Not one word of it made any sense ’ : Hyperbolic synecdoche in the British National Corpus

A distinct metonymic pattern was discovered in the course of conducting a corpus-based study of figurative uses of WORD. The pattern involved examples such as Not one word of it made any sense and I agree with every word. It was labelled ‘hyperbolic synecdoche’, defined as a case in which a lexeme which typically refers to part of an entity (a) is used to stand for the whole entity and (b) is described with reference to the end point on a scale. Specifically, the speaker/writer selects the perspective of a lower-level unit (such as word for ‘utterance’), which is quantified as NOTHING or ALL, thus forming a subset of ‘extreme case formulations’. Hyperbolic synecdoche was found to exhibit a restricted range of lexicogrammatical patterns involving word, with the negated NOTHING patterns being considerably more common than the ALL patterns. The phenomenon was shown to be common in metonymic uses in general, constituting one-fifth of all cases of metonymy in word. The examples of hyperbolic synecdoche were found not to be covered by the oftquoted ‘abbreviation’ rationale for metonymy; instead, they represent a more roundabout way of expression. It is shown that other cases of hyperbolic synecdoche exist outside of word and the domain of communication (such as ‘time’ and ‘money’).


Introduction
WORD is one of the most frequently occurring noun lemmas in the English language, ranking at number 35, between STATE and FAMILY, in a large representative corpus of present-day English (Leech et al. 2001).Its considerable frequency illustrates the importance of the metalinguistic function in language; since communication itself is such a central human activity, we have a need to refer to it often.The sheer frequency of WORD, however, makes one wonder what the attraction of referring to the word level itself might be-specifically, whether it really reflects a need among speakers of English to talk about individual words and their meaning.An lexemes.Recent work (Ädel 2014) has shown that metonymy is very common also in the abstract noun WORD, to an extent that likely exceeds that of concrete body part nouns.There are also studies of body part nouns used figuratively for linguistic action, such as Goossens's (1995b) work on 'mouth', which draws on corpus-based dictionary data.Furthermore, there is large-scale corpus-based work on WORD specifically, but looking at metaphor; see Martínez Vázquez (2005).There is also relevant work on WORD (and LANGUAGE) from the perspective of lexical polysemy; see Goddard (2011).
The current study was carried out against the backdrop of a larger study applying corpus-based methods in studying the use of WORD involving both metonymy and metaphor.It was in the process of classifying the synecdoches of the larger data set that an intriguing pattern of hyperbolic expressions, such as Not one word of it made any sense and I agree with every word, was discovered.Thus, the study came about as a result of the characteristically serendipitous nature of corpus work (cf.Fillmore 1992: 35;Johns 1988).The expressions in question were labelled 'hyperbolic synecdoche', and the following research questions were posed: 1. How can hyperbolic synecdoche be defined?2. In what lexicogrammatical patterns involving word does hyperbolic synecdoche appear?
3. How frequent is hyperbolic synecdoche involving word in a large, representative corpus of present-day English?
4. To what extent does hyperbolic synecdoche function in accordance with the 'abbreviation' rationale offered for prototypical metonymy?
5. Is hyperbolic synecdoche restricted to word, or is it also found in other nouns?Question 1 will be covered next, before we can turn to questions 2-3 in Section 4, and questions 4-5 in Section 5.

Definition of hyperbolic synecdoche
Metonymy, along with its subcategory synecdoche, and hyperbole are both figures of language.When language is used figuratively, it can be said that "speakers mean something other than what they literally say" (Gibbs & Colston 2012:1).In the case of metonymy, "one entity is used to refer to, or in cognitive terms 'provide access to', another entity to which it is somehow related" (Littlemore 2015: 4).The metonymic relationship is typically such that we "take one well-understood or easily perceived aspect of something to represent or stand for the thing as a whole" (Gibbs 1994: 320).The type of metonymy that predominates in the case of WORD is synecdoche, which occurs when "a term normally referring to part of an entity is used to stand for the whole entity" (Deignan 2005: 56).Metonymic WORD is highly polysemous, and thus differs from prototypical examples of metonymy, such as Wall Street for the financial industry in the US, or tongue for a specific language.It has a range of semantic mappings at different levels within the single domain of 'communication'.While the core meaning of WORD refers to a small unit of language, it is extended-often occurring in multi-word units-through metonymy to refer to larger units of language of various sizes, such as phrases, speech acts, utterances, entire conversations, arguments, quotations and news (see Ädel 2014 for details).In this study, the default, literal referent of WORD was taken to be the grammatical word, as in "The Shorter Oxford Dictionary says that the word 'empiric' means based on observation…" or "Some of the words on the pages were faint, barely legible" (examples from the BNC; emphasis added).Although notoriously difficult to define, word is described in a dictionary definition as "a lexical unit", "an item of vocabulary" which "in most writing systems [is] normally separated by spaces" (Oxford English Dictionary), and by a linguist as "an entity that is located between morphemes and phrases in the structural hierarchy" (Julien 2006: 618).
It has been noted in the literature that metonymy can be used for hyperbolic effect.In a recent book on metonymy, Littlemore (2015:94) observes that many hyperbolic expressions "rely on WHOLE FOR PART metonymies", as in 'The whole town is livid'.The type of word synecdoche that is examined here is characterised by hyperbole, the traditional definition of which is as "a figure of amplification or attenuation by which the speaker signals emotional involvement through an exaggerated formulation" (Norrick 2004(Norrick : 1730)).Hyperbolic expressions have been described as "intensifications for evaluative or affective purposes" (McCarthy & Carter 2004: 150), or as a device that "magnifies and upscales reality" (ibid.: 158).An example is when the listener who hears I almost starved to death when I stayed at my aunt's house!'corrects' itthat is, up-or downscales the assertion to accord with reality-to something like 'My aunt did not feed me nearly enough so I was hungry' (ibid.: 158; the invented downscaling is simplified here compared to the original).This phenomenon has also been characterised as describing the world "in terms of disproportionate dimensions" (Haverkate 1990: 103), which, incidentally, is in contrast to metaphor, where there is instead a discrepancy between the propositional content and the extralinguistic reality-an "empirical falsehood" (ibid.: 102).Rather than represent an upscaling of reality, hyperbolic 'distortions' can also be "pressed to the extreme of counterfactuality or absurd, wild impossibility" (McCarthy & Carter 2004: 161). 2 The disproportionate dimensions in the case of hyperbolic synecdoche specifically include the extreme quantities ALL (100%) and NOTHING (0%).Attention is drawn to the building blocksin this case, words-in order to stress ALL or NOTHING, as in the following examples: 3

NOTHING examples
ALL examples (1) There are no words to describe the profound horror of these beasts, but "aaaaaargh!" will do for now.
(2) When I arrived I couldn't speak a word of Spanish, didn't know anything about the photographic scene in Madrid...
(3) Poor Sybil believed every word of Mildred's story about the frog and she burst into renewed and even noisier sobs… (4) On Monday Busacher made Stefan go over again every word of his conversation with Gesner.
The hyperbolic synecdoche in these examples is used to express affective meanings that are often highly evaluative. 4The examples are associated with a particular lexicogrammar, for instance in being used with quantifiers such as every [single] and no or not a/an (see further Section 3.2).A lowerlevel part for the whole (represented by word here) is selected and used in an expression that represents a less direct way of putting it: In (2), for example, I couldn't speak a word of Spanish is used rather than, say, I couldn't speak Spanish.Attention is drawn to the most salient building block of language-the word-in order to stress the speaker's inability to communicate in Spanish.Similarly, in (3), the subject is said to have believed every word of Mildred's story rather than simply believed Mildred's story, again, to magnify the strength of the belief.Of course, since words do not have truth values (unlike propositions), it is impossible to 'believe' language at this level.The examples above display a 'zooming in' and 'concretizing', whereby a lower-level unit is selected and quantified at either end of the scale: 'none whatsoever' or 'every single one'.It would be possible to replace the hyperbolic synecdoche in (2) with 'at all', and that in (3) with 'fully' or 'completely'.
Some linguists, such as Norrick (2004Norrick ( : 1729)), make a distinction between hyperbole and/or overstatement and 'extreme case formulations' (ECFs), arguing that "ECFs make claims involving the end points on scales, [while] overstatement more generally occurs any time a speaker makes a claim higher (or lower) on some scale than warranted".In the sense that hyperbolic synecdoche involves the end points on scales (as in He didn't say one word), it is a good fit to Pomerantz' (1986) original category of ECFs. 5ne very interesting observation that has been made about about ECFs is that they are "not necessarily heard as absurd or counterfactual and often display a degree of conventionality (e.g.x was absolutely covered in mud)" (McCarthy & Carter 2004: 157).As we will see in the next section, the examples of hyperbolic synecdoche involving word are rather conventionalised.This is also sometimes the case in hyperbolic expressions in general: Like metaphor, hyperbole can become so conventionalised-or 'dead', to use a common metaphor-as to be not perceived as exaggeration, as in the case of "I have tons of work to do" (cf.McCarthy & Carter 2004: 151).In fact, conventionalised patterns are very common in metonymic extensions of WORD in general (cf.Ädel 2014), and are by no means restricted to the hyperbolic synecdoche set.As a point of comparison, consider the following example in which WORD is not placed at either extreme end of the scale: (5) It was agreed that John Major, departing from precedent, should say a few words at nine-thirty, and announce that the national YC executive and officers were to be… Here, say a few words is used to mean 'make a (brief) public announcement' and, even though a few is not necessarily very far from NOTHING on the scale, it still is not placed at the end point of the scale.
The example provides further evidence of highly patterned data in WORD metonymies, where specialised senses tend to be signalled syntagmatically.

Material and method
To examine the role of metonymy in WORD, data were retrieved from the British National Corpus (BNC), which represents a broad sample of different genres and speakers in 100 million words of spoken and written British English (see e.g.Aston & Burnard 1998 for a description of the corpus).The current study is based on a total of 1,874 examples from the BNC, which represent a random sample of 10% of the total instances of word.Note that it was deemed beyond the scope of the current study to also examine the sample of the plural form words, although see Ädel (2014) for a complete coverage of metonymic patterns involved in both the singular and plural forms of the noun.
Once retrieved by means of a concordance program, the word examples were managed through a custom-made coding tool, for which the first step was to mark the literal and the figurative examples (for criteria, see 3.1). 6This occasionally required more co-text than the one sentence retrieved for the sample set, which meant returning to the full corpus data; for this purpose, the online search interface provided at http://corpus.byu.edu/bnc/ was used.Next followed an analysis of the patterns of usage realised in the figurative examples and a time-consuming process of creating a classification for the different semantic mappings (e.g.'WORD for news/rumour', as in word spread quickly).It was in the process of analysing the patterns of usage that the high frequency of negated examples were noted, as in nobody said a word.It was by examining these examples together, as a group, that the category of hyperbolic synecdoche emerged.Each of the nearly 2,000 examples was examined at least twice, and the more than 700 figurative examples at least three times.The identification of patterns in the data was speeded up through the use of a search function for word strings available in the coding tool.Once the examples classified as hyperbolic synecdoche formed a complete set, with no further examples added or discarded, they were exported into a separate database, where their lexicogrammatical forms were identified.

Results
The results will be presented in the same order as the research questions, starting with lexicogrammatical patterns and ending with the frequency of the pattern.

Lexicogrammatical patterns
The term 'lexicogrammatical' is often used to describe patterns at the interface between lexis and grammar.Corpus-based research in particular has played an important role in uncovering "complex interactions between word sets and grammatical variation", pointing to "lexicogrammatical associations [which] usually operate well below the level of conscious awareness, yet [...] are highly systematic and important patterns of use" (Biber 2006: 202).
The present data set was found to represent a restricted range of conventionalised lexicogrammatical patterns.Among the NOTHING types, four general patterns were found.The most common one is a highly variable negated pattern in which word is in direct object position, with an animate agent subject.This pattern encompasses examples such as ( 6), ( 7) and ( 8): (6) My boss never said a word to me.(7) Howard doesn't believe a word of this crap.
(8) The Princess--looking very thin again--merely toyed with her food and hardly uttered a word during the hour-long trip.
Although this pattern is the least fixed in the group, the numerous examples that belong to this category display clearly patterned behaviour.Negation devices include the quantifier no, the negator not, pronouns such as nobody and neither (neither said a word), sentence adverbials such as never and hardly, and even an example of a negative adjectival prefix (unable to speak a word of English).The verbs controlling the direct object fall into two main groups: the larger one including verba dicendi (28 occurrences of SAY; 6 each of SPEAK and UTTER; 4 of LISTEN TO, etc.) and the smaller one including cognitive verbs (9 occurrences of BELIEVE; 5 of UNDERSTAND; 1 of REMEMBER).The word word itself can be modified by an optional of-phrase, such as of this crap in ( 7).Even though it is the case that metonymy is at work considerably more often than metaphor in connection with word, there are two examples in this hyperbolic category involving multi-word expressions which represent 'metonymy within metaphor', that is, cases in which "a metonymically used entity is embedded within a (complex) metaphorical expression" (Goossens 1995a: 172): (9) He carried on for two days but finally, in excruciating pain, had to go to hospital when a splint was put on his damaged limb.He never breathed a word to Laura because he knew her concern would force him to stop work immediately.(10) ...and to be left alone just to sit for perhaps only five minutes in unthinking apathy was her sole remaining desire.To this end she never wasted a word or gesture...
The expression to breathe a word to somebody about something is a conventionalised form of metaphor in metonymy, and is even listed in online metaphor dictionaries, whereas the more general expression to (not) waste a word is less obviously idiomatic in character.
The second most common NOTHING pattern is a relatively fixed manner adverbial, without a/another word, which typically modifies a motion verb (such as LEAVE), occurring most often in a fronted position before the subject, as in the following examples: (11) Without a word, both men left the dock.(12) Then, without another word, he hauled himself up the ladder, leaving her wide-eyed, speechless, and so confused that she wanted to scream.
The two remaining NOTHING patterns are comparatively uncommon, amounting to approximately 5% of the hyperbolic synecdoche set when taken together.The consequential one word [from NP] [that] S and S [=undesirable action] pattern represents an implicit 'if -then' structure, typically used with the pragmatic force of a threat, as in ( 13): (13) One word from Rainald that you are not proving useful to us and young de Tracy dies.
The last NOTHING pattern functions as a time adverbial typically involving a verbum dicendi, which is set in contrast to a sudden or unexpected event, as in ( 14). ( 14) "You will lead the march, I shall be taking the salute.Any questions?" Before anyone could say a word he had turned on his heel and strode off looking to left and right as he went.We stood and stared after him for a few seconds, then...As a whole, the NOTHING group exhibited a greater range of patterns than the ALL group, in which only two different patterns were found, one of which involves every word: (15) In former times he might have been a nobleman's factotum, now I dare say he is a film director's personal assistant, and he leans forward obsequiously to catch his companion's every word.(16) Some of you may think this is a fairy story, something I've made up, but every word of it is true.
The every word subcategory displays greater variability than the most frequent NOTHING pattern, for example in terms of occurring both in subject and object position and in terms of co-occurring with a range of different verbs, even though, as in the case of the most frequent NOTHING pattern, the largest group also consists of verba dicendi (including SAY, SPEAK, TELL and READ), followed by cognitive verbs (UNDERSTAND, BELIEVE and FORGET).Even though there is quite a bit of variability, it is still far from operating according to an open-choice principle (cf.Sinclair 1991: 110ff.).Further semantic subpatterns are also evident, for example in that the truth and falsehood of what someone has said is invoked, as in ( 16), and that the verbs refer to an agent's attention in some way (expressed through CATCH, ABSORB and GO OVER), and that the verbs refer to words that stand for requests that require following (as in OBEY).The hyperbole in these examples is occasionally strengthened through additional modifications, for example in the form of intensifying adverbials (as in "She asked the stewardess for a magazine and religiously read every word until they reached their first destination").
The other ALL pattern involves in every sense of the word or in all senses of the word, as exemplified in ( 17).
(17) I believe that I am a competitor in every sense of the word, always ready to take up a challenge.
The hyperbolic expression in ( 17) functions as a metalinguistic comment on the appropriateness of the term competitor. 7Of course, the term sense here is not applied with linguistic rigour; only one (non-obsolete) sense is listed for competitor in the OED.Instead, it is applied to remove all doubts that the speaker is a true competitor.
As a final comment on the lexicogrammatical patterns found in the hyperbolic data set, it is interesting to note that the negated NOTHING category predominates so clearly.It would seem that the presence of words is the default (a wish to communicate, much in the spirit of the Gricean communicative principle), while the absence of words is notable.However, as pointed out by an anonymous reviewer, it could also be that there are other expressions to describe verbosity-not necessarily involving word.Further analysis is needed to shed light on this, and see for example whether it holds in hyperbolic synecdoche in general, and not only for expressions involving word.

The frequency of hyperbolic 'word'
One of the advantages of using a representative corpus is that it makes it possible to make generalized statements about the frequency of a specific phenomenon.We can pose the question 'How common is hyperbolic synecdoche?' and actually propose a reasonably defensible answer.
7 Metalinguistic comments form an adverbial category described by Quirk et al. (1985: 618-620) as making "overt reference to the linguistic medium, such as in the selection or coining of an appropriate expression".
Hyperbolic synecdoche might be perceived as a rather marginal phenomenon, but the present data show that it actually accounts for as many as one fifth of the metonymic word examples; with 151 examples out of a total of 725, the percentage of hyperbolic synecdoche involved in metonymic word amounts to 20.8%.Thus, the pattern is persistent and represents a considerable portion of the total occurrences.

Discussion
Equipped with basic information about hyperbolic synecdoche involving word, we can now pose more general questions concerning its function and distribution in (British) English.Below I discuss whether the rationale for hyperbolic synecdoche is congruent with the rationale given for metonymy in the literature, and whether hyperbolic synecdoche is a 'natural' pattern also in other nouns in English, and thus not restricted solely to word.

Implications for the 'abbreviation' rationale given for metonymy?
Why do we use metonymy, according to the literature?Several different answers have been given, going back as far as the time of the ancient Greeks, but one common answer involves abbreviation.Consider the following extracts (with added emphasis), taken from several different scholars, the first from an encyclopaedic entry on metonymy, and the last from a textbook, thus strongly indicating that this is the established view: "[M]etonymy is a conceptual and syntactic abbreviation device"; it is "a strategy to extract more information from fewer words" (Nerlich 2006: 111) "Let us [...] consider why anyone would say The kettle is boiling instead of the straightforward The water in the kettle is boiling.It seems that what we do when we coin a metonym is that we mention some salient feature of some entity and, expecting that the decoder can work out what entity we have in mind, we neglect to mention it.This neglect may be due to a wish to economize with words."(Warren 1992: 72) "Metonymy is a convenient way of making verbal passages shorter by avoiding including details" (Alm-Arvius 2003: 159) We can also note that the subtitle of Littlemore's (2015) recent book on metonymy includes the word "shortcuts": Hidden shortcuts in language, thought and communication.This is also one of the justifications for Littlemore & Tagg's (2016:2) study of metonymy in text messaging specifically-where there is "a need for speed and a lack of space".
Judging from these quotes, the rationale given for metonymy is essentially to make language more condensed, or the linguistic message briefer.Or, to put it in more cognitively-oriented terms, it is to give the language user quick access to entities.Deignan (2005: 54) also describes metonyms, such as university for the decision makers within a university (as in ...in a bold move, the University decided that when making a permanent appointment, it would split Mr Spencer's post on the industrial model of a chief executive and a chairman), as serving "to avoid unnecessary wordiness".The abbreviation rationale seems valid for many cases of prototypical metonymy, but does it apply in all cases?While the kettle and university examples clearly represent abbreviations, it is less clear how wheels for 'car'-also an oft-cited example of metonymy-represents an abbreviation.Referring to a car directly would be equally 'brief', and there is no immediately apparent way of specifying the exact referent by expanding the noun phrase8 ('the wheels of the car', for example, would not work). 9here does this leave our hyperbolic synecdoches?Do they fit in with the economizing rationale?It seems that hyperbolic synecdoche typically entails a somewhat roundabout and 'wordy' way of putting things.By explicitly mentioning the part, the expression becomes longer.It is often possible to rephrase the hyperbolic expressions in the data set, such that they are made shorter, as illustrated in the following examples: (18) Robert hardly said a word tonight.
'Hardly SAY a word' can be replaced with 'SPEAK' or 'BE quiet', thus rendering the message as 'Robert hardly spoke tonight' or 'Robert was quiet tonight' (19) Utter so much as a word about last night's work and you will… 'UTTER so much as a word about' can be replaced with 'MENTION', thus rendering the message as 'Mention last night's work and...'.
(20) Me ma went and dropped me there [at the poorhouse] then died without so much as a word.This expression can be replaced with 'without contacting me'.
Unlike The kettle is boiling for The water in the kettle is boiling and university for 'the decision makers within a university', these examples really do not offer a straightforward possibility of extending the word metonymy (the supposed abbreviation), making the exact reference explicit.In fact, the wordiness described here applies not only to hyperbolic examples; there are many other examples in the full word data set that cannot be characterised as abbreviations, especially cases in which word is involved in multi-word units that have become conventionalised.Take, for example, the fixed phrase the word of God for 'the Bible'; referring to 'the Bible' as the word of God requires more words. 10Another case in point is examples which are euphemistic in nature:11 opting for they exchanged words for 'they quarrelled' provides an indirect, and thus more polite, way of describing an event.
How can these discrepancies be explained?Are we simply dealing with different types of metonymy, with different motivations?For example, is it the case that single lexical items that are not highly polysemous are more likely to function as abbreviations than highly polysemous words (such as word or eye) that tend to appear in lexicalised multi-word units?In other words, should we make a distinction between single-lexeme metonymy and multi-word metonymy?If it is the case that hyperbolic synecdoche is typically manifested in multi-word units, one could argue that it therefore represents a marked, or categorially different, case to be treated separately from the prototypical single-lexeme kettle-or university-type examples.At the same time, however, we also need to consider just how 'marked' the multi-word case may be.It turns out to be the case that a considerable part of the metonymic WORD data in generalnot just the hyperbolic examples-occurs in multi-word units (see Ädel 2014), so multi-word metonymy may not be an uncommon phenomenon.In fact, previous corpus-based research on body parts confirms that multi-word expressions are not uncommon (e.g.Hilpert 2007;Deignan 2007;Deignan & Potter 2004); it has even been suggested that multi-word configurations are the norm in figurative expressions: "there is a strong tendency towards the development of multi-word expressions when words are used nonliterally" (Deignan & Potter 2004: 1238).This strongly suggests that theoretical accounts of metonymy need to account for multiword configurations alongside the traditional single-lexeme examples.
In sum, considering the prevalence of hyperbolic synecdoche-it represents one fifth of the metonymy data-it does not seem defensible to view it as representing a non-prototypical category, for which there is no requirement to meet the general abbreviation criterion.Instead, we may need to rethink the rationale in the first place, entertaining the possibility that there are different types of metonymy which behave in somewhat different ways.We can add to that the fact that recent research has shown that hyperbole itself occurs highly frequently even in everyday conversation in English, which means that we cannot dismiss it as a special effect or marginal phenomenon.12

Hyperbolic synecdoche in other domains
Is hyperbolic synecdoche restricted to word and the domain of communication, or is it found in other domains as well?While it is beyond the scope of the present study to carry out a systematic study of hyperbolic synecdoche in other lexical items, I can at least point to some existing patterns, supported by attested examples from the BNC.The fact that it is not difficult to generate examples through introspection and by searching for strings such as every single in a corpus means that this phenomenon is clearly not uncommon, and is likely widespread enough to be of interest to semantic theory and rhetorical studies.
Two semantic domains in which robust patterns of hyperbolic synecdoche are found are 'time ' (examples 21-23) 24) "W-won't you want some of it?""Not a penny!" retorted Aunt Clarabel stoutly."What would I be needing money for... (25) ...they're trying to screw me for every single penny they can get out of me.
The 'money' examples nicely illustrate the two ends of the scale: all and nothing.Penny is the salient unit that stands for money, equivalent to word in the domain of discourse.In the domain of time, however, there seems to be a greater range of options, including units such as seconds, moments and even days.It is reasonable to assume that smaller units, that is, those at the lower end of a rank scale or a meronymy, would be more likely to be chosen for hyperbolic synecdoche, thereby strengthening the hyperbole.Interestingly, there is a subcategory of hyperbolic synecdoche expressions that could be said to function as markers of suspense or detective fiction, as in ( 26) and ( 27); these often have the pragmatic force of a threat.( 26) "This one is valuable," he said firmly."You don't touch a hair of his head."(27) ...take slow, calm, soft breaths and don't move an inch until you are sure that the coast is clear.
Some of the examples are so conventionalised that they are largely restricted to negated NOTHING contexts; we rarely talk about somebody touching every hair on somebody's head, or somebody moving every single inch at the gym.Placing such examples in an ALL context is likely to create a special effect, perhaps rendering the expression idiomatically and pragmatically more effective by contrast to the predominant pattern.An example of an uneven distribution between ALL and NOTHING contexts is found in expressions involving the 'part' noun drop, for which NOTHING contexts occur more than 20 times in the BNC (illustrated in 28), while only one ALL example (29) occurs.
(28) "I don't suppose you have any Scottish blood?" "Not a drop, as far as I'm aware."(29) Every single drop of water going in to your house has been expensively treated to make it drinkable, so do you really want to spray it all over your car?
The 'part' noun drop is sometimes, as in ( 29), complemented by an ofphrase specifying the 'whole', which in this case involves some type of liquid, such as alcohol, blood or rain.Some instances of hyperbolic synecdoche can be described as lexical quantifiers; like hyperbolic metaphorical expressions such as tons of work or mountains of money, they show similarities to partitive constructions, referring as they do to an extreme scale within a restricted set.This can also be compared to lexical quantifiers from selected semantic fields that denote number, which often display hyperbolic uses, and where numerical quantifiers such as dozens of, millions of and hundreds of "seem particularly hyperbole-prone" (McCarthy & Carter 2004: 176).
Let us consider a final example which should be characterised not as hyperbolic synecdoche, but rather as hyperbolic metaphor.Shred is a noun which often occurs in hyperbolic patterns, in particular in the sense listed in the OED as "a fragment, small piece, little bit, scrap (of something immaterial)": (30) You've made a lot of accusations but you haven't come up with a single shred of evidence to back them up.
This type of expression would seem to be especially useful when dealing with non-count nouns, for which the options of cardinal number one or the indefinite article are blocked.Although the expression a single shred of evidence in its syntactic structure very much looks like hyperbolic synecdoche, it is in fact a hyperbolic metaphor.Crucially, there is a crossing of domains with shred (if derived from the 'fragment of textile' meaning) and evidence (essentially an abstract epistemological representation), which is not the case in instances such as (she had not uttered) a word of reproach, where both word and reproach belong to the general domain of communication.

Conclusion
Hyperbolic synecdoche has been defined here as occurring in cases in which a lexeme which typically refers to part of an entity (a) is used to stand for the whole entity and (b) is described with reference to the end point of a scale.Specifically, in using hyperbolic synecdoche, the speaker selects the perspective of a lower-level unit (the part for an entity), which is quantified at either end of the scale of NOTHING (0%)/'none whatsoever' or ALL (100%)/'every single one'.Examples of hyperbolic synecdoche can be characterised as 'extreme case formulations' (ECFs), since they refer to end points on scales.Hyperbolic synecdoche involving word was found to exhibit a restricted range of lexicogrammatical patterns.Among these, the negated NOTHING patterns were considerably more common than the ALL patterns, accounting for some 80% of the cases.Hyperbolic synecdoche was also shown to be a relatively common phenomenon; in a large, representative corpus of present-day British English (the BNC), one-fifth of the metonymic examples of word were found to represent hyperbolic synecdoche.An ongoing study 13 has found that the comparable figure for PENNY is even greater, with as much as 68% of the figurative examples qualifying as hyperbolic synecdoche.
The oft-quoted 'abbreviation' or 'economizing' rationale for metonymy was found not to apply to the examples of hyperbolic synecdoche; by contrast, this set represents a more roundabout and 'wordy' means of expression.It was suggested that the abbreviation rationale covers many prototypical, single-lexeme examples such as university for 'the decision makers within a university', but that words that are highly polysemous (such as word or eye) and tend to occur in multi-word units require alternative explanations.The point was made that multi-word metonymy is not an uncommon phenomenon, to which several recent studies testify, which means that accounts of metonymy need to deal with multi-word configurations alongside the traditional single-lexeme examples.
Although no systematic study of hyperbolic synecdoche outside of word and the domain of communication has been carried out here, it has 13 So far, the study has resulted in a paper presentation ('There's not a penny in your pocket, but we believe every single word you say': The extremes of hyperbolic synecdoche in the domains of money and language) at ICAME 35 at Nottingham University in May 2014.
nevertheless been shown that similar patterns do exist in other domains, for example with reference to time and money.Further research is needed to investigate systematically the forms hyperbolic synecdoche takes and its distribution in different semantic domains.An interesting project for future work would be to identify specific semantic properties shared by those lexemes for which hyperbolic synecdoche is possible.
Once we have a better grasp of how hyperbolic synecdoche operates in English, we should proceed to investigate its distribution in other languages.One possible way of starting such an inventory would be to draw on parallel corpora to examine the ways in which expressions of hyperbolic synecdoche are translated into other languages, following studies such as Mol (2004), for which parallel corpus data were used to examine equivalencies between English and Norwegian in the use of metaphor and metonymy involved in head and heart.
By way of a final word, it seems appropriate to emphasise the crucial role played by naturally-occurring corpus data and corpus-linguistic methods in this study.To the best of my knowledge, the phenomenon of hyperbolic synecdoche has not previously been described in any detail.It was only detected after much examination of examples within a large collection of metonymic extensions of WORD, and not until all of the examples had been classified in different ways did it become clear how frequent the phenomenon actually was.Corpus work is thus extremely valuable in that it has the potential to provide a more authentic, and typically also more complex, picture of the ways in which linguistic phenomena operate.
and 'money' (examples 24-25): (21) ...it was the old Ferrari driver [...] who simply waded into the flames and undid Niki's harness: not a second too soon.(22) She dived for the basin and with only moments to spare was instantly and humiliatingly sick into it.(23) How old was the girl in the doorway?Not a day over twenty-five, Charlotte reckoned.(

Table 1 .
Table 1 lists the patternsdivided into NOTHING and ALL types-and the proportion of examples of hyperbolic synecdoche that they represent.Each pattern is exemplified below, in decreasing order of frequency, following the structure of the table.Distribution of "NOTHING" and "ALL" patterns found (numbers in parenthesis represent raw frequencies)