146 O L G A F I S C H E R A N D W I M VA N D E R W U R F F
(38) |
Witodlice næfde |
godes |
gelaðung paulum |
to |
lareowe: |
gif |
se |
||
|
truly |
not-had |
god’s |
church |
Paul |
as |
teacher |
if |
the |
|
halga martyr stephanus swa ne |
bæde. |
|
|
|
|
|||
|
holy |
martyr Stephen |
so not |
asked |
|
|
|
|
|
|
‘God’s church would not have had St Paul as teacher, if the holy martyr |
||||||||
|
Stephen had not asked for this.’ |
(ÆCHom I, 3 202.113) |
|
|
|||||
In ME this preterite subjunctive was often replaced by a periphrastic construction with a modal verb, but the old subjunctive form itself, which had now fallen in with the preterite indicative, remained in use too, and served by itself as a modal marker especially in present-tense contexts:
(39)But soore wepte she if oon of hem were[SUBJ] deed, | Or if men smoot [INDIC] it with a yerde smerte
‘But she would weep sorely if one of them died or if someone hit it smartly
with a stick.’ |
(Chaucer, GProl. 148–9) |
Since past-tense indicative modal verbs were also used in such hypothetical situations in OE, the past tense of modals acquired a similar role, as we can see from the translation of ME wepte into PDE would weep, and the past tense of modal verbs gradually lost their ‘pure’ past-time reference; for details, see below. When the hypothetical situation was placed in a past-tense context, the pluperfect came to be used with the same modal colouring; this was a new development in ME:
(40)I dar wel seyn, if she hadde been a mous, | And he a cat, he wolde hire hente anon
‘I dare say, if she had been a mouse, and he a cat, he would have caught her
at once.’ |
(Chaucer, MillT 3346–7) |
Note again, that PDE would use a (plu)perfect in both main and subordinate clause, whereas in ME the past modal wolde alone could still function there.
3.3.4 |
The story of the modals |
|
In the previous section, we have looked at changes in the way mood was expressed in English, especially the relation between subjunctive and modal verbs, but we have not yet considered the modal verbs themselves and the changes that took place there. The ‘modal story’ is particularly interesting because the original modal verbs have changed much more radically in English than in any of its sister languages. In English the modals have developed into what Warner (1993: 49ff.) has called ‘anaphorical islands’, i.e. they show an ‘independent “word-like” status’, with non-transparent morphology, in contrast to full verbs which have transparent morphological inflections of person and tense. The modals in other Germanic languages, on the other hand, have retained most of their verbal features. Additionally, the story is of theoretical interest because it has been used to
Syntax 147
support a generative linguistic view of change whereby certain grammar changes may have been ‘radical’, i.e. the idea that seemingly unrelated changes on the surface may be related to one, deeper and more abstract change in the base. Such evidence is important, since it may not only tell us more about how syntactic change takes place but may also serve as empirical evidence for the existence of such an abstract rule system, particularly for the degree of abstractness of this system; more generally, it may tell us more about the role the theory of grammar plays in change.
The idea of a radical change was first proposed by David Lightfoot (1974, 1979), who saw the modals as a paradigm case. His groundbreaking work has been followed by others but has also led to reactions from linguists who believed that the change was gradual rather than radical (notably Warner, 1983, 1993; Plank, 1984). Lightfoot’s story briefly is as follows. In OE and ME the core modals willan, *sculan, magan,*motan and cunnan behaved like any other verb, and there is no reason to assume that they belonged to a special category, set apart from the category Verb (see Section 3.2.4). The descendants of these modals in PDE, will, shall, may, must and can, on the other hand, are no longer verbs but must be considered to belong to a separate category, namely Aux(iliary). Thus the pre-modals (as Lightfoot terms them) could occur in positions where they now no longer occur: they could be used in both finite and non-finite position, they could be found on their own with a direct object NP or complement clause, and they could be combined with another modal. The examples in (41) from OE and ME illustrate this:
(41) a. |
as infinitive |
|
To conne deye is to haue in all tymes his herte redy (ME, Warner, 1993: |
|
199, Caxton The Arte and Crafte to knowe Well to Dye 2) |
b.two modals combined
&hwu muge we þone weig cunnen?
and how may we the way |
can |
‘And how can we know the way?’ |
(OE, Jn (Warn 30)14.5) |
c.as present participle
Se |
ðe |
bið butan |
willan |
besmiten |
oððe |
se ðe |
willende |
on |
slæpe |
||
He |
who |
is |
without |
will |
defiled |
or |
who |
willing |
in |
sleep |
|
gefyrenað, |
singe |
<XXIV> sealma. |
|
|
|
|
|
||||
fornicates, |
sing |
24 |
|
psalms |
|
|
|
|
|
||
‘Whoever is defiled against his will or who, willingly, fornicates in his sleep, let him sing twenty-four psalms’ (OE, Conf 1.1(Spindler) 46)
d.as past participle
Wee wolden han gon toward tho trees full gladly, if wee had might (ME, Visser §2042, Mandeville 196, 34)
e.with an object
He |
cwæð |
þæt |
he |
sceolde |
him |
hundteontig |
mittan |
hwætes. |
He |
said |
that |
he |
owed |
him |
(a) hundred |
bushels |
of-wheat |
|
|
|
|
|
|
(OE, ÆHom 17 26) |
||
148 O L G A F I S C H E R A N D W I M VA N D E R W U R F F
f.with a clause
Leof |
cynehlaford, |
ic |
wille, |
þæt |
þu |
beo |
æt |
minum |
gebeorscipe |
dear |
liege-lord, |
I |
will |
that |
you |
be |
at |
my |
banquet |
‘Dear lord, I would like you to be present at my banquet’ |
|
||||||||
|
|
|
|
|
|
(OE, ÆHomM 14 (Ass 8) 185) |
|||
In the course of the OE and ME periods a number of ‘unrelated’ changes took place that isolated the pre-modals from the other verbs (cf. Lightfoot, 1979: 101–9):
(42)
(i)the pre-modals lost the ability to take direct objects
(ii)the pre-modals were the only preterite-present verbs left; all others of this class were lost
(iii)the past-tense forms of the pre-modals no longer signal past-time reference
(iv)the pre-modals alone take a bare infinitive; all other verbs start taking to-infinitives
These changes are believed to be unrelated because they are accidental (especially (ii) and (iv), which concern the behaviour of verbs other than the pre-modals) and/or because they do not happen at the same time. The changes had a common effect, however, in that they resulted in the isolation of the pre-modals: they became ‘identifiable as a unique class’ (Lightfoot, 1979: 109). The evidence for this category change is to be found in the fact that the pre-modals now underwent a second phase of changes, which were related and which did take place simultaneously (Lightfoot, 1979: 110):
(43)(i) the old pre-modals could no longer appear in infinitival constructions
(ii)the old pre-modals could no longer occur as present participles
(iii)the old pre-modals could no longer occur as past participles
(iv)the old pre-modals could no longer occur in combination (except in some dialects, such as Modern Scots)
The simultaneity of these changes, according to Lightfoot, provides evidence that a deep, radical change must have taken place in the abstract system, which dissolved the verbal status of the pre-modals (i.e. they became a new category, that of Auxiliary) and thus forced the four characteristics given in (43) upon them. The simultaneity, therefore, is crucial.
There are a number of problems with this story. First of all, all the characteristics given in (43) involve losses, and such negative evidence is very difficult to date. The evidence would have been more convincing if, due to the category change to Aux, the modals began to occur in new constructions, but this is not the case. Also note that the first change under (42), the loss of direct objects after pre-modals, is really on a par with the changes under (43). It too involves a feature that would be the result of a category change from Verb to Aux. Since losses are difficult to spot in time, (42i) could as easily have been placed under the changes of (43). Indeed, examples of modals with a direct object are found quite late, i.e. after 1500, the time of the purported change. Visser (§§551, 557–8) notes examples with can until 1652, with may until 1597 and with will until 1862. Another aspect that remains hidden under the notion of ‘losses’ is the interesting fact that (43)
Syntax 149
mostly involves ME losses and not OE ones: the use of infinitive forms and of past and present participles was actually more frequent in ME than in OE. With a story of loss, one would expect the frequencies to be the other way around. The paucity of these forms in OE may be due to a lack of data, although that is unlikely (see Plank, 1984: 314), but the new forms in ME may also be due to changes taking place elsewhere: for example, the occurrence of two modals in combination is found mainly with shall and begins to occur only after shall has developed into a future-tense auxiliary (see Visser §§1685, 2134). Warner (1993: 101) also notes that the pre-modals became more verbal rather than less in ME, with shall, can and may developing full verb inflectional endings such as third-person -eþ in both the singular and the plural in southern texts, and the occurrence of certain non-finite forms that had not been attested in OE (see also below).
Another aspect that has been questioned is whether the changes in (42) are really unrelated and accidental. If we start from the assumption, as many linguists do, that already in OE the pre-modals were set apart from other verbs as a group (see Warner 1993: 152, 97ff.), then the changes under (42) can easily be seen as related. We saw above that in OE the past-tense modals could be used to express present-time modality, so in that respect they differed from ‘normal’ verbs. As to verbal complementation, (42iv), not much changes here. In OE there was only a restricted class of verbs that could take a bare infinitive. This class comprised the modals, verbs of physical perception (‘see’, ‘hear’) and causatives (OE lætan, biddan, hatan). With a few exceptions (i.e. there were some verbs that could take both bare and to-infinitives: e.g. þencan ‘think’ in OE, and in ME also helpen, maken), all other verbs took only to-infinitival complements. There is no evidence that the to-infinitive encroached on the domain of the bare infinitive in ME. It is true that the to-infinitive became much more frequent in ME, but this is due to the fact that it started replacing that-clauses, the distribution of the bare infinitive itself remaining relatively unaffected in ME (Los, 2005). As to the loss of all other preterite-present verbs, (42ii), Harris & Campbell write (1995: 179): ‘But if auxiliary variants of the modal verbs already existed, it was the entire class of preterite-present verbs that was lost, and it was no accident.’ Indeed, if the pre-modals were already looked upon as a subgroup in OE, then this very fact may have pushed the other preterite-present verbs out of the system. Harris & Campbell suggest that the modals in OE fell into two homophonous categories, one an auxiliary and the other a fully lexical verb. The OE examples given in (41) in fact illustrate this well: thus the infinitive cunnen (41b), the present participle willende (41c), sceolde with a direct object (41e), and wille with a object clause (41f) are all examples of the pre-modals used with full referential meaning, i.e. without deontic or dynamic modality. When the modals began to play a more important and frequent role in the ME period due to the loss of the subjunctive, it was the homophonous lexical pre-modals that began to die out, while the truly modal pre-modals developed further, at first still maintaining their verbal status, but gradually developing into more independent ‘word-like’ elements.
150 O L G A F I S C H E R A N D W I M VA N D E R W U R F F
It is interesting to observe that the ME increase in the infinitival and participial forms of the pre-modals may have been connected with the development of periphrastic constructions to express the future and the perfect (see Sections 3.3.1 and 3.3.2). The pre-modals (which were, after all, auxiliary-like too) were caught up in this, forming combinations with the auxiliaries of tense and aspect (just as they had combined with the auxiliaries of the passive in OE), and we see constructions such as shall may and have mought occurring. The subsequent disappearance of these infinitival and participial forms presumably has to do with the fact that they were awkward to begin with (as we have seen, they were rare to non-existent in OE), and with the fact that the homophonous lexical modals, which could have given support to these non-finite forms, had become truly separated from their sisters (indeed they all eventually became obsolescent). Another problem may well have been that the modals’ tense forms were already used in OE as modality markers, i.e. they were not strict members of the tense system. This non-tense characteristic became reinforced in ME with the loss of the subjunctive and the subsequent rise of indicative past-tense modal markers to take their place (see Section 3.3.3). In other words, the modals did not sit well in a system of tense or aspect, and this made the combination with perfect have and future shall difficult. Finally, the order of the auxiliary verbs presumably plays a role in this development as well, as suggested by Warner (1993); we will look at this in more detail in Section 3.3.7.
Because of the loss of tense distinctions in the modals, we begin to witness the rise of periphrastic modals or ‘exploratory expressions’ to indicate the sense of ability and obligation in the past. These same expressions could then easily spread to other non-finite positions, which in turn may have aided the disappearance of non-finite modal forms.
(44) a. For certes, by no force ne by no meede [bribe], | Hym thoughte, he was not
able for to speede [be successful] |
(Chaucer, Phys.T 133–4) |
b.By wey of kynde [nature], ye oghten to been able | To have pite of folk that
be in peyne. |
(Chaucer, Mars 282–3) |
c.As she was bown to goon the wey forth right | Toward the gardyn ther as
|
she had hight [promised] |
(Chaucer, Frk.T 1503–4) |
d. |
. . . that thow art bounde to shewen hym al the remenaunt of thy sinnes |
|
|
(Chaucer, Pars.T 1007) |
|
Next to to be able to and to be bound to, there are other expressions in use early (to have power/might, to be to). The emergence of today’s regular periphrastic constructions such as have (got) to and be compelled to is somewhat later. It is interesting to observe too that these periphrastic modals, like their predecessors the pre-modals, follow more or less the same path. They are first used dynamically and deontically, while epistemic use is always later. Have to, for instance, developed into a deontic modal at the beginning of the eModE period, but its epistemic use (as in It has to be true that . . .) is quite recent. Similarly, there is a tendency for
Syntax 151
these periphrastic modals to lose their non-finite forms, just as their predecessors did. Have got (now often reduced to got), for instance, is finite only, and so are had/’d rather and (had/’d) better. Instances with non-finite be to now sound distinctly archaic:
(45)a. You will be to visit me in prison with a basket of provisions (Austen, Mansfield Park I.xiv.135; Denison, 1998: 174)
b.N.B. No snuff being to be had in the village she made us some. (Keats, Letters 78 p. 189 (20 Jul.); Denison, 1998: 174)
To sum up, the evidence for assuming that there was only one homogeneous verbal category in OE, which included the pre-modals, as suggested by Lightfoot, is not all that strong. It is not the case that the pre-modals developed more and more exception features in the OE and ME period; they were exceptional within the category of verbs to start with but retained their verbal status, certainly still in ME. After the ME period they became isolated more and more, losing the trappings of full verbs in the process, but this happened slowly and not in the same way for each pre-modal. It may be that the pre-modals have become so opaque as verbs that they should be considered a different category, i.e. Aux, but the problem is that it is hard if not impossible to pinpoint when such a change could have taken place. It is clear that within the verbal class there is a continuum running from full verbs to auxiliary-like verbs, where all the different features (verbal and less- than-verbal) available are distributed unevenly across the original pre-modals, other modals and other auxiliary-like verbs such as perfect have, passive be, do etc.
In some varieties of English, there is evidence that the modals have undergone further development after reaching their auxiliary-like status. In Southern American English and Scottish English, combinations of two modals can be found, as in the examples in (28); see further Chapter 7:
(46)a. I thought you said we might could get some candy.
b.If we had known, we may still could have done it.
c.He will can do it.
These combinations are first attested in the period 1750–1850. They are therefore not likely to be direct continuations of the pattern in (41b), hwu muge we þone weig cunnen? ‘how may we the way can’, i.e. ‘how can we know the way?’, which disappeared after the ME period (and in which the second modal always had a clearly infinitival form). Rather, the double modals in (46) seem to represent an innovation, which may find its origin in a reinterpretation from modal to (epistemic) adverb. This would mean that in each example there is a sequence Adverb–Modal (or Modal–Adverb), making these sentences quite unexceptional apart from the specific form that the adverb takes. Some support for this analysis comes from interrogative and negative clauses with ‘double modals’, as in (47a, b), and an admittedly very rare type with a ‘modal’ and a form of do, as in (47c):