Skip to navigation – Site map

HomeIssues21PapersThe real polysemous meaning of re...


The real polysemous meaning of real: a study in lexical pragmatics

Olivier Simonin and Sarah Bourse


Our main objective is to provide an account of the meaning of the polysemous adjective real that is theoretically grounded in post-Gricean pragmatics (Carston [2002], Recanati [2003], Sperber & Wilson [1996]). Combining lexical semantics and linguistic pragmatics is not new (see, e.g., Depraetere [2014] and Carston [2021]) and we believe that it can lead to a better understanding of the actual use of lexemes in context – especially when they are polysemous. Drawing on two pragmatic mechanisms, modulation and disambiguation (or lack thereof), we show how they interact with the lexical content of real and illustrate with genuine occurrences taken from ICE-GB (all of which we systematically annotated semantically). Little attention has been paid to the adjective real in the literature (with some exceptions: Bolinger [1972], Magnusson [2003]). Real is an untypical adjective. Its meaning is scalar when it causes a potentially gradable head noun (Filippi-Deswelle [2014], Moreau [2022]) to be semantically adjusted by indicating that a high (or high enough) degree is reached for a property or set or properties associated with that noun while, syntactically, it shows a very strong bias towards the attributive function – which we measure within ICE-GB and then explain.

Top of page

Full text

We would like to thank Denis Jamet, two anonymous reviewers and Robyn Carston for their help and comments on previous versions of this paper.


1How should linguists define the meaning of the adjective real? Lexicographers set out to list all the possible senses taken up by one word, while some linguists try to reduce the meaning of lexical items to a single abstract, underlying value (e.g. Ruhl [1989]). The adjective real is polysemous, i.e. it conveys several senses that are semantically (and historically) related. Our goal is to propose a detailed, mainly qualitative semantic account of the occurrences of real found in an English corpus, consistent with theoretical advances in linguistic pragmatism and psycholinguistic research, and to determine whether there is any potential correlation between semantic and syntactic (i.e. positional) criteria. To do so, we use the stratified 1-million-word (late 20th C. British English) ICE-GB corpus. Since this corpus includes a substantial (60%) spoken component, it is helpful to observe uses of real as they occur in genuine discourse, with performance errors and approximations. Our account of polysemous real builds on semantic and lexical concepts (polysemy, scalarity), conceptions of disambiguation, modulation (a type of pragmatic adjustment to context), and manifestness (referring to content that is communicated more or less pointedly) in post-Gricean pragmatics, within a contextualist framework (Carston [2002], Recanati [2003], Sperber & Wilson [1996]).

2This paper is structured as follows. We start and look at lexicographic characterizations of the adjective real, reviewing and discussing recent work on polysemy (1.). We then develop our own conception of the general lexical structure of real and show how disambiguation and modulation interact with its lexical meaning (2.). Lastly, we present our corpus results and use our lexico-pragmatic account to explain why the adjective real has the unusual distribution that we observe (3.).

1. Theoretical issues and the treatment of polysemy

3Polysemy is the source of an ongoing theoretical debate. After presenting some lexicographic accounts of the polysemous adjective real (1.1.), we discuss the relatively new conception championed by post-Gricean contextualist pragmaticists and its implications (1.2.), devoting a whole subsection to the interaction between diachronic evolution and the pragmatic process of modulation (1.3.).

1.1. Lexicographical considerations and real polysemy

4The OED etymological note for real (as an adjective) suggests that the word real was initially borrowed from French (from Anglo-Norman or continental French):

Anglo-Norman real and Middle French reel, real (French réel) (adjective) (in legal use) that concerns things and not people (1283), actual, concrete (early 14th cent. or earlier in Anglo-Norman), material, objective (c1370), that actually exists (c1485; in French also true, genuine, authentic (1688)).

  • 1 Here is one example in English, illustrating the (originally French) meaning ‘material, objective’ (...)
  • 2 We are making no claim here on how subsequent senses of real precisely arose from legal use. This w (...)

5The very first historical use of real was then legal (‘relating to things’, the word form being ultimately built on Late Latin realis/res, ‘things’), and can still be found in lexical units like real estate (yet with a sense for real that has narrowed down). The other senses of the adjective real in English are ultimately derived from this initial, legal use, and all of them now make up a network of meanings that the word can denote, listed under the same heading in a dictionary. If you take the meanings of the word found in Middle English, they closely parallel those of medieval French1. For non-legal uses, the Middle English Dictionary gives the following definition: “real, actual, having physical existence; of a narrative: true, actual”. There is a potential inferential path connecting the very first historically documented senses of real, since if you are referring to things (as opposed to people), they are to be found out there in the world (otherwise you would not have mentioned them), actual, concrete, and so they are material and exist actually2. Lexicographers list senses for a given word under a single word heading when all the relevant senses and definitions are interconnected semantically. That word is a polyseme. On the other hand, when you have the same linguistic form (or forms, i.e. both phonological and graphic) with meanings that are unrelated, lexicographers consider that you have distinct words that happen to be homonyms. According to the OED, around 1400, real could be used in the sense ‘royal’, making it a partial homonym for the polyseme we are studying here (we say partial since, in point of fact, real was just a variant of rial or royal, other forms used for the same word).

  • 3 The quotation and the information given here is taken from (available on 18/12/2022): https://blog. (...)

6Let’s now look at the entry for the adjective real in the online Collins English Dictionary (CED henceforth), which we have chosen because it is based on the information contained in the Collins Corpus, “an analytical database of English with over 4.5 billion words”3, including the 650-million-word Bank of English corpus, first developed in the 1980s and then expanded.

  • 4 “9. Law of or relating to movable property, such as money” (‘personal’, CED)
  • 52. an optically formed reproduction of an object, such as one formed by a lens or mirror” (‘image’ (...)
  • 6 “3. a. (of an answer in a fugue) not having the same melodic intervals as the subject, so as to rem (...)
  • 7PHRASE If you say that a thing or event is the real thing, you mean that it is the thing or event (...)

Real (ˈrɪəl)
1. existing or occurring in the physical world; not imaginary, fictitious, or theoretical; actual
2. (prenominal) true; actual; not false: the real reason
3. (prenominal) deserving the name; rightly so called: a real friend
4. not artificial or simulated; genuine: real sympathy, real fur
5. (of food, etc.) traditionally made and having a distinct flavour: real ale, real cheese
6. philosophy existent or relating to actual existence (as opposed to nonexistent, potential, contingent, or apparent)
7. (prenominal) economics (of prices, incomes, wages, etc.) considered in terms of purchasing power rather than nominal currency value
8. (prenominal) denoting or relating to immovable property such as land and tenements
real property. Compare personal4
9. physics. Compare image (sense 2)5
10. mathematics involving or containing real numbers alone; having no imaginary part
11. music
a. (of the answer in a fugue) preserving the intervals as they appear in the subject
b. denoting a fugue as having such an answer. Compare tonal (sense 3)6
12. informal (intensifier) a real fool, a real genius
13. See the real thing7

  • 8 In the CED, some senses are characterized (sometimes loosely) rather than defined.

7Leaving aside technical definitions, whose content is related to the other ones even though they hold for a specific field or domain only (like music, philosophy, mathematics or physics), some characterizations8 may overlap in some cases, with several of them being potentially in competition at the same time: (2) and (4) are close indeed, and real sympathy could be described as “true; actual; not false”, while real reason might be thought of as “genuine”. The boundary between two senses may turn out to be fuzzy, and the meaning of a word when used in context oscillates between two or even several of them (for discussion, see Stammers [2008], subsections 2.1. and 2.2.1.). There is always an element of arbitrariness involved in choosing the number of senses for a polyseme, since lexicographers can opt for grouping some uses together under one sense only or postulating additional ones. The difficulty seems to be compounded for adjectives by the fact that “adjectives are notoriously hard to divide lexicographically into senses,” (Moon [1987: 179]), much more than nouns or verbs, at least.

  • 9 For our treatment of lexicalized phrases, see 3.1.

8Interestingly, even a simple list of senses, as the one provided here, may be underpinned by some hierarchical structure: sense (1) is arguably more basic, and senses (2) to (4) could perhaps be derived out of it (historically and/or conceptually – by metonymy here). Sense (5) can be seen as a development (with words relating to consumed goods) of (4), entrenching into word meaning a lexicalized inference or set of inferences: if some food or beverage is made with genuine, traditional ingredients in a time-honoured way, it will have a distinct flavour (it should be added that the word is normally used positively, approvingly). Senses (6) to (11) are technical (the sense of real as in real property being presumably economic or legal), (12) intuitively relates to the first, non-technical, senses (but how it does so remains to be explained) and (13) signals a cross-reference to a whole lexical unit including the word real. The CED does not make explicit any kind of structural organization between the various senses of real, except for (11b), a specific understanding of (11a) for fugues, and both are joined under (11). In contrast, the OED orders senses hierarchically, with senses and sub-senses (and even finer distinctions within sub-senses), but the groupings proposed might be criticized. Compare, for instance, I.1.a (‘Having an objective existence; actually existing physically as a thing, substantial; not imaginary’) and III.8 (‘Corresponding to actuality; true. Frequently in real fact, real story9). Some degree of redundancy or overlap is almost inevitably present, which is presumably unavoidable when tracking down the slightest variations in terms of meaning since the first attestations of the senses recorded.

9Furthermore, dictionary characterizations may sometimes be lacking in precision, like – in the CED – sense (12) for real: it can act as an intensifier before a noun, but apart from being regarded as “informal”, what difference does its use imply as opposed to that of true or genuine, which are quasi-synonyms? The words are not fully equivalent semantically, and it is important to be able to distinguish between them. All in all, however, one should remember that a dictionary is primarily intended for users to understand the meaning of words and lexemes in context, and that lexicographers are sometimes compelled to make (partly) arbitrary choices in the process, for which they do not need to claim any theoretical significance.

1.2. Linguistic pragmatics and polysemy

  • 10 The traditional distinction is fraught with some (not unsurmountable) difficulties. As Victorri & F (...)
  • 11 We suggest in the next section that some of the senses in a network may not be fully lexicalized, a (...)

10How are we to account for polysemy then? And, perhaps more importantly, is it necessary to distinguish between polysemy and homonymy10, or can we simply say that both involve the same kind of ambiguity? What do we lose if we consider that they simply call for an identical general procedure that selects one potential sense among several? The latter view has been vigorously defended by Michael Devitt [2021], and others who still accept the existence of a distinction between polysemy and homonymy (see, e.g., Camp [2006]). The various senses of a polyseme share some common features with one another (which does not mean that all of them share the same set) and so they have a Wittgensteinian family resemblance overall, making up what Langacker [1987: 370] calls a complex category, the (speakers’) knowledge of which “cannot be reduced to a single characterization”. This complex (semantic) category can be modelled as a network of related senses – including “an inventory of senses describing the expression’s conventional range of usage; the relationships these senses bear to one another; schemas expressing the generalizations supported by a given range of values; and specifications of distance and cognitive salience.” (Langacker [1991: 268])11.

  • 12 See, especially, Recanati [2002: 5-7], in which he refers (p. 5) the reader to Grice [1989: 25], wh (...)

11We will present several arguments discussed by Recanati [2017] and Carston [2021] that lend support to the view that polysemy still needs to be given a contextualist pragmatic account, in spite of the lexicalization of individual senses. They agree that the conventional senses of a polysemous lexical unit are historically connected by modulation, a pragmatic adjustment process by which meaning can be enriched or modified inferentially, yielding the main apparent message derived from explicit signs – or, to be more accurate, “what is said” in Recanati’s framework (drawing on Grice)12, and an explicature (or rather a potential set of explicatures) in relevance-theoretic terms (Sperber & Wilson [1996], Carston [2002]). Modulation is then concerned not with the derivation of implicatures but with determining the explicit content of utterances: it is a kind of pragmatic enrichment that is not linguistically mandated, in that it is unconstrained by the use of linguistic forms per se. To take up one of Bolinger’s examples in his discussion of real [1972: 134], reproduced here as (1):

  • 13 In the examples, we use bold type for the occurrences of real and we underline the elements of the (...)

(1) He is a real lawyer [actually he may be just a student], the way he goes about proving his case.13

  • 14 Disambiguation is the process whereby a lexical content is selected for a specific form over its po (...)
  • 15 Noveck and Sperber go on to write, just after that sentence [2007: 189]: “The speaker’s meaning is (...)

12The word lawyer is not quite adequate to describe a student who is not yet a lawyer, and for the right referent to be selected, its meaning is modulated to denote someone, let’s say, who has all the qualities commonly attributed to a lawyer, though they lack the official title. For the moment, it is sufficient to note that modulation, along with disambiguation and reference assignment, are pragmatic processes involved in yielding explicit content14. These pragmatic processes make it possible for hearers to derive the explicit content intended for the speaker’s words, which underdetermine the overall meaning of the utterance (including implicatures too). Put differently and strikingly, “linguistic expressions serve not to encode the speaker’s meaning but to indicate it” (Noveck & Sperber [2007: 189])15. The speaker’s meaning is inferred from the linguistic meaning of the words and expressions used taken together with the context.

13Now, let us review the arguments that are intended to show that modulation, and not just disambiguation between senses, needs to be invoked to account for polysemy adequately. As Recanati [2017: 383] observes:

[W]hat characterizes polysemous expressions is less a finite list of discrete senses (the sort of thing one posits for homonymous expressions) than an open-ended continuum of senses to which it is always possible to add in a creative manner (Recanati [2004: 134-135]).
This open-endedness/continuity is manifested in use and documented in corpus studies […], so when I say that language users ‘know’ the modulation relations that hold between the senses of a polysemous expression (i.e. that they do not merely know the senses themselves), this can be interpreted not in terms of intuitions they have, but rather in terms of abilities they manifest to exploit these modulation relations creatively in new contexts of use.

  • 16 We allow for the possibility that the boundary between the two processes may be fuzzy and that ther (...)

14In other words, the various senses of a polyseme are not only committed to the memory of speakers, but they are also understood to be linked by the (inferable) conceptual connection that underpins the initial modulation, a point which becomes obvious when one considers the open-endedness of senses. We would add that modulated senses may become more or less routinized, that is an initially new modulation may be facilitated, more readily accessible because the hearer has come across it previously. In other words, modulation can still take place when a sense is not fully lexicalized, and there may be a very thin line between modulation and disambiguation (sense selection) in such cases16. Even when a word is fully lexicalized, it may not be so for some individuals, who might still infer the modulated sense intended pragmatically (Carston [2021: 15-16]) and thereby learn it, as it were. For instance, sense (5) for real (illustrated by real ale and real cheese) in the CED can be inferred from the previous ones and its various uses in genuine communicative contexts, both by very young learners and foreign speakers. Similarly, speakers may have a separate, distinct representation for real estate as a lexical unit (i.e. the whole term real estate is a lexeme for those speakers) without knowing the etymology of real or the technical, legal sense which real sometimes take, and which can still be found in real estate. That legal sense is defined by the OED (II.7.c) as: “Being or consisting of immovable property, such as lands and anything erected on or attached to this” – as in real property, for instance, which speakers might acquire (if they do not know it) by resorting to analogy with the phrase real estate, rather than straightforward modulation from another sense that they have learned.

  • 17 Or even some underspecified meaning, although this is a much stronger claim, especially if applied (...)

15Furthermore, the psycholinguistic literature suggests different activation patterns for polysemy and homonymy. Polysemy is not a homogenous phenomenon and has been argued to come in two different kinds (with, potentially, additional refinements or distinctions to be made). Regular polysemy involves run-of-the-mill – usually metonymic – relations like container/contained for glass, for instance, and provides meaning extensions that are regular in so far as they are semantically productive, as opposed to irregular polysemy (as foot in the foot of a hill, which cannot felicitously be used in the foot of a chair). With some polysemes, in the case of regular polysemy, all the senses are jointly activated and their co-activation is sustained before the relevant meaning is selected, while with homonyms and irregular polysemy, different senses compete with one another and are subject to frequency effects and contextual bias (for a summary, see Carston [2021: 4-9]). With irregular polysemy however, in contrast to homonymy, there might be an initial activation of shared features (see Brocher et al. [2016, 2018])17 or shared “elements of encyclopaedic information” (Carston [2021: 18]), which hearers or readers use before selecting a specific sense. All this again points to a different treatment in kind between polysemy (or regular polysemy minimally) and homonymy.

16Carston [2021: 19] also brings some insightful qualifications to the idea that polysemy is merely the result of conventionalized modulation:

[N]ew senses (ad hoc concepts) for a word arise through standard relevance-based pragmatic inference and, thus, that many instances of semantic polysemy have their origin in online pragmatic processes of conceptual adjustment in context (a point that is not original to relevance theory, although the specific account of it is). I say “many” rather than “all” because there are two other sources of polysemy. One is explicitly stipulated definitions for specific purposes, as in scientific, medical or legal domains (e.g., the legal stipulation that the word “child” in England denotes anyone who has not yet reached their 18th birthday). The other, which is of greater interest here, is syntax. For instance, a noun may occur with count or mass phrasal syntax, giving rise to distinct senses (e.g., “I can see two rabbits,” “I don’t like rabbit”), each of which may, in turn, be pragmatically adjusted to provide further senses (e.g., narrowings of the general mass sense to the meat and the fur senses) […]

17Technical senses imposed by definitional fiat are still connected to the other senses, but they usually involve a narrowing or include stipulations avoiding (or trying to avoid) fuzziness. Senses can also conjure up a doctrine: for someone steeped into scholasticism, it makes sense to read about philosophers wondering whether their standpoint was nominal or real (OED sense I.3: “Relating or attached to the doctrine of the objective existence of universals” – this use being obsolete except perhaps for historians of ideas). We should also allow for senses borrowed via other languages: the English language first borrowed the German compound Realpolitick (which Bismarck used to refer to his way of doing politics, which could be described as realism or pragmatism in politics), which it is now possible to write real politics, with a meaning of real that could be retrieved by semantic backformation and used innovatively, as in real economics – which would mean a practical, cynical way of pursuing economic goals. In our opinion, Carston’s second main point, concerning syntax (or rather, to be more precise, finer distinctions for individual constituents here, illustrated with count/mass nouns), should be extended to the various semantic functions a word can perform within a grammatical category: an adjective like real can be used to denote a rather abstract property (‘being real’ – as in Those women were real), or to emphasize the magnitude of some features pertaining to the head noun that follows (as in He’s a real lawyer, with real acting as some kind of intensifier). The two distinct semantic functions delineate two broad families of senses.

1.3. Diachrony, pragmatics, and the meaning of a polyseme

  • 18 Seeking to interpret Saussure in the light of his unpublished (until recently) writings, Maniglier (...)

18The picture that emerges is that the senses of a polyseme are related via modulation – or the conceptual connections underpinning erstwhile modulations, when a given sense has become fully lexicalized. The meaning of a polyseme corresponds to a network of related senses, and some of them may still need to be backed up by pragmatic inferences as routinization has not yet led to full lexicalization. There is a degree of variation between individual speakers for their own semantic representation of polysemes, which may evolve through time. An understanding of the diachronic development of polysemous words and the mechanisms (including pragmatic ones) that contribute to semantic change is essential to appreciate the synchronic situation for any network of related senses, which is an idealized average among speakers at a given moment. The semantic representations of polysemes are subject to ongoing evolution and the Saussurean distinction between the diachronic and synchronic study of language is not so much a clear-cut dichotomy as a coin whose two sides are complementary ways of analyzing language: Saussure himself emphasized the continuity or lack of interruption in the way human languages are transmitted [2002: 152-153], since they are constantly “redefining themselves” (Maniglier [2006: 376])18 as they are being spoken.

  • 19 The influence of Anglo-Norman only started to wane long after the transmission of that French diale (...)
  • 20 The use of real with narratives in Middle English is a case in point (see the Middle English Dictio (...)

19We now wish to argue that pragmatic inferences generated by similar contexts of utterance may lead to the evolution of polysemes in ways that do not simply amount to the addition of new senses through modulation, definitional fiat (for scientific or technical acceptations), and changes in terms of syntactic, grammatical or formal semantic properties. Let us consider the first stages of the evolution of real and show how a polyseme’s network can be restructured as a result of semantic extension. As the OED etymological note makes it clear, the first sense of real that does not relate to legal possession denotes the property of being actual and concrete, while at the end of the Middle English period, the adjective could also be applied to signify true when characterizing a story (see 1.1.). After initial borrowing from Anglo-Norman (or continental French), the development of some of its (earliest)19 new senses may have been encouraged by ongoing linguistic contact with the donor language and taken up in English (a case of semantic borrowing: see Durkin [2014: 8-9]), or have naturally occurred as a result of modulation in English, or plausibly as a result of a combination of both factors. As the word real came to be used in non-legal contexts and sometimes rather loosely to cover new meanings, it could be applied to an increasing number of head nouns that did not necessarily pinpoint material, concrete referents20. The various senses that emerged (like “true; actual; not false” and “not artificial or simulated; genuine”, to quote the second and fourth senses listed in CED) could then be conflated into a broader one. We propose ‘grounded in the extralinguistic world’, a general meaning that can still be specified contextually, with modulation that interacts with the syntagmatic environment of real (including the noun it modifies) and partly routinized inferential paths. In the real purpose (‘true’), considering that the phrase is not fully lexicalized, the use of a definite article and the abstract word purpose suggest that the referent, though not directly present in the physical world, is still grounded in it, pertaining to the realm of human motivations abstracted from it.

  • 21 Broadening naturally leads to the same result. A kleenex can be used to refer to (presumbaly) any k (...)

20Modulation can usefully be thought of as operating in three different ways: it can either narrow the extension of the notion or concept a word refers to (by specification), broaden it, or simply modify that notion or concept without narrowing down or broadening its extension (or, perhaps, by doing both successively). Whenever narrowing occurs, the initial sense that was modulated is still present and conveyed, though it is less relevant and therefore not as manifestly communicated, and may even fail to be entertained if not relevant enough. Consider sense (4) for real in the CED – it “not artificial or simulated; genuine” – and the first example given: real sympathy. Even though sympathy is not directly physically present in the material world, it is grounded in it since feelings have psychological (and physiological) correlates. As opposed to a mere show of sympathy, real sympathy is understood to be heartfelt: real has the lexical sense ‘genuine’, which is presumably more frequent with abstract nouns like sympathy. Narrowing can thus create a hierarchical structure within the network of meanings associated to a polyseme21: a fairly general sense can be potentially activated (weakly, usually) while a much more relevant subsense is strongly activated. Narrowings are a subset of modulations and modulations may also be routinely encouraged or triggered when a word is applied to some specific semantic categories – for instance, when real modifies an uncountable term that denotes consumables, it is likely that what the CED defines as sense (5) will be selected: “(of food, etc.) traditionally made and having a distinct flavour.”

  • 22 Those new senses still require some pragmatic inferencing, before full lexicalization.

21To summarize, recognizing and grasping the pragmatic processes interacting with synchronic lexical representations not only make it possible to understand the progressive routinization of new senses22, but also shed some light on the construction of the semantic networks that polysemes conjure up.

2. The lexical meaning of real and pragmatic processes

  • 23 Among the 7 occurrences of the word real excluded, 3 of them pertain to the category of adverbs, an (...)
  • 24 A presentation of the overall structure of the corpus is available from the Survey of English Usage (...)

22To develop our lexico-pragmatic account of the use of real in British English, we started by selecting genuine occurrences systematically from the ICE-GB corpus and analyzed each one of them individually. We deliberately chose a corpus with a manageable size for our purpose. ICE-GB is a one-million word multi-layered, parsed and annotated corpus of British English (Nelson et al. [2002]). It was compiled between 1990 and 1998 and provides a fair sample of occurrences of real in late 20th century English: a search for the word “real” provided 211 hits, from which we had to exclude 7, yielding 204 occurrences to study23. The corpus includes a substantial (60%) spoken component, and not just written material (40%). This was important to be make it possible to observe online pragmatic adjustments in real, day-to-day conversation and speech24. Drawing on ICE-GB for examples, we first establish the existence of a major distinction between two families of senses for the adjective real and offer a classification for its main senses (2.1.). We then study and discuss the pragmatic processes of modulation (2.2.) and disambiguation – or lack thereof – (2.3.) for naturally-occurring instances of real.

2.1. The adjective real and its basic semantics

23Huddleston & Pullum [2002: 527, 532] observe that adjectives typically denote properties (i.e. many are descriptive) and involve gradability (the property is normally not a matter of all or nothing). But real is far from being a typical adjective. First, it can be used to denote a property that appears to be a matter of all-or-nothing (‘being grounded in the extralinguistic world’) and it can then hardly be modified by an adverb like very.

(2) And in the real world which we all have to live in <,> a dictator a tyrant like Saddam Hussein <,> is going to listen with great amusement to those who say well we’re going to simply use economic sanctions […] (<ICE-GB:S1B-035 #067:1:C>)

(3) Love of the transcendental image of the Virgin Mary was what gave birth to the glorious cathedrals of the Gothic world and the love of real women has inspired works of art as diverse as Dante’s Inferno and the Taj Mahal in India - let alone the beautiful metaphysics of the Divine Sophia (<ICE-GB:W1A-008 #056:1>)

  • 25 Today at the request of Mr Baker , they are discussing what Mr Baker has outlined to them when he s (...)
  • 26 “Adjectifs à « deux ordres-deux valeurs »”, quoting Blinkenberg [1969: 47 et passim]. To qualify Sc (...)
  • 27 Our translation of: “ni relationnels, ni qualificatifs” (Schnedecker [2002: 7]).

24Very is not acceptable before real in (2) and (3): the adjective real is not gradable. Either the world exists or it doesn’t, and women who inspired works of art either actually lived in the past or were images of women or ideas connected to a woman allegory instead. Of course, very can be used with real when the head noun’s actual presence is a matter of appreciation: ICE-GB contains 8 occurrences (around 4%) of very real with head nouns like question, tragedy, possibility, power, problem…, in which real is gradable as speakers turn an (initial) all-or-nothing sense into a gradable one to be emphatic in their assertion (‘This is a tragedy indeed’, ‘There is indeed a problem’…)25, as in “[…] he said a social and political explosion is a very real possibility” (<ICE-GB:S2B-047#029:1:A>). Even so, Schnedecker [2002: 6] classifies réel as an adjective with a single value within a binary system (‘réel’ or not)26 and includes it within the broader class of adjectives that are neither relational nor descriptive27. The property denoted by réel is not quite descriptive: it simply indicates that something exists, and this results in the difficulty of asking a simple question with How is…? (Comment est…?, see Schnedecker [2002: 9]). This test appears to work with the sense considered in the following examples given for real in English:

(2’) How is the world?
– [?] It is real.

(3’) How were the women who inspired works of art as diverse as Dante’s Inferno and the Taj Mahal in India?
– [?] They were real.

  • 28 In his account, Bolinger likens the use of adverbial real (for which “really is felt as a pure inte (...)

25Yet real, in English, has another use, which Bolinger [1972: 96] describes as “an intensifier of a degree noun”, which he illustrates with “He’s a real dumbhead” and “I had a real fright”. Dumbheadedness and being frightened are presumably scalar and a matter of degree, and real appears to function semantically, superficially at least, like the intensifiers really and so in “He’s really/so dumb” or “I was really/so frightened”28. In this use, real appears to contribute to meaning very differently.

26The adjective real in (2) and (3) is absolute and not relative (its meaning does not change depending on its head noun), as something is either out there, grounded in the extralinguistic world or not. In contrast, in Bolinger’s examples, real interacts with the head noun to yield meaning but the sense of the adjective real does not semantically adjust to the noun in the way adjectives (especially descriptive ones) normally do (see, e.g., Recanati [2010]). For instance, the interpretation of smallness varies if one is talking about a small giant or a small man, since the meaning of small is sensitive to the head noun it modifies (i.e. ‘small for a giant’ or ‘small for a human’ here). The same holds for non-gradable adjectives like blue which refers to a different kind of blue in a blue sea and a blue sky (the shade of blue naturally also depends on geographical location and lightening conditions). In the NPs a real dumbhead and a real fright, the head noun underpins properties that are gradable (idiocy and fear) and that are deemed to be present at a very high degree, but the adjective does not appear to change or to modulate its (intensificational) meaning on account of the head noun: its (absolute) intensifying function bears on the noun (or complex term) it modifies, without being altered semantically by it.

27For adjectives, gradability is not a matter of all-or-nothing, and pertains to adjectival uses rather than an adjective per se (Huddleston & Pullum [2002: 531-532]). There are gradable adjectives and adjectives that are not usually gradable (e.g. dead, true, alive…), but even adjectives not normally understood to be gradable can be coerced to express some kind of gradability (Charreyre [1997]). Similarly, nouns can apparently be coerced to show gradable properties, even though it remains important to make distinctions between various categories of nouns and different kinds of gradability (Schnedecker [2010]). As a consequence, there is a potential compatibility between nouns in general and the intensificational meaning of real. But what type of gradability is involved when a noun is combined with the adjective real?

  • 29 Our translation of: “ce qui est spécifique au degré dans la détermination nominale, c’est qu’il exp (...)

28According to Moreau [2022: §159], “what is specific to the expression of degrees in nominal complementation is that they point to the quantity of a given property”29, the property in question being expressed by the head noun (the property is presumably not a single trait but a bundle of features that can be regarded as forming together a complex property). Fillipi-Deswelle [2014] also shows that quantification can take place with a noun (with or without complements). Building on Culioli’s conception of notions (“complex systems representing physico-cultural properties” [1990: 50]), she argues that the scale against which quantity is measured is structured around the notional organizing (or “attracting”) centre, the pole of reference for any given notion. In any event, we now wish to demonstrate that, with real, there are in fact two distinct cases for which some notional scalarity is involved.

29We start with uses that are not clearly intensificational but for which, within the given notional type, there are tokens that are better or worse representatives for the whole class. Let us consider (4) and (5):

(4) A real development gap has opened up within the Third World therefore, in the strictest economic terms, since the 1970’s. (<ICE-GB:W1A-014 #010:1>)

(5) There can be no real unity for the Conservatives under the leadership of Margaret Thatcher. (<ICE-GB:W2E-003 #007:1>)

The characterization “that deserves the name” given by the CED (from sense 3) is apposite here to paraphrase the meaning of real. To use Culioli’s terminology for (4) and (5), the tokens considered (for development gap and unity) are to be found on one side of the notional boundary – being opposed to tokens that fall beyond the purview of the notion. Put differently, the tokens are acceptable, appropriate exemplars for the category they represent.

30Note that the nouns development gap (a compound) and unity refer to entities that do not manifest themselves directly in the extralinguistic world: they are not directly tangible and therefore a matter of interpretation. In point of fact, for any gradation to take place, it is often necessary that the noun phrase modified by real refers to an entity that is only indirectly reconstructed from elements physically present in the world out there. In both (4) and (5), although the sense ‘being grounded in the extralinguistic world’ is not foregrounded, it may be weakly conveyed – in relevance-theoretic terms, communicated with a low degree of manifestness (Sperber & Wilson [1996: 59-60]). By contrast, the obvious deduction from the material signs indicating the presence of a gap or party unity is communicatively prominent: ‘the characteristics of the referent denoted by the head noun X are such that they correspond to the general understanding of what an X is’. Naturally, the initial inference is fully lexicalized, and it is highly manifest in (4) and (5). For this (now lexicalized) sense, the inferential connection with the initial meaning, from which it is derived, is presumably still present as that weakly communicated meaning is still available.

31In addition, there is a natural bias towards a high degree construal of being part of a given notional category. Without contravening contextual clues, the token is then not just felt to fit a given lexical description, but also to be a very good exemplar standing for a particular notion (it is very close to its attracting or organizing centre). In other words, the properties that define the notion (or at least some that are deemed relevant) are found in a high degree, and the use of real is intensificational.

(6) And it’s a chance to bring back Alan Ball who’s uhm a real exponent and expert on Greek football (<ICE-GB:S2A-018 #073:1:A>)

(7) if a company is in real financial trouble then it becomes impossible no matter how well intentioned to actually afford the the various steps involved in research development and the other aspects of of innovation (<ICE-GB:S2A-037 #068:1:A>)

(6) and (7) illustrate that intensificational use. If, instead of (6), the speaker had said that after having had a greenhorn, a real expert was needed, hearers would not have understood that real was used as an intensifier. The same could be said if, instead of (7), the speaker had uttered: “I would say it’s real financial trouble, not just a fancied threat.”

32Metaphorical uses of nouns, as in Bolinger’s examples and in (8), highlight gradable features through metaphor, so that real can only be intensificational:

(8) He’s got a real inner steel and toughness (<ICE-GB:S2B-003 #048:1:F>)

What (8) communicates is not that actual steel is physically present in the person meant (virtually no steel being carried by human bodies), but that the willpower and resilience metaphorically associated with steel is abundantly found in him.

33Taking up the semantic criteria that have just been introduced, we distinguish between two main families of senses, depending on whether real has a scalar function operating over the term it modifies. On the one hand, there are occurrences that convey the underspecified meaning ‘grounded in the extralinguistic world’ (sense I) – see examples (2) and (3) – and many of its subsenses, which it subsumes. Some of these subsenses may involve gradability, but with respect to the property denoted by the adjective real, and not to properties denoted by the term the adjective modifies. On the other hand, other occurrences illustrate a scalar use of real that involves the term the adjective modifies: they imply a judgement about the fitness of the referent to represent the notional category denoted by the head noun – and, potentially, complements (sense II). Within this family, we distinguish between cases for which tokens are simply deemed by the speaker to represent the category adequately, appropriately – ‘appropriate use’ (sense II.a), illustrated by examples (4) and (5) – and those for which tokens instantiate to a high degree the properties thought of as typical or characteristic of the notional category – ‘intensificational use’ (sense II.b), exemplified by examples (6) to (8).

Figure 1. The two families of senses for the adjective real

Meaning of real

I. ‘grounded in the extralinguistic world’

II. Scalar interpretation (applied to the modified term):

a. appropriate use

b. intensificational use

2.2. Modulation and loose uses of real

34We have argued that as a new modulation and the pragmatic inferences that underpin it become increasingly routinized, the corresponding lexical sense becomes entrenched within the lexicon and the initial modulation more and more conventional, eventually leaving no trace except for the conceptual connection that can still be felt between the two senses (see 1.2. and 1.3.). Synchronically, the pragmatic process of modulation can especially be observed in everyday speech, in conversations and oralized texts (e.g. social letters and lecture notes in ICE-GB), whenever speakers use a word loosely (Carston [2002: 157-160, 320-359]), creating an ad hoc concept – an “occasion-specific sense” [Carston 2019: 150] – as they speak, in order to suit their communicative needs, as in (9):

(9) […] Malachi […] is an unlikely choice for any father to give his son so that one may really doubt whether the name Malachi is a real name (<ICE-GB:S2A-036 #084:1:A>)

35The meaning of real is modulated here and the word is used loosely by the speaker to point to names actually given to people when they live, as opposed to names that are invented or made up later for some purpose (Malachi means ‘messenger of God’ and the suggestion is that it was probably used by the writer to make a point). A loose use involves producing a word (or phrase) that does not fully fit the situation if understood literally (as in The steak is raw meaning that the steak is underdone, or France is a hexagon). It is a kind of modulation (it modulates word meaning), and idiosyncratic loose uses are of much interest to pragmaticists as they show how lexical meaning interacts with pragmatics in actual speech (or sometimes writing) situations. Contextual information helps the hearer reconstruct the intended occasion-specific meaning:

(10) Uh I like to a book to be real (<ICE-GB:S1A-016 #293:1:E>)

The speaker is using a very personal definition of real here, having just admitted to liking “to read about life, you know, as it was.” (10) is a partial reformulation and a specification of that statement (with a performance error as the first to needs to be deleted for the sentence to be grammatical). When resuming talking, the speaker adds: “I suppose that’s really why I don’t like the Lord of the Rings because I feel as if that’s what you create in your imagination.” (normalized transcription for: <ICE-GB:S1A-016 #298:1:A>). In (10), real contextually takes on the specific meaning of ‘relating to the extralinguistic world as we experience it in everyday life’, an ad hoc concept created on the spot and derived from sense I. This concept is made available to the hearer because the speaker highlights the contrast between books (i.e. stories, strictly speaking) whose action takes place in our real world or in a science-fictional or fantastic universe. A very similar use of real is found in (11), with the adjective referring to fiction dealing with the actual world, as opposed to fantasy:

(11) I I think unless you with uhm with something real <,> it’s uhm you you you end up having too many possibilities open to you and it just uhm it becomes a bit of a kind of a <,,> uhm a fantastically complicated mind maze really <,> (<ICE-GB:S1A-096 #030:1:B>)

36Another clear one-off, loose use of real is highly personal, as suggested by the quotation marks:

(12) “Too “real” an exhibition is still an artificial representation and is a coercive form of information.” (<ICE-GB:W1A-012 #044:1>)

By “too “real” an exhibition”, the writer means an exhibition that seeks (too much here) to avoid appealing to the imagination of viewers. An exhibition can more or less appeal to the audience’s imagination, and real is obviously gradable here.

37In those four examples (about 2% of all occurrences of the adjective real in our corpus), a loose use of real derived from sense I evokes an ad hoc concept forged by the speaker, illustrating the productivity of this type of modulation to generate meaning pragmatically, in the course of discourse production. In addition, there are numerous other instances for which one could be tempted to classify an occurrence of real as a case of occasion-specific loose use:

(13) I mean a real language (<ICE-GB:S1A-015 #187:1:A>)

  • 30 A speaker may even successively adopt different points of view concerning what properties are essen (...)

The conversation relates to languages one learns, and the speaker makes it clear that what is considered is the learning of a language “As different from English as Turkish is”. For that speaker, a “real” language is a language that is significantly different from English and other closely-related Indo-European languages, to present an actual cognitive challenge. The sense of real here is II.a (‘appropriate use’), and what is highly personal is not so much the use of the adjective real as a highly idiosyncratic understanding of what a proper language is, for the speaker’s current conversational purpose30.

38A subsense of real can also be selected by an implicit contrast with the discourse context (14) or the situation (15), involving modulation when the subsense does not appear to be fully lexicalized:

(14) However even in real tissue <,> what we are doing is we measure an attenuation and we’re looking for small changes in attenuation here delta A (<ICE-GB:S2A-053 #116:1:A>)

(15) This is not like a real situation of course (<ICE-GB:S1B-004 #136:1:A>)

In (14), natural tissue is contrasted with synthetic tissue, also discussed, and in (15), the speaker is a teacher in class with students confronted with an experimental situation (they have to provide some measurements), as opposed to an actual one. Artificial tissue and experimental situations can both be argued to be ‘grounded in the extralinguistic world’, as they are part of it physically (they are real in that sense), and yet they are presented as opposed to other tissue and situations that are found naturally, without human intervention – which is presumably already a subsense for real. Synthetic tissue and experimental situations are quite artificial and do not occur in the natural course of events. Even though the subsense ‘natural’ is arguably fully lexical (see sense 4 in the CED), the more precise meanings of real in (14) and (15) – grown by living organisms’ and ‘occurring in the course of events without reconstruction’ – are obtained via narrowing modulations yielding subsenses from the subsense ‘natural’, assuming that those two modulations here are only partially routinized (there is no modulation when a sense or subsense if fully lexicalized).

2.3. Ambiguity and disambiguation

39Disambiguation is the pragmatic process whereby irrelevant lexical meanings are discarded to determine explicit content: “what is said” (Recanati [2004]), explicatures (relevance theorists), or what Grice [1989: 119-120] called “applied timeless meaning”. When several meanings are potentially relevant at the same time, the process of disambiguation fails to select just one meaning but yields two or more, and ambiguity arises with several interpretations conveyed at various degrees of manifestness (see Simonin [2021]). Yet, we have seen that when a polyseme activates an underspecified meaning, viz. sense I, ‘grounded in the extralinguistic world’, for real, interpretation can home in on a more specific (sub)sense (see 1.3.).

40When analyzing occurrences of real within our corpus, we realized that some of them were genuinely ambiguous as they could not be reduced to a single semantic interpretation – i.e. either I, II.a or II.b. Here are a few examples that we deemed to be ambiguous:

(16) The inadequacies of our own relationship of father and son, the real but stifled affection, the things we never said to each other, the gestures felt but unmade, were no doubt responsible for my desire to find out what I flinched from. (<ICE-GB:W2F-014 #019:1>)

(17) But with a coordinating committee of pro-Moscow organisations <,> now calling for the dissolution of parliament <,> the authorities believe the threat of a military intervention similar to that in Lithuania has become too real to ignore <, (<ICE-GB:S2B-015 #064:1:F>)

(18) “Access” means only basic reference use of a library, and while it may well be common practice to include access to photocopying facilities, it will not normally cover use of inter-library loans, searching of electronic databases, nor even manual literature searches or other lengthy enquiries which impose a real burden on staff in other libraries. (<ICE-GB:W2D-006 #139:2>)

In (16), “the real but stifled affection” refers to actual affection grounded in the extralinguistic world. Yet, it can also be understood as something that might qualify as ‘affection’ properly (‘appropriate use’, II.a), and an idea of quantity is conveyed: enough warmth of feeling so that the sentiment can be adequately described as affection. (17) illustrates ambiguity between the senses of being ‘grounded in the extralinguistic world’ (I) and ‘intensificational use’ (II.b). The threat is felt as something tangible and concrete (I), owing to objective, material elements being well-known, leaving no room for divergent appreciation of whether the threat does exist or not. The threat is “too real to ignore”, it has gone over a point beyond which it has to be taken seriously. There is a high degree in the level of threat here in (17) and the interpretation of real is also intensificational (II.b). Ambiguity between ‘appropriate use’ (II.a) and ‘intensificational use’ (II.b) is illustrated by (18). The “real burden” in question might be a heavy burden (II.b) or a burden that deserves to be called so (II.a): it is difficult to tell which interpretation is meant, as both are apparently communicated to equivalent levels of manifestness. Several senses of real are likely to be relevant in a given utterance, making ambiguity frequent. The pragmatic procedure of disambiguation, in such cases, fails to pick just one lexical sense from the others and, instead, lets them be conveyed at various levels of manifestness depending on how relevant they are.

41After we had semantically interpreted each occurrence of real – each author individually at first – and then discussed them for the purpose of classification, we still found that 15 out of a total of 204 occurrences (7,35%) could not readily be reduced to a single interpretation and thus had to be classified as ambiguous, as two different meanings appeared to us to be just as relevant, without one meaning prevailing. Ambiguity is a very pervasive phenomenon when analyzing occurrences of the adjective real, even when one discounts cases in which a non-scalar subsense may also convey the sense ‘grounded in the extralinguistic world’ (I), which is less relevant (and so less manifestly communicated) than a more precise subsense (see 1.3.).

42There is also an interesting semantic configuration with real being used with an implicit contrast that makes one of its subsenses highly manifest while sense II.a is still communicated, albeit with a lower level of manifestness. Consider (19) and (20):

(19) Unfortunately, because the real world is more complicated than the computer models, the models cannot, as yet, tell us exactly when such a threshold will be reached, or which part of the climate system is likely to flip first. (<ICE-GB:W2B-025 #071:1>)

(20) If needed, decoy puffins may be used to encourage the real birds back. (<ICE-GB:W2C-015 #100:4>)

In (19), “the real world” is contrasted with “the computer models” thanks to a comparative construction (‘is more complicated than’). Real world refers to the physical world as we know it from everyday experience, as opposed to a theoretical, mathematical version of it. In (20), the decoy puffins may well be ‘grounded in the extralinguistic world’ (I), they are not alive, but man-made, and are not regarded as counting as real birds. The question that arises then and is addressed by the speaker, is whether the words world and birds can be adequately applied to computer-modelled worlds or decoy penguins. The answer is, obviously, no. Sense II.a is also conveyed here, but not as manifestly (it is not as salient) as the subsense considered, which can either be paraphrased with the words actual (19) or true (20) that approximate the meaning of real in those two examples.

3. Corpus study: syntactic positions and meanings of real

43So far we have drawn on our corpus data to obtain a fairly representative picture of the meaning of real as it is used in a wide range of natural utterance contexts, and to show how pragmatic processes come into play to determine its precise meaning in individual utterances. Although several research topics lying beyond the purview of the present study might be worth exploring in the future, with a systematic analysis of head nouns that co-occur with the adjective real, the general syntagmatic patterns in which it is found (e.g. when part of an NP, is the NP definite or indefinite, and if indefinite, does it appear after a copula?), or of how it differs from potential competitors like true, actual, genuine… (Magnusson [2003]), we chose to use ICE-GB to test the hypothesis that there may be a correlation between the semantics of real and its syntactic position. As we noted (2.1.), real is neither a descriptive adjective, nor a typical adjective: it may therefore be less compatible with the predicative function than other adjectives, though it is sometimes used predicatively. After presenting our parameters and classificatory choices (3.1.), we discuss the syntactic distribution of the adjective real, contrasting it with that also found in ICE-GB for the whole class of adjectives (3.2.), to account for it finally (3.3.).

3.1. Parameters, technical meanings, lexicalized phrases and metaphors

  • 31 Only one of them is preposed as it occurs after the adverb too: “Too “real” an exhibition is still (...)

44In addition to providing semantic annotations for each occurrence of the adjective real, we also tagged them individually depending on their syntactic position or function: adjectives can be predicative, attributive, postpositive, or prepositive31 (see, e.g., Huddleston & Pullum [2002: 528, 550-551]), to check whether there was a correlation between the sense of occurrences of real and their syntactic position. We semantically classified occurrences according to the three main senses illustrated in Figure 1 (in 2.1.), adding the categories ‘ambiguous’ for occurrences to which a single sense could not be clearly attributed (see 2.3.), and ‘technical meaning’ when real was used in a technical sense: though presumably related to the main senses of real that we highlight here, they are restricted semantically by definitional fiat, by tacit agreement within a field (see 1.2.). We also chose to propose two tables for our results, with one that does not include lexicalized phrases built with the adjective real, and the other that does. Lexicalized phrases can indeed be counted as independent lexical units. Table 1 presents our classification of occurrences of the adjective real in ICE-GB, discounting occurrences within lexical units, based on the semantic and syntactic criteria selected:

Table 1. Classification of the uses of real in the ICE-GB corpus (discounting lexical units)






I. ‘grounded in the extralinguistic world’ and subsenses






(II. scalar interpretation applied to modified term)






II.a. appropriate use






II.b. intensificational use












Technical meaning












  • 32 We did not include them because the technical meaning arises from the combination of real with the (...)

45Lexicalized phrases like (the) real thing make up whole lexical units that include the adjective real as part of them. All of them (21) are attributive within the unit. Among them, 10 types are to be identified. 7 hark back to the early economic meaning, with real included within the economic complex terms “(in) real terms” (5), “real interest rates” (4), “real wages” (1), “real funding” (1), “real economy” (1); “real estate” (2); “real politics” (1). The other 3 types are represented by phrases used in ordinary language like “the real thing” (4), “in real time” (1), or “in real life” (1). Real is always attributive in these phrases. 15 occurrences could also have been grouped within the category we call “technical meaning”32, like “(in) real time”, which technically denotes “the actual time during which a process or event occurs, esp. one analyzed by a computer, in contrast to time subsequent to it when computer processing may be done, a recording replayed, or the like” (OED). In this lexical unit as in most others, the meaning of real pertains to the general sense ‘grounded in the extralinguistic world’ (I) or one of its specific subsenses. The significant exception is the real thing, which can evoke a typical instance of a given notion, and whose use is then clearly intensificational (II.b). The one example in ICE-GB is reproduced as (22), and is contrasted with (21), in which the real thing is simply used (as in the other two examples found in ICE-GB) to refer to something that is actual (i.e. an actual bonfire):

(21) Nobody ever had a bonfire in the sitting room – not intentionally, at any rate – while televised fireworks are a pallid substitute for the real thing exploding across the sky. (<ICE-GB:W2E-003 #069:2>)

(22) It may be a minor event compared with the real thing in the Middle East, but for the City the near collapse of business represents an impending crisis that brokers will ignore at their peril. (<ICE-GB:W2C-013 #024:2>)

“The real thing in the Middle East” is equivalent to “a full-blown collapse” here.

46The few (4) occurrences of technical terms that are not parts of lexical units come from the fields of physics (3) and philosophy (1), and all are elaborations (i.e. modulations) on the sense ‘grounded in the extralinguistic world’ (I) or one of its subsenses, which have become partly routinized (at least) within a field. The use of real in (23) is entrenched within the field of physics:

(23) The ultimate test of such modelling is comparison with measurements of the real radiation field &semi; (<ICE-GB:W2A-029 #018:1>)

Real here means “that is actually present”, as opposed to something that is postulated or experimentally simulated.

47Table 2 presents our classification of occurrences of the adjective real in ICE-GB, when including occurrences within lexical units:

Table 2. Classification of the uses of real in the ICE-GB corpus (including lexical units)






I. ‘grounded in the extralinguistic world’ and subsenses






(II. scalar interpretation applied to modified term)






II.a. appropriate use






II.b. intensificational use












Technical meaning












48In our corpus, there are only two predicative occurrences of the adjective real that conveys a meaning different from sense I (see 3.2), whether one counts or discounts lexical units that include the adjective real. We will try and explain why in 3.2. and 3.3.

49We also observed that all occurrences of real with metaphoric terms (e.g. “a real lawyer/bastard”) are attributive. What surprised us is that while Bolinger makes much of the use of real with metaphoric terms, deeming it characteristic [1972: 134-135], there were only 5 occurrences in the whole of ICE-GB (less than 2,5% of all occurrences), even when including what could perhaps be more readily conceived of as a hyperbole (“a real deluge”: <ICE-GB:W2F-019 #088:1>). These always involve scalarity and they usually point towards a very high degree (II.b), as in (8), quoted at the end of subsection 2.1. (He’s got a real inner steel and toughness), and (24) and (25), with fully lexicalized metaphoric terms:

(24) He’s a real brick (<ICE-GB:S1A-094 #141:1:B>)

(25) and uh but I don’t say anything because I’m a real worm and I’m sort of saying (<ICE-GB:S1A-082 #067:1:A>)

50However, real can also mean that the referent qualifies (or fails to qualify) as an acceptable token for the category designated by the metaphoric term (also lexicalized here):

(26) Not Not any of the real meat I don’t think <,,> (<ICE-GB:S1A-090 #046:1:D>)

The meat in question is the substance or the plot of a story, what is regarded as interesting in it. No high degree of belonging to that category is involved.

3.2. Predicative and other uses

51To see whether the data bears out our hypothesis that predicative occurrences of real may be even more uncommon than for other adjectives, it is first necessary, for the sake of comparison, to obtain the total number – within ICE-GB – of the occurrences of adjectives in each of the three main functions an adjective can have (attributive, predicative, postpositive) – excluding the prepositive position which is highly marginal in terms of frequency and is not tagged within the corpus. A query relying on its annotated structure, using the exploration tools provided by the software ICECUP, yields the following hits:

(i) attributive


(ii) predicative complements


(iii) postpositive


This means that out of 59,073 tokens, around 72% of the adjectives are attributive, 26% are predicative and 2% are postpositive. There seems to be a general bias towards the attributive position over all the others, and another against the postpositive position, both of which do not hold only for the adjective real within ICE-GB, but is confirmed for English in general.

Table 3. A comparison of the syntactic position of adjectives in ICE-GB


predicative complements



all adjectives


















52The distribution of real in ICE-GB (excluding the prepositive occurrence) confirms the general bias towards the occurrence of adjectives that are attributive over the other two kinds considered, and is even significantly more marked (183 occurrences, amounting to 90%, as opposed to 72% in the whole of ICE-GB). Predicative and postpositive occurrences of real are not excluded, but they are relatively rare: respectively 8.5% and 1.5% of the occurrences, as opposed to 26% and 2% for all adjectives in ICE-GB.

53What is striking is that an overwhelming majority of the predicative uses of real (15 out of 17) pertains to sense I, as in (27), which is typical:

(27) An Asian orator said in the second century A.D. that men had long thought that the peace was a dream and had woken up to find that it was real. (<ICE-GB:W2A-001 #053:1>)

Instead of “was real”, “had come true” is possible, but there would have been less emphasis on the material conditions for (the dream of) peace to come into being, which is precisely the point of the utterance. In the two predicative occurrences with real that means more than ‘grounded in the extralinguistic world’ (I), that meaning is less relevant (or perhaps simply irrelevant):

(28) But with a coordinating committee of pro-Moscow organisations <,> now calling for the dissolution of parliament <,> the authorities believe the threat of a military intervention similar to that in Lithuania has become too real to ignore <,> (<ICE-GB:S2B-015 #064:1:F>)

  • 33 (29) is taken from a letter written to someone who has lost a dear one. Having just evoked a devast (...)

(29) As well as this, and I’m sure you’ve felt this, there is the feeling of the love and feelings that a person has for you being more real after they’ve died. (<ICE-GB:W1B-010 #012:1>)33

Both utterances involve a degree adverb, which triggers a scalar interpretation for real. In (28), which reproduces (17), signs have become numerous enough to point towards the appositeness of using the word threat (II.a), and the use of the change-of-state verb become contributed to giving rise to a gradable interpretation. In (29), feelings about the dead are said to be more powerful, so that they are either more rightly called “feelings” (II.a) or more typical feelings, better instances of the category (II.b). (29) incidentally provides a good illustration of the ambiguous examples that we classified separately. When no degree adverb is present before predicative real, we have not found any occurrences that involve a scalar interpretation applied to the modified term (II). The same general bias may also hold for postpositive and prepositive uses of the adjective real, which also seem to resist scalar interpretations.

54Even though postpositive and prepositive occurrences of real in ICE-GB are not numerous enough to draw strong conclusions, all of them can be likened to predicative uses with an underlying relative clause (in which the adjective real is then predicative). Consider again (11) and (12), reproduced here as (30) and (31):

(30) I I think unless you with uhm with something real <,> it’s uhm you you you end up having too many possibilities open to you and it just uhm it becomes a bit of a kind of a <,,> uhm a fantastically complicated mind maze really <,> (<ICE-GB:S1A-096 #030:1:B>)

(31) “Too “real” an exhibition is still an artificial representation and is a coercive form of information.” (<ICE-GB:W1A-012 #044:1>)

In (30), “with something real” could be paraphrased with “something that is real” (in <ICE-GB:S1A-096 #030:1:B>), and in (31), “an exhibition that is too “real”” could be substituted to “too “real” an exhibition”. 3 out of the 4 postpositive and prepositive occurrences of real have a non-scalar interpretation, derived from the general sense ‘grounded in the extralinguistic world’ (I), being loosely used in (30) and (31), or having a technical meaning (‘regarded as having existence in fact, and not merely in appearance, thought, or language’, to quote partly from the OED entry, a philosophical sense that presumably also relates to the same general sense), in (32):

(32) Writing redistributes the verbal surface of the world and illuminates its underlying dynamic by dramatising, writes Beckett, in a close paraphrase of Proust, an “experience [...] at once imaginative and empirical, at once an evocation and a direct perception, real without being merely actual, ideal without being merely abstract, the ideal real, the essential, the extratemporal” (<ICE-GB:W2A-004 #091:1>)

55Only one (postpositive) occurrence yields a scalar interpretation applied to the modified term (II), as it underlines the fitness of the descriptive label used for the head noun (II.a) and co-occurs with the adverb enough, which forces or at least facilitates a scalar reading:

(33) But terminology aside it’s a problem real enough to have prompted doctors at Leeds General Infirmary <,> to organise a conference on it <,,> (<ICE-GB:S2B-038 #071:2:A>)

The postpositive adjective real in (33) is pre-modified by a degree adverb and functions just as when used predicatively (The problem was real enough to have prompted…). In short, postpositive and prepositive uses of real pattern with predicative uses and not attributive uses. But why is the attributive function the only one to license scalar interpretation (except when real appears in other functions with a degree adverb)?

3.3. Syntactic position and scalar interpretation of real

56In contrast to predicative uses of real (and presumably postpositive and prepositive ones), which strongly favour sense I, though they remain compatible with a scalar interpretation (that bears on the term which real modifies: sense II) if coerced by a degree adverb like too, enough, more (in attested ICE-GB occurrences), attributive uses of real favour gradable, scalar interpretations: a majority of attributive occurrences within ICE-GB pertain to sense II, as opposed to sense I (see Tables 1 and 2). Additionally, when such attributive occurrences yield a scalar interpretation (II), rewriting the NP with real as a sentence with be and predicative real (subject definite NP without real, be copula, predicative real), prevents the emergence of sense II through direct (lexical) activation, leading to marginal acceptability, at best:

(34) It it’s a real treat isn’t it (<ICE-GB:S1B-029 #061:1:A>)
>> ?! The treat is real.

(35) So <,,> when he restricts sacrifice to a single site he is making real change because until now sacrifices had been conducted everywhere <ICE-GB:S1B-001 #102:1:A>
>>  ?! The change is real.

In both cases, the use of real is intensificational (II.b): the meaning of (34) is very close to It’s such a treat, isn’t it? and real in (35) could be replaced with a lot of or significant: he is making a lot of / significant change. Predicative occurrences resist scalar interpretations, as we have noted. Similar observations can be found in the literature: Magnusson points out that in phrases like real leather, the adjective is a classifier (drawing on the distinction between descriptors and classifiers, as introduced in Biber et al. [1999: 508]), which selects “a category of entities from a number of potential alternatives and […] often operates in the context of a system of classification in which it may acquire highly specialized meanings which deviate from their prototypical ones. Most classifiers are restricted to the attributive function.” (Magnusson [2003: 31]). He goes on to argue that Do you want real leather? is acceptable, while The leather is real needs sufficient context to be acceptable (Magnusson [2003: 31-32]).

  • 34 This example is borrowed from Magnusson [2003: 31-32].

57For Bolinger [1967: 15], predicative uses of adjectives are more constrained than their attributive uses: contrasting *The policeman is rural with Henry is a rural policeman (with the relational adjective rural), he remarks that when the adjective is predicated (with copular be) on the head noun (policeman), “it is restricted by the CATEGORY of the noun”, assuming here that “rural is not used with human subjects” [1967: 15], while – on the contrary – the adjective does not need to be so restricted when the head noun is modified by the same adjective within an NP. In Bolinger’s own terms, predicative adjectives can only effect referent-modification (i.e. giving additional information about the referent), and attributive adjectives can produce both referent-modification and reference-modification (the reference is not denoted by the addition of properties given by the adjective and the head noun). An eager boy is a boy that is eager (referent-modification in both cases), but an eager student is someone who is eager in his studies (reference-modification), and not a student that is eager (referent-modification), since an eager student “suggests someone who is eager qua student” (Bolinger [1967: 15]). Whether reference-modification is possible naturally depends on the syntactic position of the adjective (it needs to be attributive) and the semantics of the head noun and the adjective itself. Let’s return to eager student: the most salient property associated to student is that a student (presumably) studies (‘X studies’), and the adjective eager is then understood to apply to the student engaged in studying, and so to the student’s way of studying (her/his studying is eager). An eager student is not just someone who is a student and happens to be eager. Correlatively, when someone mentions the real owner34 (of a car, for example), they do not mean that the owner is real, but that the relationship of owning is actual, real, for the person in question (who is uniquely picked out from other possible referents with the definite article).

58We postulate that reference-modification indirectly arises from the creation of a complex notion that is not the juxtaposition of the properties conveyed by the noun and the adjective (modulated to adjust to the head noun: see 2.1. and 2.2.): rather, the adjective directly modifies a property or a set of properties associated to the head noun. This assumption makes it possible to explain why scalar meanings applied to head nouns (II) are hardly compatible with predicative uses of real: they imply the creation of a complex notion involving a scalar judgement on a property or a set of properties associated to the head noun. A real treat is a treat that happens to be a very good representative of the category ‘treat’ in the speaker’s opinion. Of course, metaphoric terms highlight properties that are already present to a high degree, and real has an intensificational meaning (II.b) when used with them, which strongly favours attributive uses.

59But why is it still possible to obtain a scalar (in sense II) interpretation of real when the adjective is not attributive? First, there are cases where a subsense of real (I) turns out to be compatible with or is coerced into a gradable interpretation with the use of a degree adverb: real in (31) – “Too “real” an exhibition…” – designates the property of ‘avoiding to appeal to the imagination of viewers’, which has to be conceived of as gradable. Such cases are to be discounted as real does not imply an appropriate use or some intensification (sense II). Then, in (17), (29) and (33), a head noun can refer to an entity that cannot be directly observed in the extralinguistic world: the threat of a military intervention, feelings, and a (health-related) problem. The sense ‘grounded in the extralinguistic world’ (I) is activated and specified just as it is adjusted to the head noun and the use of degree adverb that co-occurs with real (too, more, enough…), yielding a gradable interpretation relating to the degree of conspicuousness at which the head noun referents make their presence manifest (the clues for the military intervention are obvious to a sufficient level to be carefully heeded, the feelings are more powerful, the problem has enough objective correlates to worry doctors…). This strongly attested presence leads to the inference that the head noun referents are good or outstanding exemplars for their class. Sense II.a or II.b is not directly activated for such occurrences, since direct lexical access to either is denied. Instead, that meaning is derived from inferences that hinge on the notion of conspicuousness, and that are facilitated or perhaps even triggered by the explicit marking of scalarity with a degree adverb: the interpretations that correspond to II.a or II.b indirectly arise from the conspicuousness in the extralinguistic world of the head noun referents, which makes them genuine instances of the category.

60This inferential connection between the gradable notion of conspicuousness and the two scalar meanings II.a and II.b may account for the better acceptability of a predicative use of real with threat, as opposed to treat (34):

(36) Global warming is a real threat.
>> The threat of global warming is

61Nonetheless, the overall meaning differs if real is used intensificationally in “a real threat”, and perhaps even slightly if is used to indicate appropriateness (II.a): the variant with predicative real would fail to be fully equivalent as the predicative adjective more readily conveys the additional meaning ‘conspicuous’ (and so ‘significant’, ‘genuine’), while being perhaps not as common – our corpus does not include instances with such scalar uses of real, without a degree adverb.

62Scalar interpretations of head nouns modified by real are possible when the adjective is attributive (as in all occurrences with metaphoric terms), or sometimes when it is predicative (and probably prepositive or postpositive): in our corpus, all such occurrences of real are modified by an adverb implying reaching some (usually high) degree. In addition, it is rarely relevant to draw especial attention to the property of being real with a predicative use of the adjective, as opposed to other properties conveyed by descriptive adjectives (which are more typical adjectives): marking explicitly, with a copula and real, the relation they express, foregrounds that property informatively and makes it salient. The combination of these two factors may partly explain why the ratio of predicative occurrences over attributive ones is so low with real, when contrasted with the bulk of other adjectives.


63Any lexicographic account of real is faced with the difficulty of defining precisely its lexical meaning, with a number of postulated senses that is arbitrary to some extent, while the characterizations for its various senses may partly overlap. However, the main objective of dictionaries is to provide definitions for users to understand the meaning of words, and not to provide a theory or account of their use in a real communicative context. This is presumably all the more difficult with the highly unprototypical adjective real, which can take on a variety of different though related senses.

64There are two main ways of treating polysemy in the literature in semantics and pragmatics: some philosophers amalgamate homonymy and polysemy (since the various senses of a polyseme are conventionalized, lexicalized), while most post-Gricean pragmaticists link polysemy to modulation, whereby hearers adjust the precise meaning of a word to contextual evidence. The facts that hearers can learn new senses for a word inferentially, that new ones can be created by speakers, and that psycholinguistic evidence shows that there is a specific treatment for polysemes support this hypothesis.

65We suggested that the network of meanings to which a polyseme corresponds could be restructured in the course of its evolution (with the construction of an underspecified sense), and that the pragmatic inferences that connect some senses to others are increasingly routinized through repeated use, until the new senses become fully lexical.

66Our own account of real postulates that there are two main families of senses for the adjective real. First, its primary sense is ‘being grounded in the extralinguistic world’, which subsumes many subsenses with a more restricted extension. Second, the sense of real may also imply a scalar evaluation bearing on the term it modifies: the referent can either fit simply the category that the term designates (‘appropriate use’) or be an outstanding exemplar for that category, with the idea of a high degree being expressed (‘intensificational use’).

67This general lexical structure interacts with pragmatic processes to determine “what is said” (Recanati), explicatures (within relevance theory) or applied timeless meaning (Grice). We highlighted the crucial role played by modulation in discourse production, with the creation of occasion-specific meanings that evoke new, ad hoc concepts (about 2% of all uses of the adjective real in our corpus). Modulation is also presumably called on when access to a new (sub)sense is routinized but not fully lexicalized.

68Another important pragmatic process for analyzing polysemy and real, disambiguation, eliminates senses that are not relevant enough. Yet when several senses are equally relevant, ambiguity comes up as those senses are communicated at the same level of manifestness. This is not uncommon and there may be an ambiguity between two of the three main senses of real (about 7.35% of all occurrences). There are also occurrences for which one sense predominates, though another one (at least) is also present despite being less relevant.

69Classifying all occurrences of real within ICE-GB according to their function and their meaning, we had to recognize the special cases of technical uses and complex lexical units (which include real). We observed that metaphoric terms were used with real sporadically only (less than 2.5% of all occurrences).

70The attributive function is very much favoured for occurrences of real, as opposed to the predicative function (the prepositive and postpositive function are very poorly represented but appear to pattern with the predicative function). A comparison with the number of occurrences for those adjectival functions within our corpus showed that the bias, shared by all adjectives, was even heavier and much more pronounced for real. The ‘appropriate’ and ‘intensificational’ uses of real do not normally occur when the adjective is predicative. The only exceptions found in our corpus are when real is modified by a degree adverb.

71We argued that this strong bias towards the attributive function can be explained by two factors. First, it is rarely relevant to draw attention to the property of ‘being real’ (in contrast to other properties denoted by descriptive adjectives) by devoting a whole predicate, with a copular verb and real, to attribute that property very explicitly to the subject referent. Second, attributive adjectives can directly modify a property or a set of properties associated to the head noun, giving rise to the scalar ‘appropriate’ and ‘intensificational’ interpretations of real. Adjectives in other positions cannot, though the adjective real can if it is coerced into a specific gradable interpretation – with a degree adverb in our corpus. The subsense of real then amounts to ‘conspicuously grounded in the extralinguistic world’, with the result that the ‘appropriate’ or ‘intensificational’ sense of real is not directly, lexically accessed, but derived inferentially.

72More generally, we demonstrated how a lexico-pragmatic account of polysemy brings into focus the actual mechanisms behind ambiguity and the communication of several related meanings at the same time (potentially communicated at different levels of manifestness), the conceptual interrelatedness of senses for a polyseme like real (since new senses arise from fully routinized inferences), and the creative use of words to convey new meanings. We also hope to have filled a gap in the literature, which has so far paid little attention to the adjective real. More research is still clearly needed, and future quantitative approaches with much larger corpora are likely to qualify our findings.

Top of page


Biber Douglas, Johansson Stig, Leech Geoffrey, Conrad Susan & Finegan Edward, 1999, Longman Grammar of Spoken and Written English, Harlow: Longman.

Bybee Joan & Thompson Sandra A., 2022, “Interaction and Grammar: Predicative Adjective Constructions in English Conversation”, Languages 7(1), 2:

Blinkenberg Andreas, 1969, L’ordre des mots en français, Vol. 2, København: Levin & Munksgaard.

Bolinger Dwight, 1967, “Adjectives in English: Attribution and Predication”, Lingua 18, 1-34:

Bolinger Dwight, 1972, Degree Words, The Hague: Mouton.

Brocher Andreas, Foraker Stefani & Koenig Jean-Pierre, 2016, “Processing of irregular polysemes in sentence reading”, Journal of Experimental Psychology: Learning, Memory and Cognition 42(11), 1798–1813.

Brocher Andreas, Koenig Jean-Pierre, Mauner Gail & Foraker Stefani, 2018, “About sharing and commitment: The retrieval of biased and balanced irregular polysemes”, Language, Cognition and Neuroscience 33(4), 443–466.

Camp Elizabeth, 2006, “Contextualism, Metaphor, and What is Said”, Mind and Language 21, 280-309.

Carston Robyn, 2002, Thoughts and Utterances, Oxford: Blackwell.

Carston Robyn, 2019, “Ad Hoc Concepts, Polysemy and the Lexicon”, in Scott Kate, Clark Billy & Carston Robyn (Eds.), Relevance, Pragmatics and Interpretation, Cambridge: Cambridge University Press, 150-162.

Carston Robyn, 2021, “Polysemy: Pragmatics and Sense Conventions”, Mind and Language 36, 108-133:

Charreyre Claude, 1997, “La gradabilité est-elle donnée ou construite ?”, in Rivière Claude & Groussier Marie-Line (Eds.), La notion, Paris: Ophrys, 53-59.

Culioli Antoine, 1990, “Sur le concept de notion”, Pour une linguistique de l’énonciation : Opérations et représentations, Vol. 1, Paris: Ophrys, 47-65.

Depraetere Ilse, 2014, “Modals and Lexically-Regulated Modulation”, Journal of Pragmatics 71, 160-177.

Devitt Michael, 2021, “Polysemy and Pragmatism’s Challenge”, Overlooking Conventions, the Trouble with Linguistic Pragmatism, Berlin: Springer Verlag, 225-280.

Durkin Philip, 2014, Borrowed Words: A History of Loanwords in English, Oxford: Oxford University Press.

Filippi-Deswelle Catherine, 2014, “Lexical and Grammatical Gradability: Surprise and Grading, Sapir and Culioli”, Pre-print (available on 25/03/2023 at:

Frisson Steven, 2009, “Semantic underspecification in language processing”, Language and Linguistics Compass 3, 111–127.

Grice H. P., 1989, Studies in the Way of Words, Cambridge, MA: Harvard University Press.

Huddleston Rodney & Pullum Geoffrey, 2002, The Cambridge Grammar of the English Language, Cambridge: Cambridge University Press.

Ingham Richard, 2012a, The Transmission of Anglo-Norman: Language History and Language Acquisition, Amsterdam and Philadelphia: John Benjamins.

Ingham Richard, 2012b, “Middle English and Anglo-Norman in Contact”, Bulletin des Anglicistes Médiévistes 81, 1-24.

Langacker Ronald, 1987, Foundations of Cognitive Grammar, Vol. 1, Palo Alto: Stanford University Press.

Langacker Ronald, 1991, Concept, Image and Symbol, Berlin: Mouton De Gruyter.

Magnusson Ulf, 2003, “What’s the real thing? Paradoxes and Prototypes of an English Adjective”, ICAME Journal 27, 29-50.

Maniglier Patrice, 2006, La vie énigmatique des signes : Saussure et la naissance du structuralisme, Paris: Éditions Léo Scheer.

Moon Rosamund, 1987, “Monosemous Words and the Dictionary”, in Cowie Anthony P. (Ed.), The Dictionary and the Language Learner, Lexicographica Series Maior, Tübingen: Max Niemeyer Verlag, 173-182.

Moreau Catherine, 2022, “L’expression du degré dans la détermination du nom”, Corela [online journal] HS-37 :

Nelson Gerald, Wallis Sean & Aarts Bas, 2002, Exploring Natural Language: Working with the British Component of the International Corpus of English, Amsterdam: John Benjamins.

Noveck Ira & Sperber Dan, 2007, “The why and how of experimental pragmatics: The case of ‘scalar inferences’”, in Burton-Roberts Noel (Ed.), Advances in Pragmatics, Basingstoke: Palgrave, 181-212.

Recanati François, 2004, Literal Meaning, Cambridge: Cambridge University Press.

Recanati François, 2010, “Adjectives: A Case Study”, Truth-Conditional Pragmatics, Oxford: OUP, 49-76:

Recanati François, 2017, “Contextualism and polysemy”, Dialectica 71(3), 379-397 :

Rey Alain (Ed.), 2019 [1992], Dictionnaire historique de la langue française, new edition, Paris: Le Robert.

Ruhl Charles, 1989, On Monosemy: A Study in Linguistic Semantics, Albany, NY: SUNY Press.

Saussure Ferdinand de, 2002, Écrits de linguistique générale, edited by Bouquet Simon  & Engler Rudolf, with the help of Weil Antoinette, Paris: Gallimard.

Schnedecker Catherine, 2002, “Présentation : les adjectifs “inclassables”, des adjectifs du troisième type ?”, Langue française 136, 3-19 :

Simonin Olivier, 2018, “Sens implicite, implicatures et principes d’inférences”, Corela HS-25 :

Simonin Olivier, 2021, “Ambiguïté linguistique, désambiguïsation et sous-détermination”, in Pennec Blandine & Vincent-Arnaud Nathalie (Eds.), Distorsions cognitives : formes, récits, imaginaire, Toulouse: Presses Universitaires du Midi, 17-31.

Sperber Dan & Wilson Deirdre, 1996, Relevance, 2nd edition, Oxford: Blackwell.

Stammers Jonathan, 2008, “Unbalanced, Idle, Canonical and Particular: Polysemous Adjectives in English Dictionaries”, Lexis 1:

Victorri Bernard & Fuchs Catherine, 1996, La polysémie : Construction dynamique du sens, Paris: Hermès.

“real, adj.”, Collins English Dictionary, Copyright © HarperCollins Publishers (available on 25 May 2023).

real, adj. (2)”, Middle English Dictionary, Robert E. Lewis et al. (Eds.), Ann Arbor: University of Michigan Press, 1952-2001; Online edition in the Middle English Compendium, Frances McSparran et al. (Eds.), Ann Arbor: University of Michigan Library, 2000-2018: <> (available on 24 August 2023).

real, adj.2.”, OED Online, Oxford: Oxford University Press, March 2023. Web (available on 25 May 2023).

Top of page


1 Here is one example in English, illustrating the (originally French) meaning ‘material, objective’ combined with the sense ‘actual’ (and possibly ‘true’), from the first half of the 15th C. : In þis story..I wil procede Of þis same seynt to telle þe lyf real, Both of his diete and eke of his wede (Middle English Dictionary, ‘real’), which we translate as: In this story, I will go on and tell you about the actual life of this saint, both of his diet and also of his clothes – since the word actual conveys both the ideas of truth and materiality.

2 We are making no claim here on how subsequent senses of real precisely arose from legal use. This would require a detailed, separate diachronic study.

3 The quotation and the information given here is taken from (available on 18/12/2022):

4 “9. Law of or relating to movable property, such as money” (‘personal’, CED)

52. an optically formed reproduction of an object, such as one formed by a lens or mirror” (‘image’, CED)

6 “3. a. (of an answer in a fugue) not having the same melodic intervals as the subject, so as to remain in the original key b. denoting a fugue as having such an answer” (‘tonal’, CED)

7PHRASE If you say that a thing or event is the real thing, you mean that it is the thing or event itself, rather than an imitation or copy.” (‘the real thing’, CED)

8 In the CED, some senses are characterized (sometimes loosely) rather than defined.

9 For our treatment of lexicalized phrases, see 3.1.

10 The traditional distinction is fraught with some (not unsurmountable) difficulties. As Victorri & Fuchs point out [1996: 11], it is sometimes difficult to determine whether two senses pertain to polysemy or relate to two distinct, homonymous words – choices being usually made depending on theoretical presuppositions or some arbitrary decision. We would argue that such senses might be represented differently by various speakers: some French speakers might feel that, for the two main senses that the verb form voler can convey (‘fly’ and ‘steal’), there is homonymy and so two distinct words, while others, well-versed in etymology, might argue that they are two distinct senses of just one word, a polyseme, since those two senses are related conceptually: the second sense (‘steal’) is historically derived from the first one (‘fly’) – see Rey [2019: 4155-4156]. The same point holds for the English verb form want, which can mean ‘lack’ or ‘desire’, which some speakers might not regard as related (even though the latter sense is historically derived from the former).

11 We suggest in the next section that some of the senses in a network may not be fully lexicalized, and still need to be backed up by pragmatic inferences.

12 See, especially, Recanati [2002: 5-7], in which he refers (p. 5) the reader to Grice [1989: 25], who distinguishes between what the speaker says and what the speaker implicates.

13 In the examples, we use bold type for the occurrences of real and we underline the elements of the context that are particularly relevant to the analysis.

14 Disambiguation is the process whereby a lexical content is selected for a specific form over its potential competitors with the same form (as would be the case for the form calf, which could denote the young of a cow or a muscle), and reference assignment provides a referent for expressions including noun phrases and pronouns (the door, I…). In addition to those two processes and modulation, saturation should also be mentioned to be fully exhaustive: it gives a value to an inarticulated constituent, as in the sentence You’re not strong enough, for which the hearer understands the task for which she does not have enough strength, though it is not made explicit in the statement (for a summary of those four processes, see Simonin [2018: subsection 1.2.]).

15 Noveck and Sperber go on to write, just after that sentence [2007: 189]: “The speaker’s meaning is inferred from the linguistic meaning of the words and expressions used taken together with the context.”

16 We allow for the possibility that the boundary between the two processes may be fuzzy and that there may be a competition between them or, perhaps, that one might merge into the other. This is a question that remains to be explored and settled experimentally.

17 Or even some underspecified meaning, although this is a much stronger claim, especially if applied to polysemy as a whole (Frisson [2009]).

18 Seeking to interpret Saussure in the light of his unpublished (until recently) writings, Maniglier states that “languages abide not because they have “always already” been established by others, but rather because they always keep on redefining themselves: to be, for languages, means to build themselves anew” (our translation of: “la langue s’impose non parce qu’elle a « toujours déjà » été déterminée par les autres, mais plutôt parce qu’elle est toujours encore en train de se redéfinir : être, pour la langue, c’est se reconstruire.”) [2006: 376; the italics are Maniglier’s].

19 The influence of Anglo-Norman only started to wane long after the transmission of that French dialect was jeopardized some time after the outbreak of the Black Death (1348), remaining strong at least till the end of the 14th C. (see Ingham [2012a; 2012b]).

20 The use of real with narratives in Middle English is a case in point (see the Middle English Dictionary definition given in 1.1.).

21 Broadening naturally leads to the same result. A kleenex can be used to refer to (presumbaly) any kind of paper handkerchief, and when referring to a Kleenex handkerchief, one is (also) referring to a paper handkerchief.

22 Those new senses still require some pragmatic inferencing, before full lexicalization.

23 Among the 7 occurrences of the word real excluded, 3 of them pertain to the category of adverbs, and 4 of them are not given with sufficient context to be interpreted adequately.

24 A presentation of the overall structure of the corpus is available from the Survey of English Usage website: (available on 27 August 2023).

25 Today at the request of Mr Baker <,> they are discussing what Mr Baker has outlined to them when he said a social and political explosion is a very real possibility (<ICE-GB:S2B-047 #029:1:A>).

26 “Adjectifs à « deux ordres-deux valeurs »”, quoting Blinkenberg [1969: 47 et passim]. To qualify Schnecker’s statement when applied to English real, it is important to note that it is used much more extensively than French réel.

27 Our translation of: “ni relationnels, ni qualificatifs” (Schnedecker [2002: 7]).

28 In his account, Bolinger likens the use of adverbial real (for which “really is felt as a pure intensifier, with no remnant of the sentence adverb” [1972: 95]) to its adjectival use in examples like He’s a real dumbhead = He’s real stupid or I had a real fright = I was real scared [1972: 96] in such cases.

29 Our translation of: “ce qui est spécifique au degré dans la détermination nominale, c’est qu’il exprime une quantité de propriété” (Moreau [2022: §159]).

30 A speaker may even successively adopt different points of view concerning what properties are essential for a token to be representative of a category. Lakoff’s illustrative utterance to argue for the validity of cluster models with the word mother is a case in point: “I have four real mothers: the woman who contributed my genes, the woman who gave birth to me, the woman who raised me, and my father’s current wife” [1987: 75].

31 Only one of them is preposed as it occurs after the adverb too: “Too “real” an exhibition is still an artificial representation and is a coercive form of information.” (<ICE-GB :W1A-012#044 :1>).

32 We did not include them because the technical meaning arises from the combination of real with the head noun, and does not just depend on the adjective itself.

33 (29) is taken from a letter written to someone who has lost a dear one. Having just evoked a devastating personal experience of death, the writer goes on to underline that « Their presence is greater and not diminished. » (<ICE-GB:W1B-010 #013:1>)

34 This example is borrowed from Magnusson [2003: 31-32].

Top of page


Electronic reference

Olivier Simonin and Sarah Bourse, The real polysemous meaning of real: a study in lexical pragmaticsLexis [Online], 21 | 2023, Online since 13 November 2023, connection on 02 December 2023. URL:; DOI:

Top of page

About the authors

Olivier Simonin

Université de Perpignan – Via Domitia, CRESEM, France

By this author

Sarah Bourse

Université Toulouse Jean Jaurès, CAS, France

By this author

Top of page



The text only may be used under licence CC BY-SA 4.0. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.

Top of page
Search OpenEdition Search

You will be redirected to OpenEdition Search