then whether the social interaction of Lang4and the sense of identity it promotes in its speakers link in some way to their thinking: ‘languages reflect cultural preoccupations and ecological interests that are a direct and important part of the adaptive character of language and culture’ (Evans & Levinson, 2009). It might be comparatively easy to demonstrate this in terms of social interaction. The terms of respect in Japanese or the alternative pronouns for social status in Thai undoubtedly go with particular social values. Language is bound to mirror the way the society functions. But does language lead or follow? The disappearance of the Old English word sweostersunu (‘sister’s son’, i.e. nephew) attests to the decreased importance of the uncle/nephew bond but is hardly responsible for it.

Extending Lang4 to L2 users raises a long contested issue of how community goes with language. Does the individual L2 user effectively belong to two communities or do they belong to a different community specifically of L2 users, as in the case of ELF? On the one hand Mackey (1972: 554) claims ‘An individual’s use of two languages supposes the existence of two different language communities; it does not suppose the existence of a bilingual community’; on the other Brutt-Griffler (2002) proposes ‘the multi-competence of the community’ and Canagarajah (2007: 930) insists ‘Linguistic diversity is at the heart of multilingual communities.’ The Lang4 link between language and thinking needs to take multilingual communities into account, not just those which employ a single language.

– The Lang5 sense of language refers to an individual’s mental knowledge: ‘a language is a state of the faculty of language, an I‑language, in technical usage’ (Chomsky, 2005: 2). The classic competence/performance distinction (Chomsky, 1965; 1995) in part distinguishes Lang5 mental knowledge from Lang3 sets of sentences in so far as performance refers to ‘the actual use of language in concrete situations’. However Evans (this volume) points out ‘An important consequence of adopting the usage-based thesis is that there is no principled distinction between knowledge of language, and use of language (competence and performance, in generative grammar terms), since knowledge emerges from use’.

This sense of language is perhaps the one most used in language/thinking research, hardly surprisingly since it involves a concept of mind: the crucial connection is between individual mental knowledge of language and individuals’ thinking. In methodological terms, a description of Lang5 individual knowledge is neither the same as the general rules of the language described in Lang2 nor as the pattern regularities in the individual’s Lang3. The valid relationship is between the individual’s linguistic competence and the individual’s thinking.

There is often therefore a tension between language as an abstract codified system, as a community possession and as knowledge in an individual mind. Many see the community Lang4 and the individual Lang5 meanings as two sides of the same coin: ‘although languages are thus the work of nations … they still remain the self-creations of individuals’ (Humboldt, 1836/1999: 44); ‘le langage a un côté individuel et un côté social, et l’on ne peut concevoir l’un sans l’autre’ (language has an individual side and a social side and one cannot imagine one without the other) (de Saussure, 1915/1976: 24). This is indeed the basis of functional grammar (e.g. Halliday, 1978). The slogan of interactionist psychology was B=f(P, E) (Lewin, 1936) – behaviour is a function of Person and Environment, translatable into ‘language is both internal Lang5 and external Lang4’. Yet whole theories of linguistics have based themselves exclusively on the internal Lang5 meaning, say Chomsky (1965), just as whole theories of psychology have based themselves on the external Lang3 sense, whether behaviourism, connectionism or emergentism.

The importance for the L2 user side of the language/thinking debate is the relationship between the two language systems in the same mind. Cook (2003b) talked of an integration continuum between the poles of total separation and total integration of the two languages, similar to Weinreich (1953)’s distinction between coordinate and compound bilingualism. Neither pole is completely achievable: total separation is impossible since the two languages coexist in the same mind; total integration is impossible as the speaker would be unable to control which language they were speaking. The point on the continuum at a given moment may vary according to the aspect of language involved, say phonology, syntax or semantics, and by the choice of bilingual or monolingual mode (Grosjean, 1998). So establishing the Lang5 of an individual to link to their thinking may be trickier when the more complex language system of the L2 user is involved. While it would be attractive to take the well-known Chomskyan line that we need to study the pure water of monolingualism rather than dip a bucket into the polluted River Charles of bilingualism, this does not work if bilingualism is everyone’s potential state and if bilingual minds are more common than monolingual ones.

Lang6 is ‘language as action’, as expressed say in Schegloff et al (2002: 5), ‘People use language and concomitant forms of conduct to do things, not only to transmit information’, long part of the British school of linguistics from Malinowski (1923)’s phatic communion through Austin (1962)’s performative utterances to Halliday (1978)’s functions of language. It also forms part of Vygotskyan theory in that language development consists of the child internalising external social action (Vygotsky, 1934/62).

Do the things that people do with language differ from one language to another? A major assumption of 1970s communicative language teaching was that people had to be taught to complain or to argue rather than the vocabulary and grammar for complaining or arguing. It survives in the Common European Framework (2001) for language teaching devised by the Council of Europe, which measures language proficiency through ‘can-do’ statements such as ‘I can order a meal’ and ‘I can recognise familiar names’. One interpretation of Lang6 for language/thinking research would be to see how different ways of speaking linked with different ways of acting, corresponding to the Lucy (1997)’s behaviour-centred approach), such as Bloom (1981) starting from the problems that Chinese speakers encounter with counterfactual statements – If I had stood for election, I would have won.

The problem of L2 users for the language/thinking debate again concerns whether we think monolingual action or L2 user action is the norm. Businessmen like the Arab and Danish L2 users of English described in Firth (2009) are not using English in the same way as their monolingual colleagues; they carrying out L2 acts that do not necessarily have monolingual counterparts and which they might well be incapable of doing in their first language. Similarly codeswitching is a resource available to L2 users that enables them to make subtle points about social status and topic by changing language (Romaine, 1994). Before linking L2 users’ language to their actions and their thinking, we need to know what is distinctive about them that is lacking from those of monolinguals.

To sum up, the links between language and thinking vary according to the sense of ‘language’ being used. A claim about abstract standard Lang2 is not the same as one about social community Lang4 or individual mental Lang5, particularly if L2 users and their communities are taken into account. For example Sapir says ‘we see and hear and otherwise experience very largely as we do because the language habits of our community predispose certain choices of interpretation’ (Sapir, cited in Whorf (1941a/1956: 134). Perhaps language/thinking research needs to take all of these aspects of language into account; perhaps it needs to concentrate on one or two. But it cannot duck the responsibility to specify exactly which meaning or meanings of language it is utilizing and whether one language is involved or more than one.

Take the idea of grammar, now apparently known to psycholinguists as morphosyntax.  The grammar of the Lang2 grammar book is not actually known to individual speakers. Firstly it is a generalization, based either on large numbers of sentence and people or on the insights of gifted grammarians; secondly it includes all the rules of ‘the language’ rather than the subset of the rules any individual might use; thirdly it is unlikely to be in a form that actually corresponds to the mental Lang5 knowledge of syntax in an individual’s mind; fourthly it does not include complex language systems known and used by L2 users. Many examples over the years have shown how out of step these three ‘grammar’s can be; the Lang2 book grammar of English will insist on the object case in non-subject positions yet the current UK Foreign Secretary says people like you and I; such crucial Chomskyan test-cases of syntax as the differences between eager/easy to please and the subjacency principle in *Who is that she met worrying?  turn out to be unknown to many of the native English-speaking population: the standard institutional Lang2 grammar reflects the Lang5 knowledge of individual speakers imperfectly. Any syntactic property that is tested for fit against one person’s thinking needs to come out of the Lang5grammar of that person, not out of a general Langgrammar.

2. Aspects of language in relationship to thinking

Language not only has many interpretations, it also has many facets, reflected in the spectrum of linguistic specialities from phonology to syntax to lexicology, and a dozen more. None of these boast a single uncontested theory or an agreed set of descriptive terms and categories. Phonology encompasses theories as diverse as optimality theory and lexical phonology, some relying on terms such as phoneme, distinctive feature or constraint ranking, some rejecting them out of hand. The relationship of language to thinking depends not only upon the meaning of language employed but also on the specific aspect of language involved and the methods of analyzing it. The links between optimality phonology and thinking are certainly likely to be different from those between componential semantics and thinking. Research into language and thinking needs careful specification of the aspects of language that are used. Out of the thousands of possible aspects of language, one or two are bound to link to some equally arbitrary aspect of thinking by chance, unless the choice is severely constrained by the theory or area of language.

syntactic categories and grammar

According toWhorf, the categories of noun and verb affect Standard Average European thinking by dividing the world into things and actions, unlike the thinking of Hopis; ‘English and similar tongues lead us to think of the universe as a collection of rather distinct objects and events corresponding to words… as goes our segmentation of the face of nature, so goes our Physics of the Cosmos’ (Whorf, 1941b/1956: 241). His nouns and verbs seem to be defined in traditional terms, represented, say, by the Port Royal Grammar (1753: 90), ‘those words, which signify the objects of our thoughts ... those, which signify the manner of thinking’ or by Cobbett (1819: 12-15), ‘Nouns are the names of persons and things ... Verbs express all the different actions and movements of all creatures and of all things, whether alive or dead’.

Linguists used to inveigh against such semantic definitions of parts of speech on the grounds of their woolliness (the noun fire is hardly an object, the verb seem hardly an action, etc), and overlap (request is both a noun and a verb, up can be a preposition up the hill, a noun ups and downs and a verb up the grant). No two people would agree on whether many words were nouns or verbs by these subjective definitions; at best there is a statistical tendency that a noun will refer to an object, a verb to an action (Lyons, 1966). Instead linguists have preferred to define syntactic categories in terms of structural properties; Fries (1952) defines Word Class 1 as words that can be inflected for number and can be preceded by a determiner; i.e. nouns are defined structurally rather than notionally. Other linguists have treated noun and verb as primitive terms – ‘certain fixed categories (Noun, Verb, etc) can be found in the syntactic representation of the sentences of any language’ (Chomsky, 1965: 28); these are then substantive universals – ‘items of a particular kind in any language must be drawn from a fixed class of items’ (ibid) – a built-in aspect of the human mind. Even Sapir (1921: 96) has a universal leaning: ‘There must be something to talk about and something must be said about this subject of discourse… No language wholly fails to distinguish noun and verb, though in particular cases the nature of the distinction may be an elusive one’. Oddly enough many of those who reject the Chomskyan idea of innate universals of language are quite happy to use Nouns and Verbs as self-evident universal categories. But even Chomsky’s definition refers to syntactic properties; it is an empirical question whether there are two groups of word-classes that can be isolated in all languages on the grounds of syntactic properties (Robins, 1952).

Much language/thinking research still relies on a semantic opposition of Noun and Verb, for example Evans (this volume) ‘… verb … designates entity as an event; noun … designates entity as an object…’ The view that nouns and verbs relate to things and actions is derived more from the semantics-based school tradition of grammar-teaching than from the linguistics tradition, in which the connection between parts of speech, i.e. syntactic categories, and thinking starts from their syntactic nature; as Sapir puts it, ‘A part of speech outside of the limitations of syntactic form is but a will o’ the wisp’ (Sapir, 1921: 96). Linking syntactic categories to thinking would involve first demonstrating that categories such as noun and verb are not universal, as implied say in Chomskyan theory but refuted by their absence in Straits Salish (Jelinek (1995) cited in Evans & Levinson (2009)) or by Nootka studied by Sapir himself (Robins, 1952), and secondly correlating thinking with syntactic definitions of the categories: what does syntactic rather than semantic nouniness correspond to in thinking? Researching the relationship of a syntactic category to thinking is not the same as investigating the effects of a semantic feature of a group of nouns. When Evans & Levinson (2009) say ‘each word class we add to the purported universal inventory would then need its own accompanying set of syntactic constraints’, they are putting the cart before the horse: a word class is precisely the label for a set of words that have the same syntactic constraints, not a concept in search of its syntactic characteristics.

Much language/thinking research with nouns has concentrated on the article system, alias determiners. In English the written and spoken forms of the articles are:  the /I~/, a~an /eI~~Qn~n/, and Ø (zero article, i.e. the name for the lack of an article). These precede the noun, or, as The Infant’s Grammar (1824) puts it,

… two articles small… stood at the door,
That when the Nouns came, they might run in before.

The main English article system is a complex intertwining of:

- definite/indefinite the man/a man (singular), Ø the men/men (plural),

- countable/uncountable a man man,

- singular/plural a man/the man/the men/Ø men

- first/second mention a man came in/the man spoke.

Some uncountables can be quantified with classifier-like phrases, a cup of tea or a pile of sugar. Subtle, through everyday, uses abound, say the function/place difference between he went to hospital and he went to the hospital, or the Ø article before certain professions; he’s professor of English/he’s a teacher of English. And phonology also enters in through the meaning difference between stressed and unstressed forms; John is the // man to watch versus John is the /I:/ man to watch.

The English article system is notoriously hard for speakers of Chinese and Japanese to acquire, as my postgraduate students demonstrate in everything they write; today’s marking yielded a research and an evidence. But it is not just mass versus count nouns that gives them problems nor the absence of articles from their first languages so much as the sheer complexity of the English article system, attested by the vast literature on the L2 acquisition of English articles, say Butler (2002), Master (1994) and Thomas (1989). Evans (this volume) makes a three-way distinction between a introducing ‘a referent which the hearer is held to be unable to readily identify’, a designating ‘a unitary instantiation of the referent’ and the introducing ‘a referent which the hearer is held to be able to readily identify’. This takes in the discourse context but does not mention the zero article, equally part of the article system, i.e. the plural form of a with unidentifiable nouns is zero a man came in/Ø men came in. But it does not include the other uses of a, say for countability.

Imai and Gentner (1997) rely on ‘the grammatical distinction between count noun and mass nouns’, i.e. countability. This distinction is signalled by the presence of a before book, a book, and its absence before sand, *a sand, and the plural form books but not *sands.On the one hand this reduces a complex syntactic meaning-related formal system to an either/or semantic choice. On the other it establishes two classes of nouns, count and mass, on semantic grounds rather than seeing these as syntactic features of nouns that vary continuously (Bollinger, 1969). It is always for instance possible to create countable forms for mass nouns with specialised meanings, the sands of time, the waters of the Nile, or uncountable forms for count nouns there is too much book in school. The syntactic properties of nouns are not seen as important in language/thing research, only the meanings associated with them, or rather a small range of the meanings of articles in English. While Imai & Gentner (1997) claim to be talking about ‘count/mass syntax’, they are actually talking about count/mass semantics: the syntax is far more complex.

Nor is it only nouns that are treated semantically rather than syntactically. Take the prepositions discussed in Coventry et al (this volume) where in is glossed as ‘containment’ and on as ‘having support’. The linguists Leech & Svartvik (1975: 83-85) talk of ‘in-type’ prepositions including into, in, out of, through involving an area or volume, and ‘on-type’ prepositions such as onto, on, off, across, along involving a line or a surface; volume and surface contrast in He was living on a desert island and He was born in Cuba. As with nouns, the linguist sees a complex network of meanings, usually finding it difficult to make hard and fast divisions between pairs such as in and on. Testing out the meanings of prepositions as meaning-pairs against thinking simplifies prepositions down to isolated atoms of meaning rather than components of many molecules; prepositions may contrast with many other prepositions not just with a preposition that is in some contexts its antonym, in a plane, on a plane, inside a plane, with a plane, inside a plane, by plane and so on, each with a meaning that contrasts with the others to a greater or lesser extent.

Let us take the small sample of in and on phrases in a beginners English coursebook (Cook, 1980):

London is a city in England, 4 o’clock in the afternoon, it arrives in London at 10.15, in my class three people speak French, in January

on Wednesdays, on the wall, on the map, end on/begin on, on your left, what’s on?

Most of these ins and ons are in fact temporal, in the afternoon and in January, not spatial. Other uses are directional, arrive in and on your left. The container/surface distinction is thus only one small facet of the use of in and on. Our knowledge of in and on is encyclopedic, to use Evans’s terms (this volume); even beginners in the language need to know multiple meanings for in and on. Do ‘container in’ and ‘surface on’ separate out sufficiently from this web of meanings to be testable cross-linguistically? Levinson (2003: 38) feels ‘in general it is hard to find any pair of spatial descriptors with the same denotation across languages’.

Primarily, preposition is a structural category, like noun, defined by its appearance before noun phrases – in a moment, on a whim; ‘Typically, prepositions function as the first constituent of a prepositional phrase’ (Greenbaum, 1996: 159). The contrast with other languages is not just between certain of the preposition’s meanings but also in how it behaves syntactically, at the extreme with the postpositions of Japanese that occur after the noun phrase nihon ni (Japan in) rather than before in Japan. Prepositions have a complex set of syntactic behaviours in English, going with particular verbs come on/come in and adjectives dependent on/disappointed in; a distinctive aspect of English grammar is the prevalence of phrasal and prepositional verbs, the syntactic generalization underlying Talmy’s particular claim for Satellite-based languages. One English peculiarity is the habit of prepositions getting stranded in movement-constructions, What was he talking about? Prepositions are not independent units like content words but have a syntactic function like articles; in many ways they sit in the middle of the continuum between closed-class function words and open-class content words. Evans (this volume) states, ‘the preposition at, together with an NP that encodes a temporal point, encodes a point-like degree of extension’: it is precisely the prepositional phrase consisting of Preposition + Noun Phrase that conveys the meaning, not the preposition in isolation. Unless cross-linguistic comparison takes such basic information on board, research results may be skewed by the other syntactic and semantic features of preposition oppositions like in and on across languages.

The basic syntactic differences between language have seldom featured in the language/thinking debate, such as the variable order of Subject, Verb and Object, the contrast between configurational languages with phrase structure and non-configurational languages without (Hale, 1983), the preposition/postposition split, or indeed the presence or absence of Subjects. Whorf himself claimed ‘Hopi can and does have verbs without subjects, a fact which may give that tongue potentialities, probably never to be developed, as a logical system for understanding some aspects of the universe’ (Whorf, 1941b/1956: 243). Since Chomsky (1981), this has been known as the pro‑drop or null subject parameter; 97% of languages allegedly have verbs without surface Subjects, the well-known exceptions being English, German and modern written French. Yet no-one appears to have correlated such a well-attested syntactic difference between languages with differences in thinking. One instance where a syntactic difference per se was indeed correlated with a thinking difference was Bloom (1981)’s attempt to link the presence of if-clauses in English such as If he had not gone to the forum that day, Caesar would have lived longer with ability to reason counterfactually, though Yeh and Gentner (2005) more recently have claimed the effect is related to processing.

To sum up, the relationship between language and thinking has often been conceived in terms of semantics, not of syntax, of the meanings of sentences not of their form, as if syntax were an unimportant appendix to language. Syntax is seldom treated in its own right as conveying syntactic meaning but as a crude set of semantic meanings; gender for instance is not seen as agreement between words in the sentence but as semantic versus arbitrary meaning. Yet the way that we think could be as influenced by the way we construct our sentence as by what we mean by our words, as the writing-direction research of Tversky et al (1991) suggests. It is one thing to deduce a relationship between syntax and thinking based on a formal syntactic feature, whether the effects of SVO (apart from Goldin-Meadow, So, Ozyurek, & Mylander ( 2008)), the categories of noun and verb or the presence or absence of articles; it is another to deduce a relationship from a semantic idea incorporated in a complex web of syntactic devices, such as gender and mass/count. The chapter in this volume by Gathercole shows how great a harvest such use of syntax can yield. What is being asked for is then in the spirit of Brown (1986: 482): ‘Relativity is the view that the cognitive processes of a human being – perception, memory, inference, deduction – vary with the structural characteristics – lexicon, morphology, syntax – of the language he speaks’.

And, to return to the L2 user issue, the syntax of L2 users is not the same as that of monolinguals, whether in their first or second languages as documented in many recent publications such as Jarvis and Pavlenko (2009) and Cook (2003a). Any research into language and thinking has to be on its guard either to study pure monolingual native speakers or to base itself on the syntax of L2 users.

Lexical items and sets

The bridge between language and thinking has usually been perceived as the lexicon. Famously Whorf (1940/1956: 216) based arguments for linguistic relativity on the many words for different kinds of snow in Inuit, the scarcity of ‘snow’ words in English and the reduction of ‘cold’, ‘ice’ and ‘snow’ to one word in Aztec. Yet Sapir (1921: 181) pointed out ‘the linguistic student should never make the mistake of identifying a language with a dictionary’.

The area of colour research investigates how the set of colour terms in the lexicon of a language relates to ‘objective’ measures of colour such as hue and saturation as seen in Athanasopoulos (this volume). Like syntax, there are a range of theories and tools for handling lexis and its importance varies according to one’s theory. For instance, while Chomsky’s theory is often seen as syntactic, its incarnation as the Minimalist Program treats syntactic structure as a projection from the properties of each lexical item in the sentence (Chomsky, 1995).

Overall, however, there is a major slippage between the way that many psychologists see vocabulary and that adopted by most linguists. One preliminary difference is already seen in the term lexical item used in the heading. The category of word prevalent in psychology is extremely hard to define except in the trivial, writing-based, sense of letters surrounded by spaces.The unit used by linguists is more often lexical item or lexical entry which may well be more than one word look up, say, or in front of, or lemma in which the various alternate forms are stripped off; it contains far more information than a word’s reference, say a verb’s argument structure and collocability. Mental Lang5 lexicons are organised in lexical items or entries, not in the words typically found in Lang2 dictionaries.

In psychology by and large, words are regarded as having single core meanings: dog means , not ‘to follow someone closely’ or ‘a complete flop’: one word, one meaning. Levinson (2003: 35) sums this up as the assumption that ‘corresponding to a lexical item is a single holistic concept’. Red and blue are taken to refer to particular colours/hue/saturation on some sort of visual scale or field rather than say to the Labour and Conservative parties, snooker balls or York City Football Club. Evans (this volume) for instance distinguishes red ink from red squirrel. This tells us rather little about the wider set of colour terms that red belongs to: in the case of squirrels, the opposition is between red and grey, in ink probably between red, blue, black and green, in roulette between red and black, in traffic lights between red, green and amber. A word has many meanings in different contexts and relates to the meanings of other words.

More crucially treating red as a unique visual perception only covers its denotative reference to colour rather than its many other uses; it assumes that there is a core aspect of the physical world denoted by the colour term however this may vary cross-linguistically. Defining the language component of language and thinking seems to mean enumerating what physical shade on the colour or hue scales etc the words actually refer to before relating them to different thinking. The central belief is that all that words do is refer to things and actions; hence the emphasis in psychologists’ research is on content nouns with clear (and usually drawable or at least visible) physical reference, as opposed to structure words like to with syntactic meaning or words with abstract meanings like nation. Imai and Gentner (1997) for example frequently refer to ‘names’ and ‘nouns’ as if they were the same – ‘substance names and object names’, ‘children initially learn object names rather than names for relations of properties’, ‘extending their names on the basis of shape’. Discussions of colours frequently talk of ‘colour naming’, for example Regier & Kay (2009). For most linguists there is no simple bond between an object in the world and the noun that refers to it: most nouns are not names; most nouns have multiple meanings; most meanings do not have clear denotations in the physical word; most meanings relate to other meanings rather than being independent – the basic structuralist claim; children’s language acquisition is only marginally a matter of acquiring names.

To linguists, not only do words with unambiguous visual referents form a small subset of the lexicon but also reference is only one aspect of meaning. Take the word blue. One aspect of blue is indeed the peculiar physical shade it refers to. But blue exists in a network of other information. For an English person, its five most common associations are sky, black, green, red, white, sea, colour, yellow, eyes, aristocracy (Edinburgh Word Association Thesaurus, no date) – ignoring the student who supplied me with kind of, an association only available to fans of the Miles Davis album Kind of Blue. While this confirms a relationship with other colours on the physical scale, it also relates blue to the superordinate level colour in Roschian terms (Rosch, 1977) and to adjective-noun collocations, blue sky/sea/eyes/blood, the syntagmatic level of association (Deese, 1966). Blue is not only an adjective but also a noun, as in Klein blue or Oxford Blue, or, in the plural, a kind of music the blues or the nick-name of Birmingham City and Chelsea football teams. Meanings can be deconstructed into features – blue has the feature [+colour] etc; looked at in networks (Cruse, 1986) – blue is the opposite of red in political talk; studied in collocations – blue joke, blue funk, turn blue, black and blue; and counted – blue is 7987th in frequency on the British National Corpus, red 11605th.

Doubtless the counter-argument is that blue has a central meaning from which all the others derive: see how people relate this meaning to thinking and that’s all that needs to be done. But some linguists precisely deny that words have central meanings; polysemy is the norm. Many of the everyday meanings of blue do not relate to its supposedly core referential meaning; why should a joke be blue (it’s yellow in Chinese)? An intellectual be a blue stocking? Rather the word blue has multiple meanings and connections, of which its denotative link to a quality in the physical world forms a small part. Relating referential meanings of limited sets like colour words to thinking takes in one small aspect of the multitude of possible relationships. It is not that it is unusual or untypical of words to have such a range of senses: cognitive blends are the norm (Fauconnier, 2003).


The intention of this epilogue was to point out that, despite the exciting and novel work presented in this book, we have not achieved a proper discussion of language and thinking unless we have spelled out what we mean by language. This is not to say that one interpretation of language or one descriptive view of syntax should be preferred; many alternative avenues can be explored, as Feyerabend (1975) would recommend. But it is dangerous to take language for granted; lack of explicitness in the discussion often means it falls back by default on folklore and common-sense rather than the scientific study of language practiced in the twenty-first century.

The intention was also to highlight the importance of including people with more than one language in the discussion of linguistic relativity, as already suggested in the Prologue to this volume. The chapters in this book admirably demonstrate the complex nature of thinking in L2 users and the lively research field that is growing up around it. The message is that L2 users have different language systems and think differently from monolinguals. Researchers have to accept that it is the rule to know more than one language rather than the exception; monolinguals are not the pure natural form of human language but people who have been unable to achieve the human potential of knowing more than one language. Future research on linguistic relativity will have to take its stand on this question, amply posed by the papers in this volume.


Anderson, B. (1983), Imagined Communities. New York: Verso

Austin, J.L. (1962), How to do Things with Words. Oxford: Oxford University Press

Biber, D., Finegan, E., Johansson, S., Conrad, S. & Leech, G. (1999), Longman Grammar of Spoken and Written English. Harlow: Longman

Biko, S. (1978), I Write what I Like. The Bowerdan Press. Reprinted by Penguin, London. 1998

Bloom, A.H. (1981), The linguistic shaping of thought: A study in the impact of language on thinking in China and the West. Hillsdale, NJ: Erlbaum Associates

Bloomfield, L. (1926), ‘A set of postulates for the science of language’, Language 2, 153‑64. Reprinted in M. Joos (ed.), Readings in Linguistics I. Chicago: University of Chicago Press, 1957

Bollinger, D. (1969), ‘Categories, features, attributes’,  Brno Studies in English, 8, 37-41

Brown, R. (1986), Social Psychology. New York: Free Press. 2nd edition

Brutt-Griffler, J. (2002), World English: a Study of its Development. Clevedon: Multilingual Matters

Butler, Y.G. (2002), ‘Second language learners’ theories on use of English articles’, Studies in second language acquisition, 24, 451-480

Canagarajah, S. (2007), ‘Lingua Franca English, multilingual communities and language acquisition’, Modern Language Journal, 91, 923-939

Chomsky, N. (1957), Syntactic Structures. The Hague: Mouton

Chomsky, N. (1965), Aspects of the Theory of Syntax. Boston, Mass.: MIT Press

Chomsky, N. (1981), Lectures on Government and Binding. Dordrecht, Foris

Chomsky, N. (1995), The Minimalist Program. Boston, Mass.: MIT Press New Horizons

Chomsky, N. (2005), ‘Three factors in language design’, Linguistic Inquiry. 36, 1, 1-22

Cobbett, W. (1819), A Grammar of the English Language. Reprinted Oxford University Press, 1984

COBUILD dictionary (1995), Collins COBUILD English Dictionary. Harper Collins

Cook, V.J. (1980), People and Places. Oxford: Pergamon Press

Cook, V.J. (ed.) (2003a), Effects of the Second Language on the First. Clevedon: Multilingual Matters

Cook, V.J. (2003b), 'Introduction: the changing L1 in the L2 user's mind'. In Cook (ed.), 1-18

Cook, V.J. (2008), ‘Review of Handbook of Bilingualism: Psycholinguistic Approaches. Ed. J.F. Kroll & A. M.B. de Groot’, Language, 84, 1, 196-199

Cook, V.J. (2009), ‘Multilingual Universal Grammar as the norm’. In I. Leung (ed.), Third Language Acquisition and Universal Grammar. Bristol: Multilingual Matters, 55-70

Cook, V.J. (2010), 'Prolegomena to second language learning'. In P. Seedhouse & S. Walsh (eds), Conceptualising Language Learning, Palgrave MacMillan, 6-22

Cook V.J. & Bassetti B. (eds.) (2011), Language and Bilingual Cognition. New York: Psychology Press,143–190

Cruse, D.A. (1986), Lexical Semantics.Cambridge: Cambridge University Press

Deese, J. (1966), The Structure of Associations in Language and Thought. John Hopkins Press

de Saussure, F. (1915/1976), Cours de Linguistique Générale. Ed. Bally, C. and A. Sechehaye (1915). Critical edition by T. de Maurio. Paris: Payothèque, Payot

Edinburgh Word Association Thesaurus (no date), http://www.eat.rl.ac.uk/, consulted 11/12/09

Evans, N. & Levinson, S. (2009), ‘The myth of language universals: language diversity and its importance for cognitive science’, Behavioral and Brain Sciences, 32, 429-448

Fauconnier, G. (2003), The Way We Think. Basic Books

Feyerabend, P. (1975), Against Method. London: Verso

Firth, A. (2009), ‘Doing not being a foreign language learner: English as a lingua franca in the workplace and (some) implications for SLA’, IRAL, 47, 127-156

Fries, C.C. (1952), The Structure of English. Harcourt Brace

Gentner, D. & Goldin-Meadow, S. (Eds.) (2003), Language in Mind: Advances in the Study of Language and Thought. Cambridge, MA: The MIT Press

Goldin-Meadow, S., So, W.-C., Ozyurek, A., & Mylander, C. (2008). The natural order of events: How speakers of different languages represent events nonverbally. Proceedings of the National Academy of Sciences, 105, 9163-9168.

Gordon, R.G. Jr. (ed.) (2005), Ethnologue: Languages of the World. 15th edition. Dallas, Tex.: SIL International. Online version: http://www.ethnologue.com/

Greenbaum, S. (1996), Oxford English Grammar.Oxford: Oxford University Press

Grosjean, F. (1998), ‘Transfer and language mode’, Bilingualism: Language and Cognition, 1, 3, 175-176

Hale, K. (1983), ‘Warlipiri and the grammar of non-configurational languages’, Natural Language and Linguistic Theory, 1, 5-47

Halliday, M.A.K. (1978), Language as Social Semiotic. London: Edward Arnold

Humboldt, W. von (1836/1999), On Language. Translated by P. Heath. Cambridge: Cambridge University Press

Imai, M. & Gentner, D. (1997), ‘A cross-linguistic study of early word meaning: universal ontology and linguistic influence’, Cognition, 62, 169-200

The Infant’s Grammar (1824), London: Corner of St. Paul’s Church-yard: Harris and Son. Reprinted Scolar Press (1977)

Jarvis, S. & Pavlenko, A. (2009), Crosslinguistic Influence in Language and Cognition. Abingdon: Routledge

Jelinek, E. (1995), ‘Quantification in Straits Salish’. In E. Bach, E. Jelinek, A. Kratzer & B. Partee (ed.), Quantification in Natural Languages. Kluwer, 487-540

Jespersen, O. (1933), Essentials of English Grammar. London: Allen and Unwin

Kuno, S. (1973), The Structure of the Japanese Language. Boston, Mass.: MIT Press

Laitin, D.D. (2000), ‘What is a language community?’, American Journal of Political Science, 44, 1, 142-145

Leech, G. & Svartvik, J. (1975), A Communicative Grammar of English. London: Longman

Levinson, S.C. (2003), ‘Language and mind: let’s get the issue right’, in Gentner and Goldin-Meadow (eds), Language in Mind. MIT Press, 25-46

Lewin, K. (1936), Principles of Topological Psychology. McGraw-Hill

Lucy, J.A. (1992), Grammatical categories and cognition: a case study of the linguistic relativity hypothesis. Cambridge: Cambridge University Press

Lucy, J.A. (1997), ‘Linguistic relativity’, Annual Review of Anthropology, 26, 291-392

Lyons, J. (1966), ‘Towards a “notional” theory of the “parts of speech”', Journal of Linguistics, 2, 2, 209-236

Mackey, W.F. (1972), ‘The description of bilingualism’, in Fishman, J.A. (ed.), Readings in the Sociology of Language. The Hague: Mouton, 554-584

Malinowski, B. (1923), ‘The problem of meaning in primitive languages’. In C.K. Ogden & I.A. Richards, The Meaning of Meaning. London: 296-336

Master, P. (1994), ‘The effect of systematic instruction on learning the English article system’. In T. Odlin (Ed.), Perspectives on pedagogical grammar. Cambridge: Cambridge University Press, pp. 229-252

Oxford English Dictionary (1997). 2nd revised edition. Oxford: Oxford University Press. Online at http://www.oed.com/

Popper, K.R. (1972), Objective Knowledge. Oxford: Clarendon Press

Port Royal (1753), A General and Rational Grammar, reprinted by Scolar Press, Menston, England 1968

Quirk, R., Greenbaum, S., Leech, G. & Svartvik, J. (1972), A Grammar of Contemporary English. London: Longman

Regier, T. & Kay, P. (2009), ‘Language, thought and color: Whorf was half right’, Trends in Cognitive Sciences, 13, 10, 439-446

Robert, P. (2006), Le Petit Robert: Dictionnaire De La Langue Francaise. 40th edition. Collins, Robert

Robins, R.H. (1952), ‘Noun and verb in universal grammar’, Language, 28, 289-298

Romaine, S. (1988), Pidgin and Creole Languages. Oxford: Blackwell

Romaine, S. (1995), Bilingualism. Oxford: Blackwell. 2nd edition

Rosch, E. (1977), ‘Human categorisation’. In N. Warren (ed .), Studies in Cross-Cultural Psychology, XXXXX

Sapir, E. (1921), Language: an Introduction to the Study of Speech. Reprinted no date by: Harcourt Brace and Co, New York

Schegloff, E.A., Koshik, I., Jacoby, S. & Olsher, A. (2002), ‘Conversation Analysis and Applied Linguistics’, Annual Review of Applied Linguistics, 22: 3-31

Seidlhofer, B. (2004), ‘Research perspectives on teaching English as a lingua franca’, Annual Review of Applied Linguistics, 24, 209-239

Senghas, A., Kita, S. & Asli Özyürek, A. (2004), ‘Children creating core properties of language: evidence from an emerging sign language in Nicaragua’, Science 17 September 2004, Vol. 305. no. 5691, pp. 1779-1782

Sera, M.D., Forbes, J., Burch, M.C. & Rodriquez, W. (2002), ‘When language affects cognition and when it does not: An analysis of grammatical gender and classification’, Journal of Experimental Psychology: General, 131, 377-397

Slobin, D.I. (2004), ‘The many ways to search for a frog: linguistic typology and the expression of motion events’. In Strömqvist, S. & Verhoeven, L. (eds.), Relating Events in Narrative Vol 2, Mahwah, N.J.: Lawrence Erlbaum, 2119-257

Stow, C. & Dodd, B. (2003), ‘Providing an equitable service to bilingual children in the UK: a review’, Int. J. Lang. Comm. Diss., 38, 4, 351-377

Talmy, L. (1985), ‘Lexicalization patterns: Semantic structure in lexical forms’. In T. Shopen (Ed.), Language typology and lexical description: Vol. 3. Grammatical categories and the lexicon. Cambridge: Cambridge University Press, 36-149

Talmy, L. (2005), ‘The fundamental system of spatial schemas in language’. In B. Hamp (ed), From Perception to Meaning: Image Schemas in Cognitive Linguistics. Mouton de Gruyter

Thomas, M. (1989), 'The acquisition of English articles by first- and second-language learners', Applied Psycholinguistics, 10, 335-355

Tomasello, M. (2003), Constructing a Language. Cambridge, Mass.: Harvard University Press

Tversky, B., Kugelmass, S. & Winter, A. (1991), ‘Cross-cultural and developmental trends in graphic productions’, Cognitive Psychology, 23, 515-557

Vygotsky, L.S. (1934/1962), Thought and Language. Cambridge, Mass.: MIT Press

Weinreich, U. (1953), Languages in Contact. The Hague: Mouton

Wells, J.C. (1982), Accents of English. Vol. 2 The British Isles. Cambridge: Cambridge University Press

Whorf, B.L. (1940/1956), ‘Science and linguistics’, Technology Review, 42, 8, 229-231, 247-248. Reprinted in Carroll (1956), 207-219

Whorf, B.L. (1941a/1956), ‘The relation of habitual thought and behavior to language’, in L. Spier (ed.), Language, Culture and Personality. Reprinted in Carroll (1956), 134-159

Whorf, B.L. (1941b/1956), Languages and logic'. Reprinted in Carroll (1956), 233-245

Wierzbicka, A. (1996), Semantics: Primes and Universals. Oxford: Oxford University Press

Yeh, D. & Gentner, D. (2005), ‘Reasoning counterfactually in Chinese: picking up the pieces’, Proceedings of the 27th Meeting of the Cognitive Science Society, 2410-2415

Zandvoort, R.W. (1957), A Handbook of English Grammar. London: Longmans