1 Introduction
There are a number of contexts in present-day English where never expresses not negation quantified over time, but something closer to straightforward logical negation. This is commonly observed in a range of vernacular British Englishes, where we find examples such as (1)–(4), which are typically in the past, and where never has no quantificatory meaning. In the Survey of British Dialect Grammar, for instance, respondents at 85 per cent of schools surveyed reported use of sentences of this type in their area (Cheshire, Edwards & Whittle Reference Cheshire, Edwards, Whittle and Werlen1995: 77).Footnote 2
(1) You sure you never nicked it? (Cheshire Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998: 47)
(2) The other chaps never got the job. (Tyneside English) (Beal & Corrigan Reference Beal, Corrigan and Iyeiri2005: 145)
(3) I never done it. (Welsh English) (Thomas Reference Thomas and Viereck1985: 218)
(4) There was a car going that way and I thought he would have slowed down but he never. (London Adolescents’ Corpus, 6127int023)
This particular nonquantifier use of never is, however, unavailable in many varieties of English, not least that which is prescribed by prescriptive grammars (cf. Partridge (Reference Partridge1995 [1942]), s.v. never, and Colter (Reference Colter1981), s.v. never did it). As such, we refer to this use in what follows as ‘nonstandard’.Footnote 3
The distinction between uses of never that are nonstandard in this sense, and those which occur across a wider range of registers and varieties, is often subtle. For example, in (1)–(3) never is used in a full sentence with the nonstandard function of ‘denying the taking place of an action on a single occasion in the past, not over a timespan’ (Mazzon Reference Mazzon2004: 121). But unless this punctual meaning is highly contextually salient, speakers who lack the nonstandard use may interpret never in (1)–(3) as the ordinary temporal quantifier and hence judge such sentences as fully acceptable. It might be assumed, for example, that in (2) never quantifies over multiple job application events on the part of ‘the other chaps’. In fact, there was only one such event in this context, and the speaker uses never here simply as an emphatic negator, denying the assumption arising from the preceding discourse that these other people would have been offered the job in question in preference to the speaker. Elliptical structures such as (4), on the other hand, where never appears in clause-final position with ellipsis of the following verb phrase, are instantly recognizable as nonstandard, since speakers who lack the nonstandard use require the insertion of propredicate do in this context.
Crucially, however, there are several contexts in which never routinely fails to express temporal quantification also in varieties of English which do not attract prescriptive criticism. This has already been observed by Cheshire, Edwards & Whittle (Reference Cheshire, Edwards, Whittle and Werlen1995) and Cheshire (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998), but, contrary to what is claimed by these authors, we will see in the following that the relevant contexts are sharply circumscribed in comparison with those which allow the nonstandard use in (1)–(4).
Consider first of all the use exemplified in (5)–(7). Here never apparently functions as a straightforward sentential negator just as in the nonstandard use. An important difference, however, is that never can only be used in this way in the relevant varieties with a restricted class of telic predicates in a context where the ‘window of opportunity’ for the relevant action to be completed has now closed (see section 2.2 for further details).
(5) [He was] supposed to pick us up at lunchtime and never turned up. (British National Corpus, GVP 1260)
(6) You say he never came home that night? (British National Corpus, G3E 1607)
(7) Kids who never made it to the gig are cruising around aggressively in a heady, charged way. (British National Corpus, HH0 428)
When never combines with predicates that do not fall into this class, in a temporal context which precludes a quantificatory interpretation as in (8), the result would normally be considered ill-formed by speakers who lack the nonstandard use in (1)–(4).
(8) #I [set her up with a blind date once]. She never went for the right [guy] though. (London Adolescents’ Corpus, 6127int012)
Example (8) and others like it are commonplace, however, in the (nonstandard) varieties that allow the use in (1)–(4).
Finally, there is another major use of never that is widespread in spoken and written British English that does not involve quantification over time, but rather the categorical rejection by the speaker of the truth of the proposition of which it forms part:
(9) That's never a penalty! (http://news.bbc.co.uk/sport1/hi/football/8608559.stm)
(10) They'll never get here on time.
(11) They never ate all the strawberries?!
These various atypical uses of never have frequently been noted and commented upon, in particular the nonstandard use in (1)–(4), for instance by Beal (Reference Beal, Milroy and Milroy1993: 198), Cheshire, Edwards & Whittle (Reference Cheshire, Edwards, Whittle, Milroy and Milroy1993, Reference Cheshire, Edwards, Whittle and Werlen1995) and Mazzon (Reference Mazzon2004: 120–1), and at greater length by Cheshire (Reference Cheshire1985, Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998), but a detailed examination of their nature and historical background has never been attempted before now.
The purpose of this article is to investigate how these uses of never arose and how we should understand their relationship to one another and to the straightforward quantifier use in synchrony. We will present the following historical account.
In Old or Middle English quantifier never grammaticalizes as a nonquantificatory negation marker, which is, however, restricted primarily to verbs of cognition and to imperatives. This usage is retained in both standard and nonstandard varieties of English, but its distribution has narrowed to include only the verb know and the fossilized imperatives never fear and never mind.
In contrast to this, we argue, contrary to Cheshire (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998), that use of non-temporal never in nonstandard varieties as in (1)–(4) is an independent and fairly recent innovation, representing an extension of the older ‘window-of-opportunity’ use in (5)–(7). The latter, we suggest, arose through reanalysis of utterances containing never together with a predicate that is potentially ambiguous between an atelic reading and a specific kind of telic reading to be discussed below. In contexts where, on the telic reading, the relevant activity is expected only to have taken place once, never necessarily loses its quantificational force, given this reading.
Finally, we argue that the categorical-denial use in (9)–(11) represents a third, separate grammaticalization of the basic quantifier never, arising originally in conditional or future contexts through conventionalization of the inference from ‘at no time in the future’ to ‘under no possible circumstances’.
This article is structured as follows. Section 2 considers in more detail the distribution of never in British English, both generally and in vernacular London English. Section 3 sets out the evidence for the distribution of never in Middle English and Early Modern English, showing that, while nonquantifier negation uses existed, their restrictions suggest historical continuity with prescriptively approved, rather than nonstandard, varieties of contemporary English. Section 4 considers the development of categorical-denial never, investigating its origins in the semanticization of implicatures already regularly present in Early Modern English, but apparently only becoming lexically encoded in the nineteenth century. Section 5 concludes the article.
2 Present-day British English
This section set outs in more detail the syntax and semantics of the major uses of never in both prescriptively approved and nonstandard varieties of British English. We distinguish five main readings: (i) universal quantification over time; (ii) a ‘window-of-opportunity’ reading; (iii) a nonquantifier reading equivalent to ordinary sentential negation; (iv) a categorical-denial reading; and (v) idiomatic uses with know and imperatives. While the first two and the last two are found in all varieties, the third is present only in nonstandard varieties, represented here by London vernacular English.
Standard English data are mainly drawn from the British National Corpus (BNC),Footnote 4 supplemented by the authors’ own native-speaker acceptability judgments, while data for London vernacular English come from the corpus of the project Linguistic innovators: The English of adolescents in London, 2004–2005 (Kerswill, Cheshire, Torgersen & Fox Reference Kerswill, Jenny, Torgersen and Fox2009), which consists of interviews recorded in two boroughs of London. These interviews feature 51 young speakers (aged 16–19) in Hackney and a further 51 in Havering, as well as 9 elderly speakers in Hackney (in their 70s and 80s) and 10 in Havering. To date, the corpus has mainly been used to investigate phonological innovations but there has been some work on morphosyntactic features as well (Cheshire & Fox Reference Cheshire and Fox2009).
2.1 Never as a universal quantifier over time
The basic meaning of never across all varieties of British English, in addition to negation, clearly involves universal quantification over time. This is exemplified in (12), where it is all future time that is quantified over, and (13), where it is all past time.
(12) . . .the universe will never die. (British National Corpus, CET 1575)
(13) Although karate developed in Japan, that country has never won a team world championship. (British National Corpus, GVF 144)
More precisely, the semantics of the ordinary use of never may be expressed as follows. Given a (temporal) context C, a domain D (= the set of all units of time t contained within C) and a proposition p; never(p) is true iff for all units of time t within D, p is false at t. Or, equivalently, never(p) is true iff there is no t within D such that p is true at t. Thus (12) states that there is no instant of time in the future such that the predicate ‘die’ will hold of the universe at that instant, and (13) states that there is no instant of time in the past such that ‘win team world championship’ held of Japan at that instant.
In addition to this unchanging semantic component of its meaning, quantifier never always serves to address one of two contextually determined ‘questions under discussion’ (QUDs) (Roberts Reference Roberts, Yoon and Kathol1996), each involving a presupposition to be denied.Footnote 5 As we will see shortly, this assumption-denying function is important for understanding some of the contexts in which quantifier never cannot be used felicitously. The relevant QUDs may be formulated as the ‘when’ question in (14) and the ‘how often’ question in (15).
(14) When is/was/will be p true?
(15) How often is/was/will be p true?
The presupposition associated with (14) is that there is/was/will be some instant (or longer stretch of time) at which p is true, while the presupposition associated with (15) is that p is/was/will be true on multiple separate occasions. We can see these two QUDs with their associated presuppositions at play in (12) and (13), respectively. In (12) the QUD is ‘When will the universe die?’ As a response to this implicit question, (12) serves to deny the assumption that there is some specific moment in the future such that the universe will end at that moment. In (13), on the other hand, the QUD is ‘How often has Japan won a team world championship?’ As a response to this question, (13) serves to deny the assumption that Japan has won a team world championship multiple times, by asserting that it has in fact won one zero times.
Which of the two QUDs is at play for a given utterance containing quantifier never depends on the iterability of the predicate that never negates. With predicates that are inherently non-iterable, such as ‘die’ in (12), it is necessarily the ‘when’ question that never addresses. Many stative predicates seem to be treated as non-iterable, and as such never always addresses the ‘when’ question with these, as in (16).
(16) I never liked him and I knew he hated me. (British National Corpus, BMN 2591)
If we assume (following Partee Reference Partee1973) that every tensed utterance involves a temporal variable which must be saturated explicitly or implicitly in context, then we may take it that never saturates this variable when it addresses the ‘when’ question. This explains the observation that the combination of quantifier never with non-iterable predicates appears to resist any explicit restriction on the domain D over which never quantifies, as illustrated in (17) and (18).Footnote 6
(17) #The universe will never die this year/this century/this millennium.
(18) #I never liked him yesterday/last week/last year.
With all other predicates (i.e. all those which are iterable), it seems that quantifier never must address the ‘how often’ question, as in (13). It follows that a combination of never with an iterable predicate and an explicit restriction on D is in general perfectly well-formed, as illustrated in (19).
(19) John never won a single game all last night/last week/last year.
Such restrictions on D are in fact the norm whenever never combines with an iterable predicate. In (20), for example, the units of time making up D extend from the moment of utterance back to the time of the event referred to in the since-clause; in (21), they extend over the period (loosely) delimited by lately; and in (22) over a period of just an hour and a half.
(20) Rovers have never been the same since they knocked Newcastle and Wimbledon out of the Centenary Trophy. (British National Corpus, A8N 558)
(21) It was a long while since she had heard Con laugh like that; lately, he never seemed to find much to laugh at. (British National Corpus, FPM 2615)
(22) His delivery was extraordinarily good and, though he spoke for an hour and a half I should think, his voice never failed him and every word was clear –; and bold. (British National Corpus, EW1 430)
It is also possible for the context to restrict D such that its unit-of-time members are not contiguously distributed. In (23), for example, D is composed of all the instants forming the duration of eight separate three-minute rounds of boxing.
(23) Carr, to his credit, never stopped coming forward and he caused Wharton many problems until the eighth round. (British National Corpus, AJJ 638)
Moreover, it is not necessary that the members of D each represent only an instant. This seems to be the usual interpretation, but in an appropriate context they can represent longer intervals. In (24), for example, each unit t is not an instant, but a discrete stay in St Tropez, which necessarily has a duration.
(24) I never wear a swimsuit when I go to St Tropez. (British National Corpus, H8H 2723)
Consider in this connection the example in (25). Here John's response is ill-formed, despite the domain of quantification being a period of two days – much longer than the periods in (22) and (23). An equivalent utterance with not, by contrast, as in (26), is fine.
(25) Mary: God, you look terrible. What's the matter?
John: #I've never slept longer than an hour since the day before yesterday.
(26) I haven't slept longer than an hour since the day before yesterday.
We can understand the infelicity of (25) in terms of the assumption-denying properties of never. As noted above, use of never with iterable predicates addresses the ‘how often’ question, denying the assumption that the relevant event occurs on multiple separate occasions. But this assumption is not available in (25), because a person is only expected to sleep a couple of times in two days, and (25) is ill-formed as a result. Example (26), on the other hand, is fine, because these temporal considerations are irrelevant to the distribution of not, and (27) is also fine because it is expected that a person would sleep multiple times in a six-month period.Footnote 7
(27) I've never slept longer than an hour since my accident six months ago.
Where never quantifies over relatively short periods, this is typically only felicitous when it is the cessation of some activity that is denied for the whole of the relevant period, as in (22) (‘never failed’) and (23) (‘never stopped’). This is because it is usually plausible to assume that what is perceived as a single occurrence of a given activity might briefly cease on multiple separate occasions during the time it is carried out for. On the other hand, we would usually only assume that there should be multiple separate occurrences of an activity with significant duration in the course of a long period as in (27), and not a short period as in (26).
To summarize, temporal never with a non-iterable predicate addresses a ‘when’ question, denying the assumption that there is some specific instant within the contextually determined temporal domain D, such that the proposition in the scope of never is true at that instant. With an iterable predicate, never addresses a ‘how often’ question, denying the assumption that the relevant proposition is true on multiple separate occasions within D. In combination with non-iterable predicates never does not allow any explicit restriction on D, whereas with all other predicates never does tolerate this kind of restriction. However, where D is restricted to the extent that an assumption that the event in question could happen multiple times is no longer available, an utterance with quantifier never is no longer well-formed.
2.2 The window-of-opportunity use of never
Given this background, we can better understand the window-of-opportunity use of never (henceforth never WO) in (5)–(7), repeated here as (28)–(30).
(28) [He was] supposed to pick us up at lunchtime and never turned up. (British National Corpus, GVP 1260)
(29) You say he never came home that night? (British National Corpus, G3E 1607)
(30) Kids who never made it to the gig are cruising aggressively around in a heady, charged way. (British National Corpus, HH0 428)
We will argue that examples such as these involve a nonquantificational interpretation of never equivalent to ordinary sentential negation, a reading available to all speakers, but highly restricted in terms of the aspectual properties of the predicates to which it applies.
Initially, it may seem that never in these examples expresses quantification over time just as it does normally. For example, in (28) we could say that the domain D over which never quantifies is that delimited by lunchtime on the day in question – let us say, for the sake of argument, between twelve and two o'clock on that day. Perhaps (28) says that there is no time within this period such that the man in question arrived at that time. On closer inspection, however, this analysis proves untenable.
If this were a case of ordinary quantifier never, it ought, in combination with the iterable predicates in (28)–(30), to address the ‘how often’ question – denying an assumption that the events occurred multiple times – just as in the other examples with iterable predicates in (19)–(24). This is clearly not the case in (28)–(30), however, because there is no assumption that a person would arrive multiple times for a single appointment in (28), or come home multiple times in a single night in (29), or arrive in time for a concert multiple times in (30). But if never in these examples is the ordinary temporal quantifier, then they should be ill-formed in the same way as (25), given the non-availability of the ‘how often’ question in this context.
This point is reinforced if we alter these sentences slightly, keeping D constant but changing the verb from the preterite to the perfect. The result is that never is no longer felicitous (31a), but non-temporal not is fine (31b):
- (31)
(a) #He was supposed to pick us up at lunchtime. It's three o'clock now and he's
(still) never turned up.
(b) . . .lunchtime. It's three o'clock now and he (still) hasn't turned up.
This effect is clearly linked to the ‘present relevance’ component of the meaning of the perfect, which appears to force never to deny the ‘how often’ assumption. Since this assumption is necessarily absent in the context of (31a), it is ill-formed in the same way as (25). By contrast, where there is a contextually available assumption that the relevant event could occur multiple times even in a short period, as in (22) and (23), replacing the preterite with the perfect in this way preserves felicity:
(32) He's been speaking for an hour and a half now, and his voice has never failed him.
(33) They've been boxing for eight rounds now, and Carr has never stopped coming forward.
A key feature of never WO, observable in (28)–(30), is that it is only possible when (a) there is a temporally restricted ‘window of opportunity’, given or inferable in context, in which the relevant event could theoretically have taken place at any time but didn't, and (b) that window is closed at the time of utterance. The latter requirement is precluded by the present-relevance meaning of the perfect, which is why never WO is restricted to the preterite.
It should not be concluded from this, however, that never WO is simply a by-product of some general property of the preterite, because it turns out that never WO is only possible with the preterite of verbs with a very specific set of aspectual features. At the most general level, the verb must be dynamic, not stative. We have already seen that stative predicates which are treated as non-iterable and therefore address the ‘when’ question do not tolerate any explicit restriction on D, in clear contrast to the never WO-permitting predicates in (28)–(30). This is illustrated in (34) (= (16), (18)).Footnote 8
- (34)
(a) I never liked him and I knew he hated me. (British National Corpus, BMN 2591)
(b) #I never liked him last year and I knew he hated me.
(c) #I never liked him that day and I knew he hated me.
Setting aside stative and other non-iterable predicates, among dynamic predicates we find that those which allow never WO cannot be atelic. Consider the atelic predicate watch telly in (35). With an unrestricted or only slightly restricted value for D – (35a,b) – we clearly have quantification and denial of the ‘how often’ assumption; but with D restricted to a period where this assumption is no longer available, never becomes infelicitous – (35)c.
- (35)
(a) John never watched telly.
(b) During the campaign I never watched telly. (British National Corpus, AJU 639)
(c) #I never watched telly yesterday.
Among telic predicates, those which allow never WO cannot be accomplishments, where reference is to a goal-directed process with a duration, as with eat more than half a meal in (36).
- (36)
(a) He never ate more than half a meal. (British National Corpus, BM1 1467)
(b) He never ate more than half a meal last year.
(c) #He never ate more than half a meal yesterday.
Among achievements (telic predicates without duration), never WO is not felicitous if the predicate refers to some chance event, as with forget to get the hen-food in (37).
- (37)
(a) She never forgot to get the hen-food. (British National Corpus, ABX 2961)
(b) She never forgot to get the hen-food last year.
(c) #She never forgot to get the hen-food yesterday.
Finally, even if never WO is used with non-chance achievements, it is only felicitous if the predicate refers to the completion of a specific task, not merely to some process coming to an end and resulting in one of several possible outcomes. The predicate win as much as half the popular vote in (38) does not meet this ‘completive’ requirement and therefore does not allow never WO.
- (38)
(a) (While they existed,) the party never won as much as half of the popular vote.
(b) . . .over the 1950s the Tories never won as much as half of the popular vote. (British National Corpus, FB5 790)
(c) #In yesterday's election the Tories never won as much as half of the popular vote.
Never WO is thus possible only in the context of non-chance completive achievement predicates in the preterite. Further examples of never WO with predicates that meet this requirement are given in (39)–(41).
(39) John woke up yesterday, but he never got out of bed.
(40) Earlier today, John made his way to the end of the highest diving board, but he never jumped.
(41) John set off up the mountain yesterday morning, but he never reached the top.
The centrality of the notion of completion to the predicates which allow never WO can be seen from the fact that finish (with either an overt or implicit verbal complement) is among these:
(42) ‘We never finished yesterday's kiss,’ she explained. (www.fanfiction.net/s/1908088/3/Loves_Hate)
Similarly, predicates which would otherwise not fall into this category, such as the activity-denoting go, can be made compatible with never WO by the addition of in the end or similar, as in (43).
(43) I never went in the end + I'm so glad cos my friend texted me when she got there + said I wouldn't have got on. (www.cabincrew.com/ccnetwork/printer_friendly_posts.asp?TID=29396)
This example comes from an online conversation by flight attendants about whether the speaker should risk trying to catch a specific scheduled flight which she has only a slim chance of being allowed on. In this context it is especially clear that never is a nonquantifier, because there is no temporal domain for it to quantify over: there was only one possible time when the speaker could have gone – the scheduled time of the flight. Instead, never here (and in all the examples of never WO we have looked at in (28)–(30), (39)–(42)) is a straightforward sentential negator.
To sum up: never may function as a (nonquantificational) sentential negator in all varieties of British English. However, outside nonstandard varieties it can only do so under very limited circumstances, specifically with non-chance completive achievement predicates in the preterite, in a use we term ‘window-of-opportunity’ never (never WO).
2.3 Never as a sentential negator in vernacular London English
We now turn to consider the less restricted, nonstandard use of never. The historical development of this and never WO is addressed in section 3, while the evolution of the categorical-denial use in (9)–(11) is discussed in section 4. Like never WO, the nonstandard use of never illustrated in (1)–(4) expresses sentential negation without quantification over time. What marks out this nonstandard use is its much wider distribution relative to predicate type. It is possible with all of the predicate types in (34)–(38), including in the contexts in (34)c–(38)c, where both a straightforward quantifier interpretation and never WO are unavailable. In the London Adolescents’ Corpus we find examples with statives (44), atelic activity predicates (45), chance achievements (46) and non-completive non-chance achievements (47), as well as examples with predicates where never WO is found across British English varieties (48)–(49).
(44) the police just like they come up . like when I was younger . they just flung me over the car for no reason [. . .] ‘you must have weapons on you . you must have this’ . I never had nothing on me . they'd searched me top to bottom I never had one thing on me neither did any of my friends [. . .] and I made them apologise . I said ‘that's not right . you can't do things like that’ (London Adolescents’ Corpus, 6127int009)
(45) Sue: so did you do your GCSEs at school? what subjects did you take?
Andrew: think it was all of them [Sue: mhm] think it was erm food technology art and design but I never done drama I hated drama well it's because I don't like having my photo taken or anything like that (London Adolescents’ Corpus, 6127int001)
(46) I had to deliver stuff round erm the west. westm. area like westminster and that and I got hit by a taxi so. I never got hurt or nothing just knocked me off my bike (London Adolescents’ Corpus, 6127int007)
(47) well in .. in art yeh I used to get bare As yeh . well the next day they never put me back in art because I was being . cos I was talking too much (London Adolescents’ Corpus, 6127int017)
(48) they caught up quick but they never finished it (London Adolescents’ Corpus, 6127int001)
(49) I remember he invited me out to a Blue concert I never went in the end I was like ‘no’ (London Adolescents’ Corpus, 6127int042)
Although these examples clearly show the wide distribution of this nonstandard use of never, they nevertheless give a strong impression that, like never WO, it is restricted to past-tense environments. This may, however, simply be a reflection of the fact that these are the contexts where it is clearest that never is not functioning as a temporal quantifier. With a present-tense verb, for example, there are very few contexts in which an ordinary quantifier interpretation of never is not a possibility. Even with stative verbs, a habitual interpretation where never quantifies over separate occasions is often available in context, as in (50), in which D comprises all the occasions when the team in question need to drive over the spot where an IED [improvised explosive device] had been placed.
(50) An IED was once placed there before, and the team never like driving over it. (British National Corpus, A77 1374)
Examples such as (51)–(53) from the London Adolescents’ Corpus suggest that this could be obscuring the possibility of nonstandard never also in present-tense contexts.
(51) Ryan is so cool because I can never say that he's trying to be someone else [Lola: yeh] I can never say that he's trying to be black or he's trying to fit in here (London Adolescents’ Corpus, 6127int034)
(52) know what I'm saying you feel like you're the one . that's why I can never say that I'm moroccan. I can never say it (London Adolescents’ Corpus, 6127int036)
(53) Sue: you go to pubs?
Alan: no I've been once and that was like last week I tried it out [. . .] and I liked it but it was just the smoke I never like the smoke in the pub (London Adolescents’ Corpus, 6127int046)
In (51)–(52), which come from two different speakers, never is conceivably the ordinary quantifier, in which case the interpretation is that on each occasion where it might be expected that the speaker would say ‘Ryan is cool’ or that she is Moroccan, this is not in fact possible for her. The context does not favour this interpretation in either case, however, and it seems more likely that never is a simple sentential negator here, and can should have simple present, rather than habitual, interpretation. This is more clearly the case in (53), where the context makes it impossible that never quantifies over multiple visits to the pub. So we have some initial indications that nonquantifier never is not in fact restricted to the past. Whether its co-occurrence with the present tense is a widespread feature of the relevant varieties must remain a question for future research, however.
Finally, let us consider the syntactic differences between this use and the two uses discussed previously. As we saw in (4), nonstandard never may appear in clause-final position with ellipsis of the following verb phrase. Further examples from the London Corpus are given in (54)–(56).
(54) Sue: was it painful when you had it done?
Kelly: no it weren't actually I was expecting it to hurt but it never (London Adolescents’ Corpus, 6127int077)
(55) but that's the reason why she never cos she knew what it was like (London Adolescents’ Corpus, 6127int009)
(56) Ahmed: I never went for that
Mark: you did
Ahmed: no I never (London Adolescents’ Corpus, 6127int043)
Contrast this with the equivalent patterns for varieties which lack nonstandard never, where insertion of propredicate do is required in ellipsis contexts with a full lexical verb:
(57) *Mary always went to France on holiday but John never.
(58) Mary always went to France on holiday but John never did.
In nonstandard varieties like London English, then, never suffices to satisfy the licensing requirements on VP ellipsis and does so without requiring insertion of propredicate do. VP ellipsis is generally thought to require licensing by an overt Aux-element (member of T or equivalent) (Johnson Reference Johnson, Baltin and Collins2001: 440). Propredicate do appears in those contexts where this licensing requirement would not otherwise be fulfilled, as is the case in (57), where never, as a phrasal adverb, is not a potential licenser. However, Potsdam (Reference Potsdam1997) shows that a negative head on its own is also sufficient to license VP ellipsis. This is the case in subjunctive clauses, as in the elided VP of the second clause in example (59), where there is no auxiliary as a possible licenser.
(59) We think that Mary should present her case but we will ask that Bill not. (Potsdam Reference Potsdam1997: 538)
The parallelism between (54)–(56) and (59) suggests that never has been reanalysed as a head in these varieties, whereas it had previously, as in other varieties, been only a phrasal adverb adjoined to TP or VP. This change amounts to an instance of Spec–head reanalysis and phrase-to-head reanalysis (van Gelderen Reference Gelderen2004a, b). Here, generalization of meaning appears to go hand in hand with a shift in syntactic structure from a lexical adverb to a functional head.Footnote 9
2.4 Categorical-denial never
The fourth reading that we distinguish is categorical-denial never, which, like never WO, is not restricted to nonstandard varieties. Examples of this reading were given in (9)–(11), repeated here as (60)–(62), where it can be seen that, unlike never WO and nonstandard never, categorical-denial never has no restriction to particular tenses or to predicates with particular aspectual properties.
(60) That's never a penalty! (www.news.bbc.co.uk/sport1/hi/football/8608559.stm)
(61) They'll never get here on time.
(62) They never ate all the strawberries?!
As noted in section 1, categorical-denial never does not quantify over time. However, its meaning still seems to involve quantification, in this case over possible perspectives on a state of affairs. Example (60), for instance, can be paraphrased as something like ‘given all the facts and assumptions available to me, there is no process of reasoning by which I can reach the conclusion that this particular incident merited a penalty’. Similarly, in (61), the speaker is both predicting that the people in question will not be on time and suggesting that there is in fact no set of possible circumstances such that they could be on time, given the relevant facts and assumptions. Lastly, (62) gives an example of a common non-literal use of this type of never, where the hearer infers not that the speaker rejects the relevant proposition entirely, as suggested by the literal content of her utterance, but merely that she is very surprised to be entertaining this proposition as a possibility. The evolution of this use of never is discussed in section 4.
2.5 Idiomatic uses
Finally, there are two idiomatic uses of never, which, again unlike nonstandard never, are found across a wide range of British English varieties. The first of these is with the preterite of the verb to know, as in (63), where never has no temporal or quantificatory meaning, being interchangeable with not.
(63) I never knew Marie had a husband. (British National Corpus, A74 740)
The second is with the fossilized expressions in (64) and (65), where, just with the verbs to fear and to mind, and only in the imperative, never functions as a straightforward negator.
(64) Never fear.
(65) Never mind.
As we will see in the following section, these restrictions to the context of imperatives and the cognitive predicate to know will prove relevant to a proper understanding of how the various nonquantifier uses of never in contemporary English relate to apparently similar uses in earlier stages of the language.
2.6 Conclusion
In this section we have distinguished four uses of never which diverge from the core meaning of universal quantification over time. The first, never WO, which is found across a range of British English varieties, is a marker of pure negation that is restricted to the preterite of non-chance, completive achievement predicates. The second is a nonstandard use of never, also as a simple negator, but without the aspectual restrictions of never WO. This use is found predominantly with the preterite as well, but may also be possible with other tenses. Its syntax also diverges from that of standard never in that it has the auxiliary-like property of allowing ellipsis of a following VP without insertion of propredicate do. The third use, categorical-denial never, quantifies not over time, but over possible perspectives on a state of affairs, often with the force of surprise or disbelief. Finally, we looked at idiomatic nonquantifier uses of never with the preterite of to know and in the fixed imperative expressions never fear and never mind, which, like the categorical-denial use and never WO, are characteristic of a broad spectrum of British English varieties, not just nonstandard ones.
3 The historical emergence of nonquantifier never
Since never derives etymologically from a negative element plus a temporal element (ever), it seems likely that the temporal quantifier use represents the original meaning, while the other uses represent various grammaticalizations of this negative temporal adverb. Grammaticalization of negative temporal adverbs as sentential negators is reasonably common crosslinguistically, as the following cases exemplify:Footnote 10
(i) Old Norse eigi ‘not’ < (ni) aiw-gi ‘not ever’ (Burridge Reference Burridge1993: 212; Holthausen Reference Holthausen1948; Jóhannesson Reference Jóhannesson1956; de Vries 1957–61)
(ii) Portuguese nunca ‘never’ > Portuguese-based creoles (e.g. Cape Verdean Portuguese Creole) ka ‘not’ (Naro Reference Naro1978: 330–3; Teyssier Reference Teyssier and Bouvier1986)
(iii) Old English itself already had reinforcing nō or nā ‘never, not at all’ < ne + ō, ā ‘always, ever’ (Horn Reference Horn1989; van Kemenade Reference Kemenade, Tieken-Boon van Ostade, Gunnel and van der Wurff1998)
How old the nonquantifier uses described in section 2 are and whether they represent continuity with Old and Middle English is a more difficult question. It should be noted in connection with the nonstandard use that there is a widespread tendency, particularly on the part of non-linguists, to assume that any usages which diverge from the prescribed standard must be relatively recent innovations. In reality, however, such an assumption often turns out to be unfounded once the historical record is checked in detail; see Rickford et al. (Reference Rickford, Wasow, Zwicky and Buchstaller2007) on the ‘recency illusion’. In the present context, Cheshire (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998) suggests that it is indeed false to see the present-day nonstandard use of never as a recent innovation. Instead she suggests that use of never as a pure negator is in fact a direct continuation of the situation in Old and Middle English, pointing to earlier examples similar to (66)–(69), where never clearly does not quantify over time (see also Nevalainen Reference Nevalainen2006: 259).
(66) And his wiff saide she wost neuer.
‘And his wife said that she didn't know.’ (Book of the Knight of La Tour-Landry 79, ?c1450, OED, s.v. never)
(67) Alle his wies were went, ne wist he neuer whider.
‘All his men had gone, he didn't know where.’ (The Romance of William of Palerne 208, c1375,
Middle English Dictionary, s.v. never)
(68) Wiste he nevere what to sein.
‘He didn't know what to say.’ (John Gower, Confessio Amantis 1.2324, a1393, Middle English Dictionary, s.v. never)
(69) Þei wist neuer whiderward to fle.
‘They didn't know where to flee.’ (John Lydgate, Troy Book 4.6434, c1425 (a1420), Middle English Dictionary, s.v. never)
(70) My Lord, I did never know these Men were in the Army when I carried the Message to my Lady. (Trial of Lady Alice Lisle, Helsinki Corpus, cetri3b, 1685)
Cheshire (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998) argues that, already in Middle English, never was on a grammaticalization path to becoming an unemphatic sentential negator. She cites as further evidence the phonological reduction of never to ne'er (cf. also nare a, narey a ‘no X’). On her account, the rise of prescriptive grammar led to the stigmatization of this as an ‘illogical’ use of never. Effectively then, she claims that the contexts in which never can be used in prestige varieties of English have contracted under prescriptive influence over the last three hundred years. The urban vernaculars have been left untouched by prescriptive influence and therefore reflect a situation once found more widely on the sociolinguistic continuum. She suggests that this mirrors the loss of negative concord in English, which, according to one traditional view, has become stigmatized in some varieties of English under the influence of prescriptive grammar, but has remained in urban vernaculars across the English-speaking world.
Recent research suggests, however, that negative concord was in already decline in late Middle English and that the role of prescriptive grammarians was only incidental, speeding up a process that was otherwise underway anyway (Nevalainen Reference Nevalainen2006). Moreover, Cheshire's claims concerning the role of normative statements in the development of never are difficult to demonstrate, since mention of never in early prescriptive grammars is actually rather sparse. Sundby, Bjørge & Haugland in their dictionary of eighteenth-century English normative grammar (1991) give only one instance of a grammar requiring never to be replaced by not, namely the following example from Charles Wiseman's A Complete English Grammar on a New Plan (1764):
(71) What, and did never any one come to relieve you before than? (Wiseman 1764)
Never is clearly being used as a temporal quantifier here, so why Wiseman prefers not in this case is unclear. Whatever the reason, there is little evidence that nonquantifier never declined in certain varieties of English as a result of eighteenth-century prescriptive pressure. In fact, use of never as a simple marker of negation is not objected to by prescriptive grammarians until considerably later. The earliest example we have been able to find comes from a work from 1867, in which the author states ‘Never is often improperly used for not, but it means something more than “not”. . . any one can see that its application to an event that, in the nature of things, could take place but once, is of questionable propriety’ (Gould Reference Gould1867: 98).
More importantly, a closer look at the distribution of nonquantifier never in pre-modern English in fact reveals significant continuity with prescriptively approved, rather than nonstandard varieties of English. The clear majority of examples of nonquantifier never listed in the Middle English Dictionary are found either with verbs of cognition, especially witan ‘know’, as in (66)–(69), or with imperatives, as in (72)–(74).Footnote 11
(72) Misdrede ʒow neuer.
‘Do not be afraid.’ (The Romance of William of Palerne 1567, a1375, Middle English Dictionary, s.v. never)
(73) Lat neuer þis lawles ledis lauʒ at his harmys.
‘Do not let these lawless people laugh at his injuries.’ (The Siege of Jerusalem 495, ?a1450(a1400), Middle English Dictionary, s.v. never)
(74) Neuer shake Thy goary lockes at me. (1623 William Shakespeare, Macbeth III. iv. 49, OED, s.v. never)
As noted in section 2.5, nonquantifier never remains possible to this day in a wide range of English varieties, including those which would generally be considered standard, in combination with the verb know and in the fossilized imperative constructions never mind and never fear. Further examples are given in (75)–(80).
(75) The 56-year-old from New Zealand has spent a month in Northern Ireland with the family he never knew he had. (British National Corpus, K2E 294)
(76) In fact, I never knew he had any – at least during the ‘Carry On’ years. (British National Corpus, J0W 1502)
(77) He never even knew the baby was his. (British National Corpus, HGY 2709)
(78) Never mind, say market optimists: the first quarter of 1991 promises to make up much lost ground. (British National Corpus, ABG 2613)
(79) He would tell a player, ‘Never mind what the other team does – this is what you are going to do.’ (British National Corpus, ABG 2613)
(80) Never fear, you can still buy an attractive neck-tie at King's Cross's Tie Rack emporium. (British National Corpus, AAV 1060)
3.1 Nonquantifier never in historical corpora
In order to shed further light on this issue we examined all the tokens of nev*/neu*/ner in both the Early Modern part of the Helsinki Corpus (550,000 words from a range of text types dating from 1500 to 1710) and the Corpus of Early English Correspondence Sampler (450,000 words covering 23 letter collections dating from 1418 to 1680). Among the total of 513 tokens of never examined in the former and 437 in the latter, there are many examples of the standard quantifier use, as in (81)–(82), as well as examples of the nonquantifier use with know, continuing Middle English usage as in (66)–(69) above. However, cases such as those found in nonstandard varieties today, where never acts as a straightforward negator with an unrestricted range of past-tense predicates, do not occur.
(81) I bring certaine remedy that has beene taught, and proued, and neuer fayl'd. (Thomas Middleton, A Chast Mayd in Cheape-Side, Helsinki Corpus, ceplay2b, 1611–13)
(82) And it is aparent to me, that no ambistious ends moued Luther; for in all the cours of his life he neuer sheawed ambistion. . . (Letters of the Lady Brilliana Harley, CEECS, 1584)
Historical documents thus offer scant evidence that the nonstandard use of never goes back to Middle and Early Modern English. Instead, the continuity appears to be between earlier English and prescriptively approved varieties of present-day English, where the occurrence of nonquantifier never with verbs of cognition and imperatives is retained, albeit having contracted just to the verb know and the fixed expressions never mind and never fear.
Never WO, on the other hand, is attested at least from the early modern period. The Corpus of Early English Correspondence Sampler has the following examples:
(83) I wish you may rit to Dr. Hud about your trunke you left with him, for it never cam to Mester Busbey. (Letters of Isaac Basire, CEECS, 1661)
(84) I never had that letter yr father tells me you ritt; but Franck I had both of yours. (Letters of the Aston Family, CEECS, 1670)
Both of these examples concern items for which the window of opportunity in which they might reasonably be expected to have arrived has closed at the time of utterance. Note that in both cases the events denied are non-chance, completive achievements, which could not have occurred more than once.
Nonstandard nonquantifier uses of never seem to be a much later development. As we have seen, they are noticed and criticized by Gould (Reference Gould1867), but as far as we can tell there are no attestations in the novels of Charles Dickens (died 1870), for example, despite the large amount of vernacular London English (with, for instance, many examples of negative concord) in his writings. This is suggested by a search of the string I never in the University of Stuttgart online corpus of Dickens novels (3.4 million words), which produces 662 tokens, many of them in the context of nonstandard dialogue, but none of which feature nonstandard never (even though this is frequent in the present-day London vernacular; see (44)–(47), (53)–(56)). Clearly, then, nonstandard never had been innovated by the third quarter of the nineteenth century, but it does not appear to have become widespread until slightly later. The earliest examples cited in the OED are from novels of 1896 (85) and 1909 (86); thereafter attestations become frequent.
(85) ‘What was that you found the other day and didn't bring to me?’ ‘Nuffin’.’ Dicky withdrew a step. ‘It's no good you a-tellin’ me that. . . Wot! After me a-takin’ you in when you was ’ungry. . . then you goes an’—an takes your findin's somewhere else!’ ‘I never!’ protested Dicky stoutly. (1896 A. Morrison Child of Jago xii. 120, OED, s.v. never)
(86) ‘Davy,’ said Marilla ominously, ‘did you throw that conch down on purpose?’ ‘No, I never did,’ whimpered Davy. (1909 L.M. Montgomery Anne of Avonlea xvii, OED, s.v. never)
The available evidence therefore points to: (i) an early (Middle or even Old English) innovation of nonquantifier never restricted to verbs of cognition and imperatives, which is retained in prestige varieties of present-day English, albeit just with the verb know and in the fixed imperative idioms never fear and never mind, but which should not be seen as the precursor of nonstandard never; (ii) a later innovation, by the mid seventeenth century at the latest, of never WO, which is retained in all varieties of present-day British English; and (iii) a late innovation (perhaps as late as the mid-nineteenth century), in nonstandard English only, of a punctual use of never that is not restricted by predicate type.
This naturally raises the question of how each of these innovations came about. A definitive answer to this question for each of these innovations is beyond the scope of the present article, but we can suggest the following hypotheses for the two more recent developments.
Concerning never WO, the simplest explanation for its strikingly narrow distribution today is that this is a faithful reflection of the original context in which the semantic–pragmatic reanalysis that brought it about took place; see Detges & Waltereit (Reference Detges and Waltereit2002) for discussion of the notion of semantic reanalysis and its relation to structural reanalysis. That is, we may suppose that the generation of acquirers that innovated never WO did so by reanalysing ordinary quantifier never in the context of predicates that were, at least potentially, non-chance, completive achievements; and, having reanalysed never as a nonquantifier in this context, they continued using it in this restricted context. They did not generalize this use to a wider set of contexts, and neither did subsequent generations of acquirers (except those acquiring nonstandard varieties – see below).
A prerequisite for any reanalysis is the presence in primary linguistic data of a ‘bridging context’: potentially ambiguous utterances which are amenable both to the innovative analysis and to the conservative analysis of the grammars that produced those utterances. In the case of never WO, one possibility is that sentences analogous to (87)–(88) provided the necessary ambiguity.
(87) The mother had the baby on her shoulder for over an hour, but it never burped.
(88) You said your sister would look after me while I was in town, but she never called.
From the perspective of present-day English, both of these sentences can be interpreted as containing atelic predicates and the ordinary quantifier never, which therefore carries with it the assumption that the event denied could plausibly have occurred multiple times within the period specified. This reading can be made more prominent by the addition of once immediately after never in each case. On the other hand, these sentences can be interpreted as containing never WO, in which case the predicates are interpreted as non-chance, completive achievements, meaning that two assumptions are presupposed: (i) a single act of burping in (87) or a single phone call in (88) is envisaged; and (ii) the single occurrence of either of these events represents the completion of some task (that of making the baby more comfortable in (87), or establishing contact between the speaker and the addressee's sister in (88)). Prior to the innovation of never WO, only the first interpretation, with ordinary quantifier never, would have been possible. Importantly, however, even on this conservative interpretation, both the ideas in (i) and (ii) are highly relevant general-knowledge assumptions that are likely to be accessed in the context of these utterances. That is, in addition to the assumption that a baby could plausibly burp multiple times in an hour, or that a friend's sister might call multiple times during a person's stay in a strange town, an assumption will be activated to the effect that only one occurrence of each of these events is envisaged in order to complete the process of either making the baby comfortable in (87), or establishing contact between the sister and the speaker in (88). Bearing this in mind, we can understand the development of never WO as a semantic–pragmatic reanalysis of utterances such as (87) and (88), such that the predicates are no longer interpreted as atelic, but as non-chance, completive achievements, and that never thus explicitly denies the occurrence of a single completive event. As a result of this reanalysis, never ceases to quantify over separate instants of time and loses its associated assumption that the event denied could plausibly have occurred multiple times. It thus becomes a simple negator, but one which is restricted to the context of predicates which are non-chance, completive achievements.
Turning from never WO to nonstandard never, it seems likely that the latter represents a development out of the former (rather than simply from the ordinary quantifier never). Under this scenario, acquirers of the relevant varieties of nonstandard English learn that never can function as a punctual, nonquantificatory negator with the preterite of certain verbs, but they fail to acquire the restriction specifically to non-chance, completive achievement predicates, and so generalize its use to the preterite of any verb. At the same time, acquirers of varieties which approximate standard English maintain this restriction. A point in favour of this hypothesis is that it gives us a natural explanation for why nonstandard never is predominantly associated with the preterite – a generalization which would be unexplained on an account that simply said that it developed out of ordinary quantifier never. At the same time, however, the above scenario is also consistent with the evidence for the incipient spread of nonstandard never to nonpast environments (cf. (51)–(53) in section 2), if we see this in the context of the gradual mapping out of the consequences of the reanalysis of never as a simple negator (see Timberlake Reference Timberlake and Charles1977): first it is restricted to the original context of reanalysis (non-chance, completive achievements predicates in the preterite); then its distribution extends to all preterite verbs; and lastly, in what seems to be an ongoing development, all restrictions on its distribution are lost.
3.2 Conclusion
This section has argued against the nonquantifier use of never in nonstandard varieties of present-day English representing a continuation of the situation in Middle and Early Modern English. Cheshire (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998) suggested that the absence of this use of never in certain varieties of present-day English is the result of prescriptivism. But, as we have seen, there is very little evidence of nonstandard never either in direct attestations or in the prescriptive literature until the second half of the nineteenth century. By contrast, we have seen that the uses of nonquantifier never that we do find in Middle and Early Modern English parallel the situation in a broad range of present-day English varieties, where there is an enduring restriction to the verb know, certain imperatives and the class of non-chance, completive achievement predicates. We argued that this last use of never, which we have called never WO, was innovated through reanalysis of never-containing utterances that were ambiguous between an atelic and a completive achievement reading of the predicate, where, on the latter reading, never loses its quantificational force and becomes a straightforward negator. Lastly, we suggested that the distribution of nonquantifier never in nonstandard English represents a failure to acquire the aspectual restriction operative for never WO in other varieties.
4 Categorical-denial never
In this section, we turn to consider the development of the last remaining use of never distinguished in section 2, namely categorical-denial never. On the question of when this use arose, note that a ‘continuity from Middle English punctual never’ account will be no more tenable for the history of categorical-denial never than it was for nonstandard never. First of all, the fact that categorical-denial never is unrestricted in its distribution marks it out as distinct from both the earlier English punctual use, which was restricted primarily to verbs of cognition and imperatives, and the present-day nonstandard use, which occurs predominantly with the preterite. Secondly, although there is no temporal component to the meaning of categorical-denial never, as we have seen, it does seem to retain an element of quantification – over perspectives on a situation – and it is not clear how this could have arisen out of a use of never as a straightforward negator. Finally, categorical-denial never is scarcely attested in its current unrestricted form until the nineteenth century. The earliest attestation of this use (termed ‘expressing disbelief’) in the OED dates to 1855:
(89) ‘Surely, sir, it's never so!’ said Mrs. Purkis, turning to Mr. Bell for confirmation of the sad suspicion that now entered her mind. (1855 E. C. Gaskell North & South II. xxi. 282, OED, s.v. never)
Among the tokens of never in the Early Modern part of the Helsinki Corpus (1500–1710) and the Corpus of Early English Correspondence Sampler (1418–1680) there is just a single attestation of the categorical-denial use, given in (90), which comes from the mid-sixteenth-century comedy Gammer Gurton's Needle:
(90) Gogs woundes Tyb, my gammer has neuer lost her Neele? (William Stevenson, Gammer Gurton's Needle, Helsinki Corpus, ceplay1b, 1552–63)
If we take this evidence at face value, it would suggest that this use was innovated by the mid sixteenth century in at least some varieties, but did not become widespread until several centuries later.
On the question of how the categorical-denial use of never might have arisen, it is instructive to note that the ability to function as a non-temporal, categorical-denial negator seems to be a special property of never, and does not carry over to equivalent utterances in which never is replaced by the normally synonymous not. . .ever, as in (91) and (92). This suggests that the categorical-denial interpretation of (60)–(62) is not merely an implicature that arises from use of ordinary quantifier never in an appropriate context.
(91) *That isn't ever a penalty.
(92) *They didn't ever eat all the strawberries?!
[Intended interpretation: ‘Surely they didn't eat all the strawberries?!’]
Significantly, however, a reading that is not dissimilar to that of categorical-denial never does seem to be available with not. . .ever in combination with future-oriented constructions:
(93) They won't ever get here on time.
(94) You aren't ever going to believe this.
(95) Ageing beauty Jennifer Aniston says she wouldn't ever have a baby via artificial insemination. (www.buzzvee.com, 19/09/2010)
Judging from (91) and (92), it seems that ever only lexically encodes universal quantification over time, not categoricality in general. If this is correct then it must be that the literal meaning of (93)–(95) is of the form ‘there is no time t in the future such that p is true at t’, and the categorical-denial interpretation of these utterances is derived by an implicature of the type ‘at no time in the future’ +> ‘under no possible circumstances’. This implicature makes sense if we take the standard view of future time reference as universal quantification over possible future worlds (see Copley Reference Copley2002): since the outcome of future events is not predetermined, statements about the future are necessarily statements about all the possible ways that the future could be, where the number of ways the future could be increases exponentially the further from the present a particular future moment is situated. A negative future statement thus asserts that all possible future worlds are such that p is false in them, and the addition of ever makes explicit that this holds true for even for the remotest instants of time in each of these possible future worlds, subject to any explicit or implicit restrictions on the temporal domain over which ever quantifies. Since the number of possible future worlds constantly increases the further one looks into the future, a negative future statement containing ever entails the categorical-denial implicature – if you deny that something is ever going to be the case, this entails the denial that there are any possible circumstances such that it could be the case. The same goes for the (synonymous) standard quantifier use of never with future time reference:
(96) You'll never guess who's here. (British National Corpus, GV8 977)
The literal meaning of (96), assuming never here is the standard quantifier use, is that there is no future moment, such that the addressee will guess at that moment what third party is at the place of utterance; but clearly the primary intention of an utterance of (96) is to communicate the categorical-denial implicature that there is no set of possible circumstances such that the addressee will be able to guess correctly.
If this reasoning is correct, then the categorical-denial implicature should always have been available when the standard quantifier never combined with future time reference, even when categorical-denial never proper, with present or past reference, was not yet possible. This does seem to be the case. In the Corpus of Early English Correspondence Sampler there are many examples in which never combines with future-oriented verb forms, where either the literal temporal meaning or the categorical-denial implicature could be intended as the primary meaning of the utterance:
(97) For my parte, I trust I shall stay noe time here; yf I should, I would never agree to haue this man deall with the money agein. . . (Correspondence of Robert Dudley, Earl of Leycester, CEECS, 1585–6)
(98) . . .for yt was a matter assured them that the queen of England wold never attempt any thing, ether here or elleswhere, but he saw now yt was otherwyse, and that she had sent Drake to the Indyes. . . (Correspondence of Robert Dudley, Earl of Leycester, CEECS, 1585–6)
However, there are also examples where the categorical-denial implicature is clearly intended as the primary communicated content:
(99) My lord, for a particular plesur, I thank you for licensyng [my son] to come home, for suerly otherwise his carcass had never bene brought alyve hyther; he is yet, by the opinion of the physicians, not out of perrill. . . (Correspondence of Robert Dudley, Earl of Leycester, CEECS, 1585–6)
In the specific case of (99), it seems unlikely that extra time would have increased the likelihood of the son being brought home alive; in fact it would probably have decreased it. Hence the speaker must intend to communicate primarily that there are no possible circumstances under which his son could have returned home alive, rather than that his being away would have persisted indefinitely. Later examples in which the categorical-denial implicature is prominent are given in (100)–(102).
(100) for I dare say, if the horses had runn'd over me, as I laid in that nastiness, you'd never have stirred a step to save me. (Frances Burney, Evelina, letter 19, 1778)
(101) ‘Heaven will never let her die so young.’ (Charles Dickens, Oliver Twist, chapter 33, 1837–9)
(102) ‘There's somebody to speak to there, at all event’ he thought. ‘A good hiding-place, too. They'll never expect to nab me there, after this country scent. Why can't I lie by for a week or so, and, forcing blunt from Fagin, get abroad to France? Damme, I'll risk it.’ (Charles Dickens, Oliver Twist, chapter 48, 1837–9)
Our suggestion is that examples such as these, where the naturally arising ‘under no possible circumstances’ implicature is particularly prominent, led to a reanalysis whereby this reading became lexically encoded as part of a separate lexical entry for never; see, for instance, Hansen & Waltereit (Reference Hansen, Waltereit, Bunting, Desai, Peachey, Straughn and Tomková2009) for recent discussion of the diachronic semanticization of implicatures. At what point this newly innovated categorical-denial use of never spread beyond the original context of reanalysis to present and past-tense predications is unclear (a genuine instance of categorical-denial never with future time reference would in any case be impossible to distinguish from a standard quantifier use with a categorical-denial implicature), but as soon as it becomes possible in non-future contexts as in (89)–(90), it is clear that we are no longer dealing with an implicature of categorical denial arising from a literal meaning of temporal quantification, but rather categorical denial as part of the encoded meaning of the lexical item.
5 Conclusion
This article has examined a number of instances in which the function of never diverges from its core meaning of negation plus universal quantification over time. In addition to this core temporal function, we have distinguished four non-temporal uses of never in present-day English: (i) ‘window-of-opportunity’ never WO, in which never functions as a simple sentential negator, albeit one that is restricted to the preterite of non-chance, completive achievement predicates; (ii) nonstandard never, which is also a simple negator, but has no aspectual restrictions, although it too is found predominantly with preterite verbs; (iii) categorical-denial never, which quantifies not over time but over possible perspectives on a state of affairs; and (iv) fossilized or idiomatic uses of punctual never with the verb know and the imperative expressions never mind and never fear.
We have argued that these four non-temporal uses of never are the result of three distinct innovations. First, never grammaticalized in Old or Middle English as a marker of sentential negation restricted to verbs of cognition and imperatives. As we have seen, there is little evidence to support Cheshire's (Reference Cheshire, Tieken-Boon van Ostade, Tottie and Van der Wurff1998) suggestion that present-day nonstandard never is a continuation of this situation. Instead, it is the fossilized uses in (iv) that are the present-day outcome of this first development. Second, never WO (use (i)) arose through reanalysis of the ordinary quantifier never in sentences whose predicates were potentially analysable as non-chance, completive achievements. In prescriptively approved varieties of English this simple negator use of never continues to be restricted to this context, but in nonstandard English it has spread to the past tense of all verbs, and possibly beyond (use (ii)). Finally, categorical-denial never (use (iii)), which has no tense-aspect restrictions, is the result of the semanticization of the ‘under no possible circumstances’ implicature that is regularly associated with use of ordinary quantifier never with future-oriented verb forms.
More generally, the developments explored here serve as a reminder that, although it is often illusory to assume that nonstandard usage is recent and standard usage old, occasionally such intuitions prove well-founded.
