Using Early Modern English (1640-1710）(part of the Helsinki Corpus of English Texts), Gotti (2003: 267-300）finds that shall expresses volition at a ratio of 11 out of 10,000 words, and prediction at a ratio of 39.4 out of 10,000 words across various text types.This forms a sharp contrast with Coates’ findings about the frequency of the various modal uses of shall in contemporary English based on LOB (Lancaster-Oslo-Bergen Corpus): shall expresses volition at a ratio of 2 out of 10,000 words, and prediction at a ratio of 10 out of 10,000 words.This shows that shall is diminishing in expressing both deontic and epistemic modality.Coates (2003: 331-348）finds that women use epistemic modal forms (including MVs）more frequently than men based on a corpus of spontaneous conversations between women friends collected over 15 years or so.Coates’ conclusion is drawn as follows:
In all kinds of contexts, where we need to be sensitive to the face needs of others, where we need to qualify assertions to avoid total commitment to a particular point of view…then epistemic modal forms are the valuable resources for speakers, the resources we should never under-estimate.
MVs are supposed to possess a number of properties.Heine (1995）argues that the following properties are criteria in the analysis of MVs:
a.There is some force that has an interest in an event either occurring or not occurring (F).
b.That event is to be performed by some agent (A).
c.The event is dynamic (D).
d.The event has not yet taken place at reference time and, if it does take place, that will be at a time later than reference time (L).
e.The event is non-factual, but there is a certain degree of probability that it will occur (P).
Coates (1995: 59）takes the same line with Heine, but adds that another crucial distinction between forms expressing root possibility and epistemic possibility in English is that the latter involve subjectivity (S).Using the above criteria, it is possible to distinguish root and epistemic MVs in English as follows in Table 2.3.
However, viewed diachronically, the development of the modal meanings of MVs does not follow the ideal model.First of all, historical evolution results in the vague criteria regarding root and epistemic readings.Traugott (1989）discovers how can came to develop an epistemic reading from the dynamic modal sense of ability.These can be confirmed through two aspects: (a）in many cases, the MV can could provide the missing negative form in the epistemic must paradigm (the invariant form can’t could express “it’s necessarily the case that…not…”); (b）can is used in interrogative constructions to express epistemic possibility (e.g.Can it be true?).Also, regional discrepancy results in the vague boundary between MVs as well as between the modal meanings of individual MVs.For instance, can is used in American English as an epistemic sense of prediction, which is not common in British English (cf.Coates, 1995: 63).
(56）We hope this coding system can be useful.
Example (56）means “we hope there is a chance that this system will be useful” in American English.In contrast, a British speaker tends to say “we hope this coding system will be useful”.For many British, can is subjective and totally lacks objectivity; Americans, nevertheless, regard can as a hedge and do not think that it harms either subjectivity or objectivity that the MV contains in such a case.
As a result of the vagueness between the root and epistemic readings of MVs, some linguistic devices have come to act as the means of ambiguity resolution.Denison (1992: 229-254）explores the use of may have, might have, may well, and might well through the historical perspectives.According to Denison (ibid), have is an auxiliary implying unreality or lack of fulfillment; well has the interpretation of “it is highly possible”, or “it is indeed possible” and is used counterfactually; the combination or blending (in terms of historical linguistic variations）of may + have, might + have, may + well, and might + well makes it possible for may or might to be encoded and decoded as epistemic rather than deontic.Consider the following:
(57）They might have recognized us.(“It is tentatively possible that they recognized us”)
(58）English may well have become bankrupt had it not had the influx of vocabulary.(“It may be the case that English became bankrupt without the influx of vocabulary”)
Examples (57）and (58）indicate that the use of multiple modal devices in English is partly a result of historical development.
On the other hand, seen synchronically, linguistic devices for certain modal senses are quite rich in English.Coates (1995: 64）points out that there is a trend for the MV can to co-occur with syntactic features such as inanimate subject and stative verbs to support an epistemic reading of the MV.Also, epistemic possibility has many exponents, notably maybe, perhaps, I think, possibly, probably and the MVs may, might and could.Moreover, people tend to double the use of modal devices in communication.Take the following for example:
(59）That may be yellow fever, I’m not sure.
(60）It may be a few minutes late, but I don’t know.
(61）I think it’s unlikely actually but he might do it today.
In Examples (59）and (60), the modality pattern is MV (may）+ clause, and in Example (61), clause (I think）+ adjective (unlikely）+ MA (actually）+ MV (might).Obviously, the synchronic cause for the use of multiple modal devices is language users’ intention of clarifying their attitude or judgment.