1 Introduction
Will be -ing and shall be -ing are among the least discussed expressions in English to refer to the future. The few studies that address these constructions tend to focus on synchronic usage, primarily their alleged status as exceptional uses of the progressive (see e.g. Leech Reference Leech2004; Huddleston & Pullum et al. Reference Huddleston and Pullum2002; Williams Reference Williams2002). On their diachronic development, we find a few brief remarks on the timing of the historical emergence of will be -ing and shall be -ing. According to Mustanoja (Reference Mustanoja1960: 591) and Strang (Reference Strang1970: 208), both patterns have been attested since at least Middle English, while Visser (Reference Visser1963–73: 2412) dates them even earlier. The only attempt we are aware of to hypothesise how these constructions emerged is a short section in Samuels (Reference Samuels1972).
We suggest that will be -ing and shall be -ing merit closer scrutiny. Firstly, the variant with will is in fact encountered in a variety of discourse types in Present-Day English. Following are some typical examples found in contemporary usage:
(1) This store will be closing in 5 minutes.
(A standard announcement in UK stores at around 5.25 pm).
(2) This train will be calling at Preston, Chorley, . . .
(automated announcement, Transpennine Express trains, UK, 2005–present).
(3) In a few hours’ time President Bush will be giving his State of the Union Address. (BBC Newsnight, 20/1/2004).
(4) Will you be needing the OHP? (asked of a speaker before giving a conference paper)
(5) Your car will be waiting for you when you arrive.
(6) BCNU (= Mobile phone texting abbreviation for ‘Be seeing you’)
Corpus data, moreover, confirm that in recent times there have been significant frequency developments in standard English. Will be -ing, especially, has raised its profile among future expressions during the last century. Shall be -ing, too, seems to have grown in use at around the same time, though to a lesser extent.
Furthermore, we find that this is one part of the progressive paradigm where there is clear regional divergence between standard BrE and standard AmE. Will be -ing is markedly more popular in BrE. This could be one reason why it has not been commented on much by American linguists.
The main area in which shall be -ing and will be -ing have been previously discussed – and in which we would like to advance the debate – is that of meaning. It has been argued that certain instances of these constructions are characterised by a meaning that is not aspectual, or at least not progressive. Whereas (5) above is clearly progressive, examples such as (1) and (2) would be deemed ‘progressive in form but non-progressive in meaning’ by Wekker (Reference Wekker1976: 116), as not indicating duration by Palmer (Reference Palmer1990: 151), as ‘perfective’ by Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 172) and as applying ‘to a single happening viewed in its entirety’ by Leech (Reference Leech2004: 67). In other words, the prototypical features of progressive aspectual meaning, or ‘progressivity’,Footnote 3 such as ongoingness and duration, are alleged to be absent.Footnote 4 A less popular view is that cases such as (1) and (2) can still be construed as in progress, albeit in an extended sense (see Hirtle Reference Hirtle1967 and Williams Reference Williams2002).
It is sometimes contended that shall be -ing and will be -ing imply further meaning, beyond the notion of aspect. According to Hirtle (Reference Hirtle1967) and Leech (Reference Leech2004), for instance, the realisation of the future situation is construed as ‘a matter of course’, Huddleston & Pullum et al. (Reference Huddleston and Pullum2002) speak in terms of the future as ‘already decided’, whereas Samuels (Reference Samuels1972) and Gachelin (Reference Gachelin and Schneider1997) deem it ‘pure’ or ‘colourless’ future (see also Declerck, this issue). These accounts share a common thread: the idea that the future situation ‘will come to pass without the interference or the volition of anyone concerned’ (Leech Reference Leech2004: 67). However, the characterisations involve subtle differences, which need to be elucidated and assessed against a body of authentic data.
Thus our article is concerned with the following questions:
(a) How frequent are shall be -ing and will be -ing in absolute terms, and in comparison to other constructions referring to the future? How frequent are they across different genres or text types? How do BrE and AmE compare in these respects?
(b) How can we characterise the synchronic meaning of shall be -ing and will be -ing in terms of aspect, and other implied meanings? Can we resolve the apparent anomaly of two aspectual values (namely, progressive and non-progressive) expressed by a single construction?
(c) What selectional preferences and constraints are there on the two constructions, in terms of, for example, subject selection, choice of lexical verb, and combinability with the passive?
(d) What, if anything, is the connection between the meanings, in synchronic and diachronic terms?
(e) What evidence do we have regarding the historical spread in use of will/shall be -ing?
Our study is corpus-based, embracing a range of corpora and privately collected examples from the twentieth century and earlier. Due to a paucity of examples before 1900, our comments on historical developments are necessarily speculative.
Although other modal auxiliaries can combine with be -ing to refer to future situations (e.g. She may be leaving tomorrow), we do not focus on these constructions. Nor, for reasons of space, do we discuss the related be going to + be -ing form. These patterns are, however, implicitly dealt with in some general remarks we make about the effect of combining modals with be -ing.
2 Data and methodology
Our analyses are both qualitative and quantitative. Most of the synchronic discussion is based on qualitative analysis of privately collected examples and corpus data. For the historical part, we focus on two primary sets of corpus data:
(a) The ARCHER corpus, version 3.1, consisting of 1.7 million words of English from 1650 (the end of the early modern period) to 1990, sampled in fifty-year subperiods (cf. Biber et al. Reference Biber, Finegan, Atkinson, Fries, Tottie and Schneider1994). ARCHER contains a variety of written and spoken (or speech-like) genres: journals, letters, fiction, news reportage, medicine, science, drama and sermons. BrE is included in all subperiods, and AmE for 1750–99, 1850–99 and 1950–90 only.
(b) The ‘Brown family’ of corpora: a set of matching one-million-word corpora of written BrE and AmE sampling the years 1961 (the LOB and Brown corpora) and 1991/92 (the F‑LOB and Frown corpora), with a recently added corpus of BrE sampling texts from 1931 (±three years: the B-LOB corpus). Each corpus is modelled on the design of the original Brown corpus (Francis Reference Francis1965), containing four broad genre categories: Press, General prose, Learned/Academic and Fiction.
Evidence of contemporary usage is also drawn from the 100-million-word British National Corpus (BNC), which samples BrE predominantly from the 1980s and early 1990s, and a variety of privately collected examples.
To retrieve occurrences of shall be -ing and will be -ing, we used the CQP software (Christ Reference Christ1994) on grammatically tagged versions of the corpora.Footnote 5 With its advanced query syntax, CQP allows the specification of variable patterns between the modal, be and -ing, such as noun phrases, negatives and adverbials:
(7) Will the master be having his too, m'am? (B-LOB P16)
(8) . . . we regret that she will notbe personallyappearing in Richard II. (ARCHER 1952whit.f8b)
All the examples retrieved were hand-edited to remove false positives.Footnote 6
In quantifying use of shall be -ing and will be -ing, our main method has been to normalise their absolute frequencies to a text span of one million words, to facilitate comparisons across genres and corpora of different sizes. In addition, to confirm the growth of will be -ing in recent BrE, we have assessed its frequency relative to that of other expressions of future time, notably will + bare infinitive, shall + bare infinitive, be going to and the futurate use of the present progressive. All changes have been measured for statistical significance, using the Log Likelihood test.
3 General observations on frequency
3.1 Will be -ing in BrE and AmE
According to ARCHER, despite being available for several centuries, will be -ing did not significantly expand in use in BrE and AmE until the twentieth century (table 1).
Will be -ing in BrE and AmE, 1650–1990: ARCHER data
aARCHER currently has AmE data for the periods 1750–99, 1850–99 and 1950–90 only
b‘pmw’ indicates frequency per million words
By contrast, overall use of the progressive construction in ARCHER accelerated significantly in each century from 1700 onwards, according to figures in Hundt (Reference Hundt2004: 69; cf. table A1). Similarly, in a corpus of nineteenth-century BrE, Smitterberg (Reference Smitterberg2005) finds a substantial increase in frequency of the progressive construction as a whole, but a low and declining frequency of will be -ing (Reference Smitterberg2005: 136).
That the expansion of the construction is a comparatively recent development is further supported by the Brown family corpora of twentieth-century English (Figure 1).
Will be -ing in twentieth-century written BrE and AmE (Brown family corpora): frequencies pmw (raw figures in table A2)Footnote 7
The latter dataset also highlights, more clearly than ARCHER, a significant level of regional divergence between AmE and BrE, which appears to widen as the century progresses. While use of the progressive overall continues to increase in both varieties (cf. Mair & Hundt Reference Mair and Hundt1995; Smith Reference Smith, Peters, Collins and Smith2002), the growth of will be -ing is significant in BrE only.
3.2 Shall be -ing in BrE and AmE
Shall be -ing, in contrast, never seems to have enjoyed much popularity. From its sparse appearances in ARCHER, it is just about possible to discern a slight growth in BrE: one occurrence per century, until the twentieth century, in which there are five occurrences (table 2). This nevertheless represents a small fraction of the frequency of will be -ing. In the AmE portion of ARCHER, a single occurrence of shall be -ing is found, in the second half of the twentieth century.
Shall be -ing in British English, 1650–1990: ARCHER data
A similarly sporadic usage rate is found in the Brown family of one-million-word corpora. At just one, seven and five occurrences respectively in 1931, 1961 and 1991, shall be -ing in BrE seems to have reached its peak. In AmE the frequency again trails some way behind: shall be -ing appears once in 1961, and likewise in 1991.
3.3 Shall/will be -ing and other future time constructions in BrE: a brief quantitative comparison
We now consider the frequency of shall/will be -ing relative to other expressions of future time: will + bare infinitive, shall + bare infinitive, be going to (e.g. I'm going to leave), be to (e.g. I am to leave at 4), the futurate use of the present progressive (e.g. I'm leaving at 4).Footnote 8 The survey here is mainly limited to late twentieth-century BrE (see table 3).
Constructions referring to the future in corpora of recent BrE: raw and proportional frequencies in the LOB and F-LOB corpora
Note: For be going to, be to and the futurate use of be -ing, only present tense cases were counted. Counts for the futurate be -ing involve a higher margin of error than the other constructions, due to indeterminate readings – see below.
* indicates a probability of <.05; ** indicates a probability of <.01.
The only construction significantly increasing in relation to the others is will be -ing. In fiction, for example, will be -ing is buoyant in comparison to a general decline of the combined set of constructions. Meanwhile, figures for the futurate use of the present progressive are somewhat complicated by cases of indeterminate temporal reference (e.g. Timotei . . . is introducing a Facial Scrub to its skin care range, F‑LOB E34); see Mair & Hundt (Reference Mair and Hundt1995: 116) and Leech et al. (Reference Leech, Hundt, Mair and Smith2009: 132–3), for discussion. Such cases are fairly evenly balanced at around 60–90 instances in each corpus. Even so, there is no evidence that the futurate use has increased. This is a notable difference from the clear growth of the futurate progressive in BrE across the last two centuries, as reported by Nesselhauf (Reference Nesselhauf2007 and this issue), again using ARCHER.
Meanwhile two other competitors, shall + bare infinitive and be to, have fared distinctly worse; both have become increasingly confined to formal style (see Leech et al. Reference Leech, Hundt, Mair and Smith2009: 80–1, 108), and shall + bare infinitive is now almost exclusively used with first-person subjects. It is probable that shall be -ing has been prevented from sharing in the success of will be -ing through the demise of shall in general.
Be going to is one construction we might expect to have increased in use (see e.g. Krug Reference Krug2000; Heine et al. Reference Heine, Claudi and Hünnemeyer1991; Hopper & Traugott Reference Hopper and Traugott2003). However, as noted in Mair (Reference Mair, Hickey and Puppel1997) and Leech et al. (Reference Leech, Hundt, Mair and Smith2009: 108), as far as written BrE is concerned, its gains are limited to the Press genres in LOB and F-LOB; in the parallel 1960s–90s corpora of AmE (Brown and Frown), be going to has risen dramatically in frequency, and across a wide range of genres. Similarly in two small corpora of spoken British English, Leech (Reference Leech, Facchinetti, Krug and Palmer2003) finds that the construction has expanded. No instances of be going to in construction with be -ing were found in any of the Brown family corpora.Footnote 9
There are some intriguing tendencies, which can be surveyed only briefly here. For example, the spread of will be -ing does not appear to have had much impact on the frequency of will + bare infinitive. Will + bare infinitive still dwarfs all other future-referring expressions. Its frequency in the genres where will be -ing is strongest – namely the press, especially news reportage and editorials – has even increased by nearly 10 per cent in BrE (see Smith Reference Smith2005: 268). In part this can probably be attributed to a fairly consistent propensity in newspaper writing to refer to forthcoming events, not just situations in the present and the recent past: in both LOB and F-LOB the total number of future-referring expressions is around 4,500 pmw, well above each corpus average.
In summary, the recent (late twentieth-century) growth of will be -ing seems to be confirmed by the fact that most of the future-time constructions it competes with have either declined in use or shown growth in a few registers only.
3.4 Will/shall be -ing across genres in BrE
In ARCHER, will be -ing and shall be -ing both mainly occur in speech-based or speech-like genres. In e.g. Drama, Letters and the dialogue parts of Fiction, will be -ing registers 9, 17 and 3 occurrences respectively (51, 55 and 36 pmw). Among information-oriented genres, there is one attestation in News, and none in either Science or Medicine. ARCHER thus gives the impression that the rather restricted register spread is consistent across the periods sampled, and applies to both constructions.Footnote 10 With the twentieth-century Brown family corpora, however, will be -ing presents a more complex pattern of register variation and change (see Figure 2).Footnote 11 In 1931, Fiction ranks highest in frequency. By the 1960s, Fiction has been overtaken by Press, and by the 1990s it is matched by General Prose.Footnote 12 Not unexpectedly, the construction is consistently infrequent in Learned (i.e. academic) writing: there is rarely call in this text type for predictions and discussion of future situations.
Genre distribution of will be -ing in twentieth-century written BrE: frequencies pmw in B-LOB, LOB and F-LOB
The lack of increase of will be -ing in Fiction may seem at odds with the general expansion of the construction. However, part of the explanation for this is that in fictional dialogue, where nearly all cases are found, the frequency of references to the future reduced across the century. In analysing future expressions for table 3, we found that their total frequency in BrE Fiction declined by 26 per cent.
We conclude our general frequency overview by considering the distribution of will/shall be -ing in the contemporary genres of the BNC (see tables A3 and A4). Three of the main findings are that:
(a) Five of the top ten genres using will be -ing are news-related – including newspapers as well as news broadcast on television and radio.
(b) Neither construction can be considered especially conversational, or typically ‘oral’, in character: conversation is of only middle-ranking frequency among the genres where will be -ing occurs, and near the bottom of those using shall be -ing. This is in stark contrast to conversation's high rank in the overall distribution of the progressive (see e.g. Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 462; Leech et al. Reference Leech, Hundt, Mair and Smith2009: 125).
(c) Shall be -ing is associated most strongly with formal genres, such as parliamentary debates, professional letters and formal meetings.
The preponderance in BNC news media supports the findings from the Brown family corpora above. It seems at least partly explainable by the frequent need in news reporting to refer to forthcoming, pre-planned events, as in (9). In broadcast news this frequently extends to scheduling arrangements about the news programme itself, as in (10).
(9) Vicki will be trying to repeat her double 1991 success in the junior and senior open women's 200 m individual medley. (BNC K4T 1327, Northern Echo, sports news)
(10) In a moment we’ll be speaking to him about his son's plight, but first we go over to Switzerland where our reporter, John Marshall, has been following the story. (BNC KRM 884, Broadcast news: Central TV)
4 Temporal, aspectual and other implied meanings in Present-Day English
We turn now to the meanings of will/shall be -ing in contemporary English. We consider aspectual interpretation, and a closely related set of concepts: volition-neutrality, predetermination and matter-of-courseness. Our discussion questions the discreteness of the different uses of will/shall be -ing, and argues further that non-agentivity is an additional feature. Where appropriate we compare will/shall be -ing to other future-time constructions.
4.1 Aspect: Type 1 vs Type 2
Where will/shall be -ing have been commented on previously, the main distinction made is usually between a ‘normal’ use that indicates progressivity (henceforth Type 1), as in (11), and another use, as in (12), that is considered somewhat eccentric in that it does not imply progressivity (henceforth Type 2):
(11) When we get there, they’ll probably still be having lunch. (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 171)
(12) Will you be going to the shops this afternoon? (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 171)
Either the latter use is regarded as an exception to the principle that be -ing conveys progressivity,Footnote 13 or it is accounted for by loosening the notion of progressivity.Footnote 14 Type 1 or ‘future progressive’ (Palmer Reference Palmer1979: 153) use does not normally attract further comment.
The problem with the first of these approaches is that it makes no attempt to account for the use of the be -ing form in Type 1 and Type 2 in a unified way. It seems incongruous to have two contrasting aspectual values represented in a single construction. The problem with the second approach is that it extends the notion of progressivity to the arrangement or decision that underlies prediction. Although we fully agree that Type 2 is based on some sort of arrangement, we would question whether this arrangement should be considered as being in progress in the future, since it has already been made in the past.
Most previous approaches pose two further problems. Firstly, Type 2 is in fact the more frequent in discourse, and our analysis of twentieth-century corpora suggests that it is becoming more common still: see table 4. Type 1 is compatible with progressivity if a temporal adverbial or the contextual environment provides a frame for the ongoingness interpretation.
Functions of will be -ing: frequencies in twentieth-century BrE (Brown family corpora)
Our analysis of the respective types in twentieth-century corpus data, based on aspectual criteria, shows Type 2 to be the most frequent. In this respect Type 2 is not an eccentric use, which leads us to dispute Samuels’ (Reference Samuels1972: 57) claim that Type 1 is the primary function overall. Moreover, the Brown family corpora show that across the twentieth century this use increased. A second problem is that the level of indeterminacy between the two aspectual types is so high that we must question whether they are in fact discrete categories. Consider example (13), where, without any time specifier (such as during the race), the aspectual value of I'll be competing is difficult to determine:
(13) If Honda do provide me with a factory machine, I’ll be competing on a par with the top riders in the world. (FLOB A41, reportage)
Our corpus analysis indicates that as many as 20–40 per cent of examples cannot be classified discretely according to aspect. We discuss further areas of overlap between the respective types in section 4.3.2.
4.2 The role of be -ing
How can we explain the frequent absence of progressivity in will/shall be -ing? Before moving on to the specific meaning of will/shall be -ing, we need to clarify the contribution of be -ing to the construction.
Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 162–71) provide a generally convincing attempt to unify different uses of be -ing, but treat will be -ing (in its Type 2 function) and the futurate use of the present as ‘non-aspectual uses’ of be -ing (Reference Huddleston and Pullum2002: 171–2). One may wonder why progressive aspectuality should be disallowed by future time reference. Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 171–2) briefly state that these non-aspectual uses both involve future time reference, but do not elaborate. No mention is made either of shall and other modals which similarly may convey non-progressive meaning in combination with be -ing, or of the epistemic use (see section 4.3.3), which does not involve future time reference. On the other hand, in their general discussion of the meaning of be -ing, Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 165) rightly draw attention to an ‘interpretive’ (or ‘interpretative’; see Ljung Reference Ljung1980) use, i.e. where a clause with be -ing refers back to a given situation in order to elucidate its nature. Their example (14) is from a past time context, but it is equally applicable in the future; cf. (15):
(14) When I said the ‘boss’, I was referring to you. (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 165)
(15) If all goes well there may be more ventures of this kind. Tennis teams perhaps. Athletics in the summer . . . The upshot of this is that you all carry a heavy responsibility. I want you to look on yourselves as ambassadors. You will be representing your country – Strathdonald School . . .. (W. Boyd, School Ties, p. 95)
We would question Huddleston & Pullum et al.'s claim that such examples have progressive aspectuality by virtue of the speaker taking an internal, metaphorically extended, view of the situation. The explanatory function of will be representing in (15), for instance, arguably derives from its reidentifying, and implicitly elucidating, the predication <you – carry a heavy responsibility> and the role of ambassador mentioned in the preceding utterance. At the same time, will be -ing identifies the temporal reference point with the future adverbial in the summer.
This is one illustration of the dual function of be -ing, a fact which can be traced back to its component parts. The be -ing form ‘consists of the auxiliary be + the -ing participle of a verb’ (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 151). Be indicates that the situation referred to is identified both temporally and subjectively, because it is always related to some viewpoint (see Larreya Reference Larreya1999). With the -ing participle this situation is presented as already in existence, but with no reference made to the right-hand temporal boundary of the situation (see Bouscaren et al. Reference Bouscaren, Deschamps and Mazodier1993: 28). As such, be -ing signals that the utterance is connected with some specific situation that is viewed from a subjective perspective. As pointed out by Larreya (Reference Larreya1999: 140–3), the concept of identification allows the interpretive use of the be -ing form to be integrated into the realm of aspect, rather than treated as an exceptional case. The following example, quoted by Larreya (Reference Larreya1999: 141), further illustrates the interpretive use:
(16) When thousands of protestors began a march outside East Berlin's state television . . . they had no idea they were starting one of the biggest demonstrations in European history. (Adamczewski & Gabilan Reference Henri and Gabilan1996: 59)
The implication of the situation referred to in the preceding clause is here considered by the speaker. In retrospect, the speaker is able to draw the implication of this event and to interpret its meaning. In this example, as in (15), the event is first referred to and then identified. Referential construction and identification take place in stages. The interpretive meaning arises from the need to re-identify the situation that appears in the preceding clause. However, Larreya (Reference Larreya1999: 141) notes that reference to a situation and its identification may coincide:
(17) It was pouring with rain, so Tim and Elinor stayed on in Hackney after lunch listening to a hair-raising Otello . . . (quoted by Larreya Reference Larreya1999: 141)
The event is here viewed as in progress at some specific time in the past. In addition, this type of standard use of the be -ing form is related to the speaker's point of view. As Mettouchi (Reference Mettouchi, Borillo, Vetters and Vuillaume1997: 199–200) suggests, the state of affairs referred to by the be -ing form may well not be in accordance with the subject's intention. In (17), for instance, Tim and Elinor's original intention was probably to go out, and be -ing indicates that the situation runs counter to the subject's intention.
Drawing upon Larreya (Reference Larreya1999) and Mettouchi (Reference Mettouchi, Borillo, Vetters and Vuillaume1997), we propose to consider this subjective dimension as an intrinsic feature of the be -ing aspect, which includes but is not limited to the notion of progressivity.
4.3 Pure future, predetermination and matter-of-courseness
4.3.1 Volition-neutrality and non-agentivity
One explanation that has been proposed for the emergence of will/shall be -ing is that it allows speakers to avoid volitional overtones when referring to the future. Although we consider the meaning to be richer than this (see section 4.3.2), it is undoubtedly a prominent feature.
As has been previously observed (e.g. by Coates Reference Coates1983; Palmer Reference Palmer1979: 119, 133–4, Reference Palmer, Facchinetti, Krug and Palmer2003: 16), when modals combine with be -ing, they generally lose the tendency to convey root modality.Footnote 15 Instead, the function of the modals is either epistemic (conveying judgements about situations in present time), or a predictive future time reference.
Like shall + bare infinitive, shall + be -ing cannot be used epistemically, and it thus has a narrower range of application than will be -ing. Combined with be -ing, will and shall express neither volition nor obligation respectively. The modals take on a seemingly ‘pure’ predictive meaning, which seems to confirm our view that the relation between the speaker and the predication as a whole takes precedence over the relation between the grammatical subject and the verb. We would argue that be -ing combined with will/shall also has an effect on agentivity. Firstly, we compare will be -ing with be -ing and then with will + be + passive past participle. Contrast the following pair of examples:
(18) (a) I don't listen to you when you are being morbid. (T. Williams, quoted by Mettouchi Reference Mettouchi, Borillo, Vetters and Vuillaume1997: 192)
(b) ?? You’ll be being morbid.
In (18a), as pointed out by Mettouchi, the non-progressive present would also be possible, with no implication for the duration of the situation. The temporal clause when you are being morbid is just as temporary as when you are morbid. Mettouchi correctly states that the main difference between the progressive and the non-progressive form is one of agentivity. With the non-progressive form, the situation is stative and the subject is characterised by the predication. With be -ing, the subject is presented as an agent involved in an activity (actively behaving in a morbid way). In addition, this pattern further implies a negative or reproachful attitude by the speaker. In the case of will/shall be -ing, however, such a recategorisation of the stative verb into an activity verb is not possible, as exemplified in (18b).Footnote 16 Our hypothesis is that, combined with will and shall, the be -ing form loses its potential to be interpreted agentively. The subject cannot be perceived as a volitional agent, which accounts for the unacceptability of (18b). Moreover, it is questionable whether the verb refers to an activity in (19), since the inanimate subject cannot have a volitional role:
(19) With few exceptions, your car will be waiting for you at dockside, airport, railroad station or hotel when you arrive . . . (Brown E36)
The speaker refers to a typical case, relying on knowledge of car rental companies to make a prediction (it is standard practice for customers to have a car waiting for them where they arrive). With modals, the be -ing form signals that the predication is based on the speaker's representation, i.e. on his/her knowledge of a predetermined situation, but not on the agentive role of the grammatical subject, in contrast to modals + bare infinitive. Contrast the following pairs of examples:
(20)
(21)
In (20a) and (21a), it is the subject's ability or volition that is negated. In (20b) and (21b) the speaker's judgement takes precedence over the subject's agentive role.
It follows from this that the difference between situation types is blurred by will/shall be -ing, which might explain why this construction may occur with almost any verb type as long as no active role is assigned to the grammatical subject: for example, punctual verbs (see (22)) – with the notable exception of be (see (18b)) – and stative verbs:
(22) He was speaking in shorthand in a sense and made it very clear that Scottish circumstances would be taken fully into account, and that I (sic) will be reaching a decision in due course. (BNC K5L 1550)
(23) Not surprisingly Antrim will be relying on the same side that finished that day. (BNC HJ3 2045)
(24) In fact I don't think I will be requiring your services at all. (BNC HTU 2099)
However, in the case of the passive, will be -ing is extremely restricted. In the nearly 100 million words of the BNC, just five examples were found of will be being + past participle. That each of the five occurs in a different genre, two in spoken and three in written data, suggests that it is not genre per se, but probably more general stylistic considerations, that restrict the currency of the pattern.
(25) This can, in fact, be the most productive teaching time, because the children will be involved in very real problems and their skills will be being rigorously tested. (BNC EV4 1437, academic social science)
The be -ing form is taken to introduce a viewpoint specification which overrides the situation type because aspect here is intrinsically linked to the speaker's mental representation of a future situation.
4.3.2 Predetermination and ‘matter-of-course’ use
Leech's (Reference Leech2004: 67) notion ‘future as a matter of course’ and Huddleston & Pullum et al.'s (Reference Huddleston and Pullum2002: 172) ‘already decided future’ are closely related, and both relevant to will/shall be -ing. They involve the speaker's knowledge about the situation that is predicted. As such, the prediction is represented as predetermined, and the label ‘colourless’ future (Samuels Reference Samuels1972) seems inappropriate. The use of will/shall be -ing is often said to be more tactful than will/shall + bare infinitive, because it implies that future events will happen as a matter of course – in the natural course of events – and not as a result of the speaker's, or anyone else's, involvement (see e.g. Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 217). However, neutrality may be feigned in a subtle way in order for the speaker to better manipulate the addressee: the future event is presented as incontestable, and as having been settled prior to the time of utterance:
(26) After they had gone, there was an awkward little silence, then Stephen said gently: ‘I hope everything went as well as you expected, Jo?’
‘Oh yes! Better, really. We shall be going to London at the end of this month . . . We've been offered a very good engagement, you see.’ (LOB P05)
The speaker merely reports what has been planned in accordance with the engagement. Only shall/will + bare infinitive unambiguously refers to a decision that is being made at the time of utterance.
Our contention is that with will/shall be -ing a predetermined situation is implied in utterances both of Type 1 and Type 2 aspect. In (27), we have a situation that is construed as future in progress, but also interpretable as ‘already decided’:
(27) He gave the address of the destination to him.
‘Axel will be waiting for you,’ he said to her in a whisper. ‘He'll take you to the boat. I'll be there shortly after six.’ (W. Boyd, The Blue Afternoon, p. 281)
In (27), although there is no temporal clause, Axel will be waiting for you implicitly refers to when you get there. This implicit future reference point provides a temporal frame relative to which the situation is viewed as in progress. And yet, this situation is also part of a schedule that has already been planned at the time of utterance. But the concept of predetermination may also be construed in a more abstract way, since the effect of adding be -ing is to convey the speaker's judgement based on his/her knowledge, rather than on some plan:
(28) I regret to now inform you that we will be terminating all our contracts with you as of Monday 22nd of July 1991. (ICE-GB W1B-028 #46:4)
The decision to terminate the contracts is presented as independent of the situation of utterance, i.e. neither the speaker nor the addressee is supposed to be in any way responsible for it. This decision is also assumed to be in accordance with some norm, even if the speaker expresses conventional regret. If will + bare infinitive had been used, this decision would have been felt to convey the speaker's intention and might have been considered arbitrary or debatable. With be -ing, however, the preconstruction of the predication leaves no room for negotiation, which allows the speaker to impose disputable decisions on the addressee without having to justify them. On the one hand, the addressee is spared the trouble of an unpleasant explanation, which may sound polite. On the other, the addressee may be manipulated into accepting a decision that is not as normal as it is presented to be. The use of will be -ing here amounts to an understatement.
Most linguists have shown the effect of be -ing on modals without paying attention to its effect on the speaker–hearer relationship. Palmer (Reference Palmer1979: 133–4) rightly notes that with will/shall be -ing, ‘the speaker does not want the agreement of his audience’, but the implication of this has not been investigated so far. We propose that the will/shall be -ing construction is not only a volition-disclaimer, but also a responsibility-disclaimer. As a volition-disclaimer, it has an effect on the relation between the grammatical subject and the verb, as stated earlier. As a responsibility-disclaimer, it has an effect on the relation between the speaker and his/her utterance, and consequently on the speaker's modal attitude towards the hearer.
To a certain extent, will be -ing may be considered to serve an evidential function. There seems to exist typological evidence of a formal parallel between the category of evidentiality and progressive aspect. In many languages, for example Modern Western Armenian, the evidential is derived from the perfect. Donabédian (Reference Donabédian2001: 436–7) observes that the Western Armenian progressive frequently co-occurs with the evidential, and argues that the progressive functions as ‘the non-completed modal counterpart of the perfect’. In English, the increase in use of be going to – especially in AmE – and of will be -ing – in BrE – suggests that these constructions are developing along two separate but parallel evidential paths. It is striking that the meaning of be going to and will be -ing encapsulates the three distinguishing features of evidentials – namely inference from situational evidence in the case of be going to, and epistemic inference, report or surprise in the case of will be -ing.Footnote 17 This may well account for the richness and apparently paradoxical meaning of will be -ing: it is used in reaction to directly perceived evidence as well as in pseudo-neutral quotative statements. In the quotative function, a parallel may be drawn between will be -ing and other modal devices. Modal adverbs (such as reportedly) or passive phrases (such as is reported to, is scheduled to) likewise imply that the speaker is not responsible for what has been scheduled or said. In this way, the speaker may disclaim responsibility for past or present events. Symmetrically, will be -ing signals that the speaker is not taking responsibility for future events:
(29) He will be taking part in an international conference on the space project which will meet on January 30 in London. (LOB A03, Press reportage)
(30) The former White House intern reportedly signed an agreement on Friday with St. Martin's Press for a book about her affair with President Clinton . . . The book is scheduled to be published in February, the Post said, citing a source familiar with the negotiations. (CNN online)
(31) A recent New Scientist article reported that within five years most Western countries will be issuing their citizens with a machine-readable passport that will carry with it the threat of global surveillance of innocent travellers. (www.lextutor.ca/concordancers/concord_e.html)
In these examples, the speaker dissociates himself/herself from the source of information, which is made explicit in (31) with the introductory clause ‘a recent New Scientist article reported’. This implies that the future situation will actualise as part of some plan arranged independently of the speaker, who disclaims responsibility for this arrangement. In these examples, the speaker aims to construct a future referential situation, even if he/she presents him/herself as not directly involved. However, the speaker may adopt a different stance when faced with a referential situation or a statement that runs counter to his/her norm.
(32) ‘A garment, you little thief. Mamzer! I'm sorry for your father. Some heir he's got! Some Kaddish! Ham and pork you’ll be eating, before his body is in the grave.’ (S. Bellow, Herzog, p. 131)
(33) ‘So you mean you haven't read them?’ she says, sticking determinedly to the subject of the documents. ‘You’ll be telling me in a minute you haven't had time.’ ‘Of course I've read them.’ (J. Le Carré, The Constant Gardener, p. 58)
The speaker considers the implications of the facts he/she is faced with. The meaning could be glossed as ‘If things carry on this way, I predict that this is what will happen next’ (see Leech Reference Leech2004: 69). The speaker does not want this situation to actualise and his/her prediction aims at warning the hearer against the negative and inevitable consequences of his/her behaviour, should no change occur. The speaker's prediction is therefore to be understood as an evaluative judgement about actual surprising or abnormal facts. At the same time, the speaker also intends to influence the hearer's attitude. In other words, a pragmatic relation is restored in an indirect way, by means of a negative evaluative judgement.
Thus the impression of neutrality is in fact created by the stance that the speaker adopts vis-à-vis the hearer. Because the speaker does not commit himself/herself to the prediction that is made, his/her prediction may, in some cases, sound neutral. But his/her evaluative role may also be marked, as in the last two examples.
Once again, the dual function of the be -ing form gives insight into this apparent paradox. As we have seen, the be -ing form combined with modals allows the speaker to identify a situation from his/her vantage point. Therefore, it should come as no surprise that the current state of affairs is interpreted in a negative or hyperbolic way. This kind of overstatement is only expressed with will be -ing, as opposed to shall be -ing. Shall be -ing is found in predictive statements where future actualisation is envisaged by the speaker, but not in epistemic or evaluative judgements.
4.3.3 Epistemic use of will be -ing
Coates (Reference Coates1983: 180) points out that aspectual marking forces modals to be interpreted epistemically. Tsangalidis (Reference Tsangalidis1999: 204-7) pushes further the implications of this observation. He draws upon Blokh's (Reference Blokh1994) analysis, which shows the tendency of the be -ing form and the perfect infinitive to function modally ‘under modal government’.
Consequently, the combination will be -ing is ‘underspecified for Tense’, as Tsangalidis (Reference Tsangalidis1999: 213) notes. The prevailing meaning is one of prediction about a situation that may be present as well as future. In any case, temporal adverbials or information from the context are needed to assign temporal reference to the verb phrase. In (27), for example, a future temporal locator is inferred from the context, allowing Axel will be waiting for you to be construed as referring to an implicit when you get there. Similarly in (19), future time reference is made possible by the temporal clause. However, if a temporal adverbial such as now or by now is present, the reference is to the time of utterance:
(34) ‘Should we disturb her?’ ‘I think not,’ he replies. ‘She will be sleeping by now.’ (A. Brookner, Family and Friends, p. 176)
(35) He is aware that the inhabitants of the bungalows are watching him curiously out of their windows. He knows that back at Wren House Muriel will be heaving herself to her feet with a loud sigh and asking if anyone wants anything else before she puts her weary bones to bed. (A. Brookner, Family and Friends, p. 110)
The be -ing form provides the link between the state of affairs that the speaker intends to explain and his/her mental representation.Footnote 18 In (34), the be -ing form marks a relation of temporal coincidence between the situation referred to by the verb sleep and the current state of affairs. The situation is viewed as ongoing and is simultaneously identified by the speaker as the reason why she should not be disturbed. Similarly in (35), will be heaving herself to her feet marks temporal coincidence with the time of utterance and with the situation are watching.
In (34) and (35), the speaker aims to construct a referential situation but does not have direct access to it, hence it is a case of epistemic modality. This kind of epistemic usage is typically found with third-person subjects. In (34) and (35), the speaker's spatial location is different from that of the grammatical subject. The role of the be -ing form is to establish temporal coincidence with the time of utterance. In the following examples, will be -ing also conveys epistemic meaning. However, the function of be -ing is not only to identify the situations temporally:
(36) In the drawing-room, Hal glances unobtrusively at his watch, computing some timetable of his own. ‘Yes,’ says Sofka, who has followed his glance. ‘You will be wanting to get back.’ (A. Brookner, Family and Friends, p. 118)
(37) Mrs Batley's response to this was to say quietly: ‘You’ll be wanting a wash; will you come up?’ (LOB P21)
These are second-person utterances directly indexed to the situation of utterance. In combination with want, will be -ing conveys a modal meaning. In (36), will be -ing signals that Sofka is seeking to interpret Hal's glancing at his watch in an attempt to understand his current state of mind. Hal's glance is subjectively identified as meaning ‘you want to get back’. Will be -ing serves a similar function in (37). Mrs Batley is attempting to interpret Linda's state of mind by conforming to traditional standards of hospitality.
In both (36) and (37) the speaker is commenting on a referential situation. It is in this case that will be -ing takes on an interpretive meaning. By contrast, in (34) and (35), the speaker is attempting to establish facts in order to construct a referential situation. Here will be -ing marks temporal coincidence and the situation is viewed as in progress at the time of utterance.
5 Historical development
5.1 Internal factors: possible paths of development
Although shall be -ing and will be -ing have both been in use since at least Middle English (Mustanoja Reference Mustanoja1960: 591; Strang Reference Strang1970: 208) and possibly earlier (Visser Reference Visser1963–73: 2412), we have no dating as to the emergence of the respective uses. Probably, as Samuels (Reference Samuels1972: 57) argues, the earliest uses were aspectual, i.e. representing situations as future-in-progress. Most of the seventeenth- and eighteenth-century examples in ARCHER (BrE) are of Type 1 aspectual interpretation, (38), and there are no clear cases of Type 2 until the nineteenth century (39), which is in fact from an Irish dialect speaker:
(38) Our Men have brought in several little Prizes, and will be catching up some or other daily. (ARCHER 1665inte.n2b)
(39) ‘. . . I want to know from you when you’ll be goin’ to your uncle's, at Mullaghmore.’ (1847carl.f5b)
Unfortunately, the sampling frame of ARCHER is too recent, and cases of shall/will be -ing too infrequent, to enable a detailed understanding of their functional evolution.Footnote 19 However, in the case of will be -ing in BrE, there is enough material to discern possible evolutionary developments.
(a) Contextual reinterpretation of volition-neutrality
Samuels (Reference Samuels1972) treats future-in-progress (our Type 1) as the original aspectual meaning of shall/will be -ing, and says that volition-neutrality was a feature of meaning incidentally associated with it. Volition-neutrality, he argues, later transferred to an ‘actualising’ function of will/shall be -ing: ‘especially when it is used with non-durative verbs, it focusses more attention on the action of the lexical verb (be) leaving and less on the auxiliaries shall/will’ (Reference Samuels1972: 57). The transfer of meaning was motivated by the tendency among other constructions referring to the future (e.g. I shall leave, I will leave, I'll leave, I'm going to leave) to convey ‘some degree of modal nuance’ (Reference Samuels1972: 57). He adds: ‘It [will/shall be -ing] is therefore becoming more and more used as a colourless future without overtones of intention, wish, irritation and the like, and irrespective of whether the context demands a marked aspectual form’ (Reference Samuels1972: 57).
With some qualifications Samuel's account seems plausible: it integrates the evolution of the aspectual meaning with the volitional-neutral component of meaning. In contemporary terminology we might express the generalisation of volition-neutrality as a case of ‘context-induced reinterpretation’ (Heine et al. Reference Heine, Claudi and Hünnemeyer1991; Heine Reference Heine, Joseph and Janda2003) or ‘pragmatic strengthening’ (Hopper & Traugott Reference Hopper and Traugott2003), i.e. a feature that was originally only conversationally implicated becomes routinised as a feature of the construction as a whole. Together with the partial loss of aspectual specificity (through the emergence of Type 2), this would have resulted in the meaning shift of shall/will be -ing.
The historical data in ARCHER show mixed support for Samuels’ arguments. On the one hand, all occurrences of will be -ing in ARCHER up to 1900 are in second- and third-person contexts: according to Aijmer (Reference Aijmer and Fisiak1985), cited in Traugott & Dasher (Reference Traugott and Dasher2002: 223), it is in such environments that a reported intention is most likely to be reinterpreted as a non-volitional prediction; see (40).Footnote 20
(40) Her cully then desires her to leave it off, and tells her that she might live better and more at ease in her private lodgings, telling her that if she will do so, he will not be wanting to contribute handsomely to her maintenance. (1673kirk.f2b fiction)
On the other hand, in most cases it is difficult to say whether the motivation for using will be -ing is to disclaim volition. It is frequently unclear whether the utterance would be interpreted as volitional if will + bare infinitive had been used instead:
(41) Our Men have brought in several little Prizes, and will be catching up some or other daily. (1665inte.n2b news)
(42) If you have a handsome wife, every smooth-faced coxcomb will be combing and cocking at her. (1680otwa.d2b drama)
Further, Samuels’ view that the expressive value of shall/will be -ing is limited to that of colourless future overlooks the notion of predetermination/matter-of-courseness which, as argued in section 4.3.2, is another important characteristic of the constructions. This feature is arguably detectable in early cases such as (42) above, where the speaker seems to construe the predicted event as normal and expected.
(b) Spread by analogy with the futurate use of the present
Another possible factor in the development of the aspectual meaning, as well as the notion of predetermination, is extension by analogy with the futurate use of the progressive (e.g. John is leaving tomorrow). Numerous synchronic discussions of the progressive have claimed that the notion ‘situation in progress’ can be extended to cover the futurate use (cf. Hirtle Reference Hirtle1967; Declerck Reference Declerck1991; Miyahara Reference Miyahara1996; and Williams Reference Williams2002). Declerck (Reference Declerck1991), for example, claims the ‘temporal perspective is shifted, i.e. . . . the situation is represented as if it were lying in the present rather than in the post-present sector’ (Reference Declerck1991: 67).Footnote 21 In Williams’ (Reference Williams2002) account, the mechanism seems more metonymic: by virtue of a part of the situation (e.g. the planning phase) being under way, the entire situation is represented as in progress.
It seems more difficult to extend progressivity to the Type 2 use of will/shall be -ing (although see Williams Reference Williams2002: 95–8). ‘Situation in progress’ would need to be conceptualised very schematically, since the use of will/shall ordinarily (as in Type 1) locates the orientation time of the situation in the future, not the present.
In several respects the putative developments outlined in (a) and (b) are consistent with grammaticalisation:
– That one meaning (progressivity) has been demoted, while others (volition-neutrality, matter-of-courseness, etc.) have been promoted by pragmatic strengthening. To the extent that the future situation cannot literally be in progress, but is based on the speaker's viewpoint and expectations, the newer meanings can be seen as reflecting increasing subjectification in the sense of Traugott (Reference Traugott1989, Reference Traugott, Stein and Wright1995). That is, over time the function of the construction has become increasingly based in the speaker's subjective reasoning processes.
– There is ‘layering’ (Hopper Reference Hopper, Traugott and Heine1991) in that the original aspectual meaning in Type 1 has survived and co-exists, albeit as a minority use, with Type 2.
– The functional generalisation of the construction makes its meaning overall more indeterminate. This is again supported in ARCHER, and also in our twentieth-century corpus data (see table 1 above).
A more problematic question is how to relate the constructions to the grammaticalisation ‘paths’ posited by language typologists. Each construction is a merger of a modal and a progressive construction, whereas typologists’ proposed paths of change typically involve single constructions. According to Dahl (Reference Dahl1985: 93; 2000: 10–11) and Bybee et al. (Reference Bybee, Perkins and Pagliuca1994: 141), across the world's languages the normal grammaticalisation path of progressive constructions is to develop into the superordinate category of imperfective; that is, incorporating habitual and stative meaning, in addition to progressivity. If examples like I'll be leaving tomorrow at 8 p.m. are construed perfectively (as claimed by e.g. Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 172 and Leech Reference Leech2004: 67) then it would seem that the semantic shift in English will be -ing is a counterexample to the normal path. We argued earlier, however, that the constructions have become underspecified for aspect (see section 4.1), and they are increasingly compatible with stative verbs (section 4.3.1); this could then be taken as tending towards imperfectivity, as predicted in the path.
Concerning the development of modals such as will into futures, Bybee et al. (Reference Bybee, Perkins and Pagliuca1994: 256) propose a path from desire to prediction:
(43) desire > willingness > intention > prediction
While (43) fits well with will + bare infinitive, our historical data do not allow us to check whether it also applies specifically to will/shall be -ing.
5.2 Social-stylistic factors
5.2.1 Decline of prescriptive will/shall rule
Given that will/shall be -ing have been available for several centuries, we might ask why they did not significantly spread in standard BrE and AmE until the twentieth century. One factor may be the continuing influence in the standard varieties of the prescriptive rule for shall/will.
This rule has been expressed in various ways, but in its most influential form it could be summarised as follows: (i) in first person, shall expresses a volition-neutral prediction, whereas will expresses volitional meaning such as determination, insistence, threats, promises, etc.; (ii) in the second and third persons, will expresses a volition-neutral prediction, and shall has a volitional function (see e.g. Fenning Reference Fenning1771).
The rule seems only to be seriously undermined towards the end of the nineteenth and start of the twentieth centuries (Denison Reference Denison and Romaine1998: 167–8; Smith Reference Smith1996: 142). It is generally acknowledged that as the rule broke down, shall gave way increasingly to will (including its weakened form ’ll: see e.g. Barber Reference Barber1964: 134).
The weakening of the rule would presumably have benefited will be -ing. Initially restricted to second- and third-person subjects, during the twentieth century it gradually generalised to select any grammatical subject – including first person, where in the volition-neutral function, shall + bare infinitive was previously dominant: see Figure 3.
Subject selection with will be -ing in twentieth-century written BrE: absolute frequencies in the Brown family of corpora
In non-standard varieties it is more probable that the prescriptive rule on shall/will was not observed: here, presumably, will be -ing would have been free to spread much earlier.
5.2.2 Change led by conversational norms
The early spread of will be -ing seems to show a typical sociolinguistic pattern of change, in that examples from ARCHER and the B-LOB corpus suggest that it formerly predominated in characteristically oral registers (see section 3.4 above).
But is the spread of will be -ing in later twentieth-century written language similarly driven by adoption of patterns from informal speech? In the last decade a number of publications have linked the rapid pace of change in standard written English to increasing use of patterns characteristically associated with informal speech, e.g. contractions (Hundt & Mair Reference Hundt and Mair1999); the be going to future (Mair Reference Mair, Hickey and Puppel1997), and the present progressive (Smith Reference Smith, Peters, Collins and Smith2002). Similarly, features that are rare in informal speech, and therefore negatively correlated with colloquial speech, have declined in written usage: e.g. the passive (Seoane & Williams Reference Seoane, Williams, Dossena and Taavitsainen2006), and the auxiliaries shall and be to (Leech et al. Reference Leech, Hundt, Mair and Smith2009).Footnote 22 In the case of will be -ing, registers closer to the vernacular probably provided a trigger for its broader diffusion in the early to mid twentieth century.
However, evidence from the BNC suggests that by the late twentieth century, the continuing spread of will be -ing was not being driven by colloquial usage. The BNC genres with the highest frequencies are news-related varieties, such as broadcast news and press reportage, with typically more than 300 occurrences pmw; casual conversation ranks well down the list of genres, with only 149 occurrences pmw.
6 Conclusion
Although will be -ing has been around for a considerable time, it started to gain a significant foothold among future-time expressions in standard English only in the twentieth century. In an interesting case of regional divergence, our findings show that much more modest gains have been made in AmE than in BrE.Footnote 23 In AmE be going to has possibly played a bigger role in keeping will be -ing at bay. Shall be -ing has not nearly prospered to the same extent, although again it appears to be used more frequently in the twentieth century than previously. Possibly shall be -ing has been adversely affected by a demise of shall in general, which tended to favour will (and its reduced forms) as the force of the shall/will rule started to fade. Will be -ing may have benefited in this process, since – like shall + bare infinitive – it has the characteristic of prediction/volition-neutrality.
We lack a detailed historical record of the spread of the shall/will be -ing construction, but ARCHER suggests that initially it followed a typical sociolinguistic pattern of change, diffusing among spoken or speech-like, mainly informal, genres. In more recent times it has spread to a much wider range of genres, such that it occurs less prolifically in conversation than in comparatively formal genres, especially media-based ones such as news reportage, professional letters and parliamentary debates.
With respect to contemporary meanings, we have argued (with Huddleston & Pullum et al. Reference Huddleston and Pullum2002) that in combination with a modal auxiliary, as well as in the futurate use, be -ing does not primarily convey a sense of progressivity. In such cases, the situation is viewed from a prospective or subjective viewpoint. This viewpoint specification takes precedence over the relation between the grammatical subject and the verb, which subjectivises aspect. The future-in-progress interpretation of shall/will be -ing is still possible where the context or temporal adverbials, or both, induce such an interpretation. However, in very many cases aspect is not a useful discriminating feature. We would therefore suggest that Type 2 is underspecified for aspect, rather than perfective (cf. Palmer Reference Palmer1990; Leech Reference Leech2004; Huddleston & Pullum et al. Reference Huddleston and Pullum2002).
We accept the general view that there is no implication of volition or intention on the part of the subject of shall/will be -ing. However, the construction cannot simply be regarded as a ‘volition-disclaimer’, since it may be used where volition is not relevant. Rather, prediction is based on some predetermined situation: some plan may have been made independently of the speaker (see Palmer Reference Palmer1979: 134; Huddleston & Pullum et al. Reference Huddleston and Pullum2002), or, in a more abstract way, the speaker may conform to some norm (Leech Reference Leech2004). We argue that this feature has generalised across both aspectual types, again blurring the boundary between them.
The speaker is bound to adopt different subjective attitudes to his/her utterance depending on the type of discourse. Therefore, the epistemic and the ‘future as matter of course’ meanings should not be regarded as entirely distinct, but rather as equally available options provided by the predictive judgement. The combination of will with be -ing extends the possibilities of temporal reference beyond the frame of the future by allowing the speaker to form an epistemic judgement about present situations. The selection of meaning then depends on the discourse type (e.g. direct vs reported speech), on the context, and on temporal adverbials. Crucially, the sense of ‘pure future’ seems to be an illusion created by evidentiality.
Our account of the functional mechanism of change draws on the short diachronic sketch outlined by Samuels (Reference Samuels1972). An advantage of Samuels’ account is that it provides a motivation for the loss of aspectual specificity (progressivity) concomitantly with a general incorporation of volition-neutrality. Such a development is in accord with other accounts hypothesising that, over time, the meaning of grammatical constructions becomes increasingly based in the speaker's subjective reasoning processes (cf. Traugott Reference Traugott1989, Reference Traugott, Stein and Wright1995; Sweetser Reference Sweetser1990). However, Samuels’ account does not address the richness of the meaning that shall/will be -ing gained: the notion of predetermination or ‘matter-of-courseness’ – not only volition-neutrality – seems to have been implied early on, and has generalised to all future uses of will/shall be -ing. Further, the future-in-progress meaning has become marginal, as the construction becomes prevailingly aspectually underspecified.
The matching corpus data on shall/will be -ing from ARCHER and the Brown family (see sections 3.1 and 3.2) do not point to contact effects of AmE on BrE. In BrE, the incidence of both constructions is not only higher but in the case of will be -ing has increased significantly faster. Clearly, the two varieties are capable of taking separate paths, even if they share a common trait of raising their use of the progressive construction as a whole.
One type of contact that merits closer investigation in the future is that from varieties within the British Isles. It has been claimed that in Middle English the progressive, including shall/will be -ing, was more frequent ‘in the north’, and from there spread to other regions (cf. Mustanoja Reference Mustanoja1960: 586; Strang Reference Strang1970: 208; Fischer Reference Fischer and Blake1992: 251; and Mossé Reference Mossé1938: 35–6). It is perhaps a reflection of these origins that in recent (1990s) BrE, across the demographically sampled spoken part of the British National Corpus, the constructions consistently exhibit higher frequencies in the northern and western regions of the British Isles, and lower frequencies in the south-east (see tables A5 and A6). Such provisional findings clearly need further scrutiny, including comparison of the functions and uses of the construction in the respective regions, and incorporation of equivalently sampled diachronic corpora.
Appendix 1. Queries run in CQP for retrieval of shall/will be -ing from the corpora
(a) Basic query:
[word=‘will|\’ll|wo’%c] [pos=‘AV0|ORD|XX0’]{0,4} [pos=‘A.*|D.*’]? [pos= ‘PP.*|PN.*|N.*’]{0,2} [pos=‘AV0|ORD| XX0’]{0,4} ‘be’%c [pos=‘AV0|ORD|XX0’]{0,4} [word=‘.*ing’%c]’
(b) Query run on BNC, using BNCweb (CQP edition):
[word=‘will|\’ll|wo’%c] ([pos=‘AV0|ORD|XX0’] | <mw_pos=‘AV0|ORD|XX0’> []* </mw_pos>){0,4} [pos=‘A.*|D.*’]? [pos=‘PP.*|PN.*|N.*’]{0,2 ([pos=‘AV0| ORD|XX0’] | <mw_pos=‘AV0|ORD|XX0’> []* </mw_pos>){0,4} ‘be|bee’%c ([pos=‘AV0|ORD|XX0’] | <mw_pos=‘AV0|ORD|XX0’> []* </mw_pos>){0,4} [word=‘being’%c]’
Appendix 2. Additional tables
The progressive in BrE and AmE, 1650–1990: ARCHER data (based on Hundt Reference Hundt2004: 69)
Genre distribution of will be -ing in twentieth-century written BrE: raw and normalised frequencies in the Brown-family corpora
Notes:
a. Figures in parentheses represent raw frequencies.
b. In the rate of change column, ***indicates a probability of <.001
Genre distribution of will be -ing in 1990s BrE: top ten genres, plus conversation, in the BNC (estimated frequencies)Footnote 24
a‘S’ signifies spoken genres and ‘W’ signifies written genres.
bDispersion = the proportion of texts in the genre category that contain at least one instance of will be -ing.
Genre distribution of shall be -ing in 1990s BrE: highest ranking genres, plus conversation, in the BNC
a‘S’ signifies spoken genres and ‘W’ signifies written genres.
bDispersion = the proportion of texts in the genre category that contain at least one instance of will be -ing.
Distribution of will be -ing based on region of the speaker: data from the conversational, demographically sampled part of the BNC
Distribution of the progressive based on region of the speaker: estimated frequencies in the conversational, demographically sampled part of the BNC
