Hostname: page-component-6898cfcb55-w5hgh Total loading time: 0 Render date: 2026-11-11T10:00:18.084Z Has data issue: false hasContentIssue false
  • English
  • Français

Ne + infinitive constructions in Old English1

Published online by Cambridge University Press:  22 October 2012

LINDA VAN BERGEN
LINDA VAN BERGEN*
Affiliation:
Linguistics and English Language, School of Philosophy, Psychology and Language Sciences, University of Edinburgh, Dugald Stewart Building, 3 Charles Street, Edinburgh EH8 9AD, UKl.vanbergen@ed.ac.uk
Rights & Permissions [Opens in a new window]

Abstract

The occurrence of the Old English negative particle ne ‘not’ preceding a bare infinitive rather than a finite verb is a largely neglected or overlooked phenomenon. It is attested in constructions with uton ‘let's’ and in conjoined clauses with omission of the finite verb (Mitchell 1985). This article discusses evidence gathered mainly from the York–Toronto–Helsinki Parsed Corpus of Old English Prose, showing that it is a phenomenon that needs to be taken seriously in descriptions and analyses of Old English. It is argued that the factor shared by the two constructions is the lack of an available finite verb for ne to attach to. It is also found that the use of ne for the purpose of negative concord appears to be more variable with infinitives than it is with finite verbs. Whether attachment of ne to a non-finite verb in the absence of a finite one is restricted to bare infinitives is difficult to determine because of the limited evidence relating to other non-finite forms, but there are some indications that use of ne may have been possible with present participles. Finally, some implications that the ne + infinitive pattern has for the formal analysis of Old English are discussed.

Information

Type
Research Article
Information
English Language & Linguistics , Volume 16 , Issue 3 , November 2012, pp. 487 - 518
Copyright
Copyright © Cambridge University Press 2012

1 Introduction

A common generalisation about the placement of the negative particle ne ‘not’ in Old English (OE) is that it always occurs immediately before the finite verb (e.g. van Kemenade Reference Kemenade, Tieken-Boon van Ostade, Tottie and van der Wurff1999: 152; Fischer et al. Reference Fischer, van Kemenade, Koopman and van der Wurff2000: 54, 140; Ingham Reference Ingham2007: 390). It has mostly been taken for granted that this generalisation is correct and admittedly the data would appear to give good grounds for accepting it. The position of ne depends on that of the finite verb; finite verb placement varies in OE, but regardless of where in the clause the finite verb occurs, ne immediately precedes it, as illustrated in (1).

  1. (1)

Footnote 2 It is also true that instances where the negative particle is not adjacent to a finite verb are extremely rare in OE. However, Mitchell (Reference Mitchell1985) mentions two contexts where ne may occur before an infinitive rather than a finite verb. If this is the case, the commonly accepted generalisation about the placement of ne would not be completely accurate, which would have consequences for both descriptions and analyses of OE. It is important, then, to establish whether ne can indeed attach to infinitives in specific contexts, and if so, to what extent and under which conditions. This article focuses on answering those questions.

To begin, let us briefly review what is stated about these cases in the literature. The first context where ne + infinitive occurs according to Mitchell involves uton, which combines with bare infinitives to form a first-person plural adhortative construction comparable to present-day English let's. An example, with ne before the infinitive, is given in (2).

  1. (2)

The second environment identified by Mitchell concerns coordination, specifically when the finite verb is not repeated in the conjunct with the relevant infinitive, as illustrated in (3).

  1. (3)

The two contexts where the construction is said to occur are likely to be rare and very low frequency is probably part of the reason why the phenomenon has largely been ignored or overlooked so far.Footnote 3 Moreover, those who are aware of the existence of examples like (2) and (3) might with some degree of justification have written them off as anomalous. In the case of uton, Mitchell (Reference Mitchell1985: §916a) gives three examples with ne before the infinitive, but he draws attention to ‘the scarcity of examples with an intervening negative adverb’, which he thinks might suggest ‘that the use of uton to express prohibitions was not a natural one’. He raises the possibility of some degree of Latin influence – the instance in (2), for example, corresponds to Latin Non scindamus eam, with the negative adverb non before the (subjunctive) main verb – although he is quick to point out that they cannot be regarded as mechanical glosses.Footnote 4 Commenting on an example with na instead of ne, given in (4), he states that here ‘we have the more natural na’ (Mitchell Reference Mitchell1985: §916a), referring to a paragraph where he says that ‘In the prose, words (other than finite verbs) and phrases are regularly negated by na/no’ (Mitchell Reference Mitchell1985: §1614).

  1. (4)

Mitchell, then, clearly regards na rather than ne as the expected form of negation in this construction. He further suggests that analogy with ne + infinitive found in the coordination environment may have been involved in the appearance of uton ne + infinitive.

As for the coordination context, Mitchell states that ‘ne or na precedes the (first) negated infinitive’ (Reference Mitchell1985: §1602). However, he gives just one example with ne: the one from Beowulf given in (3) above. He adds a reference to two further examples, but both involve na and neither is directly comparable to the instance with ne.Footnote 5 Mitchell then points to Einenkel (Reference Einenkel1912) for more information.

Einenkel clearly regards the use of ne (rather than na/no) with the infinitive as both rare and erroneous (Einenkel Reference Einenkel1912: 208). He describes the Beowulf example as generally marked by the awkwardness of its phrasing. In any case, of course, examples from poetry should be treated with caution if there is no good evidence from prose to support the grammaticality of the construction, given that influence from metre is difficult to rule out. Einenkel does provide two examples from prose texts, given in (5) below, but they are problematic because they involve manuscript variation; not all extant witnesses have ne + infinitive.Footnote 6 Einenkel implies that we are dealing with scribal error in both cases.

  1. (5)

Scribal corruption is certainly a possibility for (5a); unintentional repetition (dittography) is a plausible error, and the other manuscript has ne scyran ‘nor decree’ instead of ne ne scrifan. For (5b), two of the five manuscripts omit ne, which could indicate that the construction was not natural for OE and that it was altered by one or more subsequent scribes for that reason. In addition, Latin influence is a possible factor in (5b), given the parallel non ‘not’ + infinitive found in the source text.Footnote 7 Of course, the translator could have chosen to use na as a translation for non if ne was ungrammatical here, so it is not clear that Latin influence can easily be held fully responsible for the use of ne in (5b). On the other hand, it is also difficult to rule out the possibility of Latin influence in cases of this type; a translator might occasionally use the most common translation for a particular word in a context where it is not appropriate, even if that seems less likely in a reasonably free translation than in a word-by-word gloss or very close translation.

Unlike Einenkel, Mitchell does not appear to think ne is erroneous in the coordination context. However, the information he provides still seems to imply that na/no is the normal form. Consequently, the evidence and discussion found in Mitchell (Reference Mitchell1985) and Einenkel (Reference Einenkel1912) might well lead a reader to conclude that, at least as far as prose is concerned, there is little that would justify a revision of the rule on the placement of OE ne.

In addition, it might be possible to explain away a few apparent examples of ne preceding something other than a finite verb by treating it as a rare scribal variant of na. The spelling <ne> can very occasionally be found in cases that clearly involve constituent negation and/or where there is another instance of ne immediately before the finite verb, as illustrated in (6).

  1. (6)

It must be stressed that this is rare. Even so, it does happen, so it might be possible to argue that an occasional example of <ne> before an infinitive is actually a spelling variant of na. However, such an interpretation would only be plausible if <na> or <no> is indeed found in the vast majority of instances.

The question, then, is whether the apparent occurrence of ne before an infinitive following uton or in the coordination context needs to be taken seriously. And if so, how common is it? Is na the norm and ne a possible but less frequent alternative, or is ne the rule rather than the exception? This study seeks to answer those questions on the basis of a collection of data on both constructions from the York–Toronto–Helsinki Parsed Corpus of Old English Prose (YCOE; Taylor et al. Reference Taylor, Warner, Pintzuk and Beths2003).Footnote 8 In the case of uton, the unparsed Dictionary of Old English Corpus (DOE Corpus; Cameron et al. Reference Cameron, Crandell Amos, Butler and DiPaolo Healey1981) has also been used. Sections 2 and 3 discuss the data for uton and the coordination context, respectively. On the basis of those data, I will show that the occurrence of ne before bare infinitives is indeed a phenomenon that needs be taken seriously in descriptions and analyses of OE, particularly in the case of the coordination context. Section 4 focuses on identifying the precise conditions under which ne may combine with a non-finite verb. I will suggest that the crucial property linking the uton and coordination contexts is the lack of an available finite verb. I will also explore whether there are any further constraints on attachment of ne to bare infinitives, especially in relation to contexts containing other negative elements, and I will discuss the very limited evidence relating to non-finite forms other than bare infinitives. While the primary concern of this article is with the more descriptive aspects of the ne + infinitive phenomenon, implications for the formal analysis of OE are touched on briefly in section 5. This is followed by the conclusion in section 6.

2 Data on ne + infinitive after uton

As seen in the previous section, the information given in Mitchell (Reference Mitchell1985) would lead us to believe that ne is a possible but less common alternative to na in the construction with uton plus negated infinitive. This section discusses the findings from the OE corpora to see whether that is borne out by the data.

Just five examples of uton ne + infinitive were found in the YCOE, all given in (7) below. No additional examples were found in the DOE Corpus.

  1. (7)

The number of instances with ne + infinitive in the uton context is very small, then. However, a search for instances with na/no found no examples at all in the YCOE, and only one in the DOE Corpus: the example already cited from Mitchell in (4) above.Footnote 9 So while, with Mitchell (Reference Mitchell1985: §916a), we might have expected na to be the more obvious choice in this construction, that is not in fact supported by the data.

As Mitchell acknowledges, the construction with ne cannot be treated as a straightforward calque (notwithstanding his suggestion of a possible degree of Latin influence): none of the examples in (7) with a known Latin source involve a negated infinitive in the Latin. Two, (7b) and (7c), do correspond to negated subjunctive verbs in the Latin, so that ‘the sense . . . of the negated Latin verb is conveyed by the OE infinitive’ (Mitchell Reference Mitchell1985: §916a), but they could have been translated with negated subjunctives. The translation in (7a) appears to be quite free,Footnote 10 and (7d) may be an original addition (Scragg Reference Scragg1992: 364). Since Latin influence could at best provide a partial explanation for only a subset of the examples, it does not look like it really helps to account for the use of ne + infinitive with uton.

The very low numbers do mean that caution is required. Moreover, it will be clear from (7) that all but one of the examples with ne are found in the Vercelli Homilies. In addition, those four examples are all from homilies for which the Vercelli manuscript is the only surviving witness. The sole example from a different text, (7a), is from Soliloquies, which only survives in a twelfth-century manuscript and might therefore not be representative of OE. In addition, the spelling <ne> for na/no may be becoming more frequent by that time; it is attested elsewhere in Soliloquies, so a single example of pre-infinitival ne could potentially be ascribed to that factor.Footnote 11 In short, just one manuscript (the Vercelli Book) gives apparently solid evidence for the existence of uton + ne + infinitive.

To complicate matters somewhat further, the YCOE data might suggest that the frequency of the spelling <ne> in the Vercelli Book for what we would expect to be <na> or <no> (in contexts other than with uton) is higher than usual. In the YCOE as a whole, the frequency of <ne> spellings for na is no more than 2 per cent,Footnote 12 but for the Vercelli Homilies specifically it could have been as high as 12.5 per cent, according to the YCOE: four instances of <ne> are parsed as the negative adverb na in that text file, against 28 instances of <na> or <no>. However, this would still not be high enough to justify explaining the four <ne> forms before the infinitive in uton constructions as spelling variants of na, especially given that the form <na>/<no> is not found at all with uton in the Vercelli Homilies. Moreover, at least three of the four apparent examples of <ne> for na have an alternative interpretation and/or are problematic for other reasons, so there is no good basis for believing that this manuscript really has an unusually high frequency of <ne> spellings of na.Footnote 13,Footnote 14

To sum up, then, there are no good grounds for treating the examples with uton as anything other than genuine cases of ne + infinitive, at least in the case of those from Vercelli. The existence of just one example with na in the DOE Corpus also suggests that <ne> really is ne in this context. Even so, the data on uton do not allow very confident or sweeping conclusions. Specifically, we cannot conclude on the basis of these examples that the construction was generally available in OE. Mitchell appears to be right in his assessment that uton was not normally used to express prohibitions. What little evidence there is for the construction is largely confined to a single manuscript. So the data suggest that ne + infinitive may have been normal with uton to the extent that uton was used in combination with negation, and it almost certainly was normal in at least one variety of OE, but they are insufficient for anything stronger than that.

3 Data on ne + infinitive in the coordination context

In section 1 we saw that, on the basis of the information found in the available literature, the evidence for ne + infinitive in the coordination context appeared to be weak. The only relatively unproblematic example cited so far, i.e. (3), was from verse rather than prose. However, the data discussed in this section will show that the evidence for this construction is actually much stronger than the information in Mitchell (Reference Mitchell1985) or Einenkel (Reference Einenkel1912) suggests.Footnote 15

There are 19 instances of ne + infinitive in the coordination context in the YCOE (plus three duplicates, which are excluded from the counts). Clearly we are dealing with an infrequent construction, as expected, but there are in fact significantly more examples of ne + infinitive in the coordination context than was the case for uton. Five of these conjuncts are introduced by the negative coordinator ne ‘nor’; they will be discussed separately for reasons that will become clear. I will first focus on the 14 instances where the coordinator is not negative (normally the conjunction and, but omitted in one instance involving a sequence of conjoined clauses). A few examples are given in (8). The instances not given in (8) or elsewhere in this section can be found in the Appendix.

  1. (8)

A slight complication arises with the data, in that three of the examples are also dependent on uton – not present in the conjunct concerned itself – as exemplified in (9).

  1. (9)

The possibility of influence from uton means that it is not entirely clear that such examples are necessarily evidence for the construction with coordination. Nevertheless, I think it is probably safe to include them here (certainly safer than assuming that uton is the crucial factor, with firm evidence for that construction being all but limited to the Vercelli Book).Footnote 16 In any case, the evidence for the existence of the construction in the coordination context would remain strong even if these three were excluded, although admittedly it would be even more highly concentrated on the language of one particular author than it already is.

While the examples come from a wider range of texts than was the case for uton, no fewer than seven are found in the works of Ælfric, and one further instance comes from a section of the translation of Numbers that has also been attributed to Ælfric.Footnote 17 One of these examples is similar to (9), i.e. it involves dependency on uton, but the remainder are straightforward cases. Several are from Ælfric's Catholic Homilies, for which we have very good contemporary evidence as well as multiple additional copies confirming the form <ne>, e.g. (8a).Footnote 18

The number of examples alone is enough to suggest that this construction was genuine, at least in the case of Ælfric's language. Moreover, the spelling <ne> for na in the works of Ælfric is vanishingly rare. There are just two instances of <ne> parsed as a negative adverb in the relevant YCOE files against more than 600 instances of <na>/<no>. The first case was already given in (6a) above. The two extant witnesses both have <ne>, but ne furðon frequently means ‘nor even’ (i.e. with the coordinator ne), so the spelling <ne> would have been comparatively easy to escape notice to a copying scribe in this context. In the second case, ne læs ‘no less’ (ÆCHom I, 30, 435.182), only the base manuscript has <ne>, while the other manuscripts have na or na ðe. This is in sharp contrast with what we find for the examples of ne with infinitives found in Ælfric's writings, where there is very little manuscript variation involved. In most cases, all surviving copies agree on ne. Even in the two cases where there is some disagreement between witnesses, the form <na> is found in just one of these two instances and it involves a single twelfth-century manuscript at that; the other case involves omission of the negation, which looks like a scribal error (examples (A1) and (A3), respectively, in the Appendix). Thus, the examples of ne + infinitive from Ælfric can definitely not be treated as occasional <ne> spellings for na. The near-absence of manuscript variation in addition suggests that the copying scribes were generally at least willing to accept the construction.

The evidence for the grammaticality of the construction in Ælfric's prose can be strengthened still further. The YCOE contains an additional instance of ne + infinitive from Ælfric found in a very similar construction, given in (10). The only difference is that the infinitival clause is appositional rather than coordinated – it explains the meaning of the preceding infinitive don ‘do’.Footnote 19

  1. (10)

Aside from the potential problem of classification with examples that involve uton as well as coordination, most examples in this context look straightforward and convincing. I have also seen little that could be attributed to possible influence from a Latin original. One example that does look unusual, though, is (11a). The use of the coordinator and is unexpected here, given that the coordinated clauses are both negative. By contrast, the reading in the Peterborough Chronicle in (11b) has the expected coordinator ne ‘nor’ instead and omits the negative particle. The possibility of error in the case of (11a) cannot be excluded. But on the strength of the other examples, the use of ne + infinitive in this context is plausible in principle, and the fact that it occurs in two versions of the Anglo-Saxon Chronicle suggests it was probably not an error.

  1. (11)

Footnote 20

There is also one example, (12), which may look peculiar at first, but it should probably be accepted and might even give us another indication that ne rather than na really was acceptable in this construction: ne might look out of place – one manuscript in fact omits it and negation is also absent from the corresponding Latin – but pleonastic negation is attested in complements of verbs like tweogan ‘doubt’ when negated (see e.g. Mitchell Reference Mitchell1985: §2039), so this is the probable explanation for the use of ne here. If so, this particular instance of <ne> is unlikely to be a spelling variant of na, since use of a negator other than ne would be not be expected in this context.Footnote 21

  1. (12)

So far we have been concentrating on examples where the coordinator was either and or unexpressed, and we have seen that the evidence for ne + infinitive in this context looks robust. The instances which involve coordination with ne ‘nor’, on the other hand, are fewer in number and look much less convincing. Two were already given in (5) above, and as said there, both involve manuscript variation. In one of the remaining three examples, (13a), ne ecan is coordinated with na wanian, strongly suggesting that it is a <ne> spelling of na. The same interpretation seems not impossible for the example in (13b), given that it is also from the Blickling Homilies. However, <ne> spellings for the negative adverb na/no are not common in these homilies – there is just one probable further case against 38 instances of na/no – so this is by no means certain. The final example, (13c) from Bede, does not survive in the most authoritative witness, and at least one manuscript does not have the construction. We may question, then, whether ne + infinitive occurred in ‘nor’ conjuncts, since some doubt can be cast on at least four out of five instances found in the YCOE.

  1. (13)

Footnote 22

In the case of conjuncts into which negation is first introduced, however, we have already seen that there are no grounds for doubting the grammaticality of the ne + infinitive construction, especially for Ælfric's variety of OE. And when we look at the frequency of na/no in this particular environment, that conclusion becomes even stronger. Among the YCOE data, only the two examples in (14) are comparatively convincing cases of na as clausal negator in the coordination context when the coordinator is either and or not overtly stated. (15) gives a potential additional example, but here constituent negation seems plausible too.Footnote 23 So we find two or three instances of na in the context where we found 14 instances of ne in the YCOE. It also does not appear to be the case that alternative negative elements are normally used instead in this context to avoid both ne and na, certainly in the case of Ælfric; the number of conjuncts of this type with other negative words/phrases is limited (see section 4.3 for data).

  1. (14)

  2. (15)

In ‘nor’ conjuncts, on the other hand, we find four examples with na, which all look straightforward; they are given in (16).Footnote 24 Given that there were just five instances with ne in this type of conjunct, most of which were doubtful or problematic, it looks as if na was probably the more normal form in this context, to the extent that the very limited data allow any conclusions.

  1. (16)

Alternatively, it might be not so much the context as the set of texts that is responsible for the difference. We have no relevant data from Ælfric on ‘nor’ conjuncts (regardless of whether the negator is ne or na), whereas three of the examples in (16) come from translations by Alfred, who is not represented at all in the data set on ne + infinitive in the coordination context. In addition, all examples in (16) involve early rather than late OE texts, even if not all survive in manuscripts from that period. By contrast, most examples with ne + infinitive are found in late West Saxon material. The only three from early OE texts – (12), plus the problematic (5a) and (13c) in ‘nor’ conjuncts – are from Bede and Gregory's Dialogues, which derive from an Anglian source, so straightforwardly early West Saxon examples are not found at all. The use of na rather than ne in the examples in (16), then, could reflect diachronic factors and/or differences between dialects. As for the mostly problematic examples of ne + infinitive in ‘nor’ conjuncts, only (5b) from the Benedictine Rule is clearly (late) West Saxon; the other four all derive from Anglian sources. The near-complete lack of evidence for the use of either ne or na with infinitives in ‘nor’ conjuncts in straightforwardly late West Saxon material could indicate that late West Saxon simply avoided using either of these negators here.

To sum up, the data leave no doubt that in the cases where the coordinator is not ‘nor’, i.e. when negation is normally first introduced in the conjunct, the norm is ne rather than na. This certainly holds for Ælfric's writing and may well have been true more generally, at least for late West Saxon. We cannot rule out the possibility that na was more common or even the norm in early texts, but since the data concerned are found in ‘nor’ conjuncts, it is equally possible that the syntactic context is responsible for the observed difference; late West Saxon seems to have avoided using either ne or na in that particular context.

4 Identifying the conditions for attachment of ne to a non-finite verb form

The data given in the previous two sections have shown that the phenomenon of ne preceding bare infinitives cannot be ignored. Under specific conditions it is clearly a grammatical option. This holds especially for the coordination context, where it is the normal option, at least in the case of Ælfric and quite possibly late West Saxon in general. The evidence for uton was more limited, but we have seen that ne + infinitive can also occur in that environment. The next question is precisely what it is about these two environments that allows ne to combine with a non-finite form.

Wallage (Reference Wallage2005: 78) suggests in relation to the construction with uton that ne occurs before the infinitive for scope reasons, i.e. that it is a biclausal structure and the higher verb (uton) is outside the scope of negation. While this account could be extended to cover the straightforward instances in the coordination context (though not all of the potentially more problematic ones, e.g. (11a), (12)), it seems an unsatisfactory approach. Even ignoring the question whether uton constructions really are biclausal, constructions with (pre-)modals are frequently treated as such (e.g. Roberts & Roussou (Reference Roberts and Roussou2003: 39), whose analysis in this respect Wallage adopts). Yet when the finite modal is not included in the scope of negation, ne is still found before the finite verb rather than the infinitive. This can, for example, be seen when sceal ‘must’ is used in commands not to do something (e.g. ÆCHom I, 19, 326.46 Ne sceal se rica . . . þone earman forseon ‘The rich man must not despise the poor man’). Given that ascribing the use of ne with infinitives to scope makes the wrong predictions for a far bigger set of data than it could potentially explain, we need to look in a different direction.

In this section, I will suggest that the property enabling ne to combine with infinitives is the absence of an available finite verb. I will argue in section 4.1 that this property is shared by both contexts where ne + infinitive is found by showing that uton cannot straightforwardly be analysed as a finite verb. In section 4.2, I will discuss one potential counterexample where ne appears to attach to an infinitive in spite of the presence of a finite verb. Section 4.3 explores whether the lack of a finite verb is a sufficient condition for ne to attach to infinitives, or whether the option is constrained further in the context of other negative expressions. Finally, section 4.4 discusses the limited data available on other non-finite forms, to see whether there is any evidence enabling us to determine whether or not attachment of ne in the absence of a finite verb is specific to bare infinitives.

4.1 The common factor: absence of a finite verb?

One property of the coordination context that might help to explain the ne + infinitive phenomenon is the absence of a finite verb for ne to attach to in the conjunct. If that is the relevant factor, though, we would expect ne + infinitive not to occur when a finite verb is clearly available. The fact that it is also found with uton, then, might at first sight seem to argue against this approach. However, the status of uton is by no means clear-cut. Mitchell (Reference Mitchell1985: §916a) mentions that it has been labelled an interjection, even if he himself still regards it as a verb. In this section I will discuss peculiarities in the behaviour of uton that are hard to explain if it were indeed a finite verb, indicating that unavailability of a finite verb is arguably a characteristic that the two environments share.

A well-known property that makes uton peculiar is its isolation. Historically it derives from the verb witan ‘depart’, but any connection with this verb has been thoroughly obscured by the time of OE records. Indeed, according to Ogura (Reference Ogura2000) witan is no longer found as a verb of motion at all in OE – the prefixed verb gewitan is used instead. So if uton were a verb, it would have just a single inflectional form, found exclusively in a specific type of adhortative construction.

Furthermore, there are indications that uton should not be regarded as inflected from a synchronic perspective. If -on were an inflectional ending, we would expect uton to sometimes occur in the form <ute> when it is followed by the subject pronoun we, since the first- and second-person plural verb endings (regardless of tense or mood) frequently reduce to -e in that context. Yet this type of reduction does not happen at all in the YCOE with uton we.Footnote 25 The non-occurrence of ute we could be accidental in the OE Bede, the Blickling Homilies and the Vercelli Homilies, since they are of Anglian origin and this reduction of the plural ending is less frequent in Anglian than in West Saxon (see Campbell Reference Campbell1959: §730); most finite verb forms in these texts retain a full ending before we. For Ælfric, however, reduced endings are very clearly the norm when a finite verb is followed by we: 110 out of 113 instances in the Catholic Homilies, 44 out of 47 in Lives of Saints and 24 out of 29 in the Supplementary Homilies.Footnote 26 Yet none of the 10 forms of uton followed by we in these texts change the ending to -e.Footnote 27 In Wulfstan's Homilies, reduced endings predominate too in this context – 16 out of 19 – whereas neither of the two instances of uton we has such an ending. So the ending remains the same regardless of whether it is followed by a subject pronoun or not, indicating that it is not behaving like the inflectional ending of a plural finite verb form, but rather as a part of a fixed, uninflected lexical unit.

It is of course unfortunate that this particular argument against treating uton as a finite verb form is strongest for West Saxon, whereas the instances of ne + infinitive with uton are nearly all found in the Vercelli Homilies, most or all of which go back to an Anglian original (see Scragg Reference Scragg1973; Wenisch Reference Wenisch1979). Soliloquies, the only other text with that construction, simply does not have any instances of uton followed by a subject pronoun, nor do the other translations attributed to Alfred. So we cannot prove that the same type of fossilisation of form is found specifically in those varieties of OE in which uton ne + infinitive is attested. However, it still adds to the overall impression that the behaviour of uton is not characteristic of a finite verb form.

Moreover, for the Vercelli Homilies there is a different aspect of the form of uton that does not quite fit ‘normal’ behaviour of a finite verb. In this text, it is rare for a plural indicative verb form to have the ending <-an> instead of <-on>: just 12 out of more than 600 plural past tense indicatives (i.e. 2 per cent) of the YCOE file ‘coverhom’ have this ending, and the proportion of <-an> forms is slightly lower still for plural present tense indicative forms normally ending in <-on> (sindon ‘are’ and preterite-present verbs). Yet the form <utan> is common for uton in this file: 36 out of 96 instances (38 per cent).Footnote 28 If we exclude the three most West-Saxonised of these homilies, i.e. homilies 19–21 (Scragg Reference Scragg1973: 203–5), that goes up to 30 out of 49 (61 per cent). Such a high proportion of <utan> spellings is difficult to account for if uton is to be treated as a finite plural verb form.Footnote 29

From a syntactic perspective too, the behaviour of uton would at the very least be peculiar if it were a verb. Notably, it is extremely restricted in its distribution. Ignoring vocatives, interjections and coordinating conjunctions, it consistently occurs clause-initially in main clauses in the YCOE data, except that it may occasionally be preceded by a subclause, left-dislocated constituent, or a constituent such as an adverb that could be treated as being in a ‘topic’ position. It is never found clause-finally, and the subject pronoun (if present) always immediately follows it.Footnote 30 To state things in terms of analyses employing a CP/IP structure, there is no evidence that uton ever occurs anywhere other than in C. This opens up the possibility that it might be base-generated in C rather than having moved there, the more so since we have seen there is reason to believe that it is not an inflected form. If so, it would be difficult to regard it as a verb.

If uton is indeed no longer a verb form, that would mean that OE would have one non-finite construction that licenses a nominative subject, whereas nominative subjects are otherwise restricted to finite contexts in OE. This could be regarded as an argument against reanalysis of uton, but constructions may have properties reflecting their origin. Moreover, it would not be an unprecedented situation. In (formal varieties of) present-day English, nominative subjects are also permitted in one specific non-finite construction, i.e. ‘gerund-participials functioning as a supplement to a clause’ as in They appointed Max, he being the only one who spoke Greek (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 1191–2). It is not a short-lived phenomenon either: the occurrence of nominative subjects in this type of absolute construction goes back at least to Middle English (see Visser Reference Visser1966: §§1078–82). Given that it is possible for a language to tolerate such a situation for an extended period of time, the continued use of nominative subjects does not block the possibility that uton may have been reanalysed.

To sum up, there are indications that uton should probably not be analysed as a verb form synchronically. The ending may superficially look like a finite plural ending, but unlike such an inflectional ending, it does not reduce when followed by the subject pronoun we. It also shows spelling variation in the Vercelli Homilies that is again uncharacteristic of the relevant inflectional ending. In addition, there is no evidence that it ever occurs anywhere other than in one fixed syntactic position in main clauses, making it unlike any other OE verb or verb form, and it is only used in one specific type of adhortative construction. So uton has become fixed in both form and syntax.Footnote 31

If uton is not a finite verb, or at the very least not one that functions normally, the two constructions where ne + infinitive occurs would both lack a (suitable) finite verb for ne to attach to. In that case, the generalisation about the behaviour of ne in OE could essentially be left intact, with the added proviso that ne can attach to bare infinitives in the absence of an available finite verb.

4.2 A counterexample?

There is one example of ne + infinitive in the YCOE which would seem to be a counterexample to the generalisation that ne only attaches to bare infinitives if no finite verb is available.Footnote 32,Footnote 33 It is given in (17).

  1. (17)

If this truly is a counterexample, that would obviously be problematic. It does not look like scribal error, even though the version in the Cotton transcript has ne before sceoldon instead; ne occurs before underfon both in the list of headings and in the corresponding heading in the main text. However, I suspect it involves a different interpretation of the text by the scribe. There is an erasure before sceoldon and ne is inserted above the line in CPHead 9.2, so the scribe appears to have deliberately moved ne away from the position before the modal. There is also a punctuation mark after sceoldon in the manuscript, suggesting that there might be a syntactic break at that point. If underfon is not dependent on sceoldon, it could be a finite form, making the sentence the equivalent of ‘Nor again should the learned who do not want to live as they learned in books that they should (i.e. live), receive the honour of the function of teacher’, i.e. underfon would be a subjunctive form in the main clause and ðæt hie sceoldon part of the relative clause (with postverbal ellipsis after a form of *sculan, which is attested in OE; see Warner Reference Warner1993: 116). In that case, the placement of ne is exactly where it should be, i.e. before the finite verb. So this apparent counterexample may well not be a counterexample at all.

4.3 Further restrictions on the use of ne with infinitives in the context of other negative elements?

So far I have suggested that the crucial shared characteristic of the two environments where ne + infinitive occurs is the absence of a finite verb that ne can attach to. But does ne in other respects follow the normal distributional pattern in this environment or are there additional restrictions on its use with infinitives? We saw in section 3 that there is relatively little evidence for ne + infinitive in ‘nor’ conjuncts – examples were rare and usually problematic. I will show that this is not just a data gap, so why is attachment of ne to infinitives apparently avoided in this context? Could it be the result of a general avoidance of ne + infinitive in contexts where negation is already marked in some other way? In that case, we would expect that, unlike finite verbs, infinitives do not combine with ne for the purposes of negative concord. So can ne + infinitive be used in negative concord patterns, and if so, to what extent? This section will address those issues.Footnote 34

First, then, let us look more closely at ‘nor’ clauses. I will concentrate on the language of Ælfric, which is where most instances of ne + infinitive in conjunct clauses were found. In the works of Ælfric included in the YCOE, there are over a hundred ‘nor’ clauses with a bare infinitive and omission of the finite verb, but none have the negative particle before the infinitive. This strongly indicates that ne + infinitive is not an available option in that context. Compare ‘nor’ clauses that have a finite verb: 267 out of 268 instances in the works of Ælfric have ne before the finite verb.Footnote 35 So whereas the use of the negative particle in ‘nor’ clauses is very close to 100 per cent in Ælfric's writings when they contain a finite verb, it is avoided when the finite verb is absent and only a bare infinitive is available.

This might suggest that we could be dealing with some kind of ‘last resort’ strategy where ne only attaches to infinitives to mark clausal negation if (a) no finite verb is available and (b) negation is not already signalled in some other way. In ‘nor’ conjuncts, omission of the negative particle would be an alternative, since the context already clearly marks the negation. However, negative phrases are actually found in some of the conjunct clauses with ne + infinitive. A particularly clear instance of this is (18) – given with context as (A5) in the Appendix – but example (9) above and (A1), (A7) and (A9) in the Appendix also contain negative elements in addition to ne. So a strict ‘last resort’ account cannot be right.

  1. (18)

A different explanation is needed for the systematic avoidance of ne + infinitive in ‘nor’ conjuncts by Ælfric, then. In all likelihood it reflects the bond between ne and the finite verb; it is a characteristic property of clitics and affixes that they must be deleted under identity if the host/stem is (Zwicky Reference Zwicky1985: 288). So just as ne is repeated when the finite verb is not omitted in a conjunct clause, it is understood together with the finite verb when a (negated) finite verb has been omitted under identity with that in the previous clause. Of course, that interpretation would rely on discarding the small number of potential instances of ne + infinitive in non-Ælfrician material that do involve conjunction with a clause containing a negated finite verb.

This does not necessarily mean that a last resort approach is completely on the wrong track, however. Even if we treat clauses with omission of a negated finite verb as a separate case, it is not clear that negative concord functions completely normally with infinitives. While we have seen that it can occur, there are comparable cases where ne is left out. Taking Ælfric as a test case again, we find three instances where ne is not used (two clear ones, and one potentially ambiguous case). One of them is given in (19).Footnote 36

  1. (19)

Two or three examples might not seem much, but we need to compare this to just three instances with negative concord in this context in work attributed to Ælfric ((A1), (A5) and (A9) in the Appendix). In addition, there are three examples with na from Ælfric (whether or not as constituent negation), given in section 3: (14a), (15) and the instance mentioned in fn. 23. Since ne can definitely be used with infinitives, we might have expected negative concord here too, in which case the number of instances without negative concord in Ælfric would go up to five or six.

The admittedly limited data suggest, then, that lack of negative concord was probably not unusual with infinitives in Ælfric's language. That contrasts with the near-obligatoriness of negative concord with finite verbs: according to Wallage (Reference Wallage2005: 239), the use of ne in clauses with negative phrases ranges from 95 per cent in main clauses to 99 per cent in subclauses in Ælfric. So it looks as if there may be some reluctance to combine ne with infinitives when not absolutely necessary. On the other hand, Ælfric does not appear to use alternative negative phrases such as næfre (or indeed na) as a strategy to avoid the ne + infinitive construction. If we take into account all conjuncts of the relevant type containing any negative word or phrase (including ne on its own), ne + infinitive is still used in a majority of instances in Ælfric's work (8 out of at most 14). And as said, he can combine the construction with other negative phrases. So if there was any such reluctance, it clearly was not an overriding factor.

It is difficult to say to what extent we can generalise from Ælfric to OE more widely because of the limitations of the data. Something similar may hold for Wulfstan, whose work provides one example with negative concord, (9), and one without,Footnote 37 but that is too little evidence to make anything much of. The Benedictine Rule, on the other hand, may avoid negative concord in this context – there are three relevant instances if we include (14b) with no Footnote 38 – but that is again very little, plus we have no unproblematic evidence to show that the ne + infinitive construction was available in this variety. And the absence of ne in two infinitival conjuncts with a negative phrase in Bede does not tell us anything, because the rate of negative concord is low in this text even with finite verbs (Wallage Reference Wallage2005: 238; Ingham Reference Ingham2006).Footnote 39

To sum up, ne + infinitive is avoided in ‘nor’ conjuncts, probably because ne is omitted together with the finite verb under identity with the negated verb in the preceding clause. In other conjuncts, ne can be combined with bare infinitives in a negative concord pattern, but on the basis of the evidence from Ælfric, negative concord looks more variable here than it is in clauses containing a finite verb. This could indicate some reluctance to use ne with infinitives when the context is already clearly negative, but if so, it is not strong enough to lead to consistent avoidance of the construction in such a context.

4.4 Other non-finite forms

If ne can be used with bare infinitives when the clause concerned does not contain a finite verb, an obvious question is whether this phenomenon is restricted to bare infinitives (with na being consistently found with other non-finite forms), or whether ne can also attach to other non-finite verb forms under comparable circumstances. This section looks at the available evidence. We will see, though, that it is unfortunately too limited to settle the issue with any degree of confidence.

Inflected infinitives are not attested with ne in the YCOE, but there are just four examples with na, and only one of those is a reasonably clear case of na in a context where it looks equivalent to ne.Footnote 40 This example is given in (20).Footnote 41

  1. (20)

It is of course difficult to draw any conclusions on the basis of one example, but it is worth noting that it occurs in precisely the type of coordination context where ne + infinitive is found and that it comes from Ælfric, whose preference for ne over na is clear for bare infinitives. It also forms a near-minimal pair with example (A8) in the Appendix (not from Ælfric). (A8) translates the same Latin sentence as (20), but a bare infinitive (forlætan) is used in the second conjunct rather than another inflected infinitive (a common phenomenon in OE; see Mitchell Reference Mitchell1985: §§924, 925), and this bare infinitive is preceded by ne rather than na. The example in (20), then, suggests that na is likely to have been the more usual form of negation with inflected infinitives, but however suggestive, one instance certainly cannot prove that ne would have been ungrammatical.

There is also no good evidence that ne can combine with past participles. The only apparent instance is more plausibly interpreted as a <ne> spelling of na, given that the finite verb (which is present) is already negated.Footnote 42 On the other hand, it is difficult to judge precisely where this construction would have been expected to surface if available. There are 11 instances of na and a past participle in constituents coded as participial phrases in the YCOE, plus four in clauses coded as incomplete, but often the participle could be regarded as an adjective rather than a verb form. If we exclude cases where na is the only other constituent in the phrase (as being less clearly clause-like) and instances where na forms part of a contrastive/emphatic construction so that ne would probably not have been an alternative,Footnote 43 we are left with just two examples.

The first of these involves an absolute construction mirroring the Latin.Footnote 44 Absolute constructions are mostly found in close translations of Latin and it has been claimed that they are not native to OE (see e.g. Mitchell Reference Mitchell1985: §§3814, 3825–31 for discussion). In Old High German, an apparently non-native negation pattern is attested in absolute constructions translated from Latin (Jäger Reference Jäger2008: 71; see fn. 4 above). The possibility of foreign influence must be taken seriously, then. On the other hand, Latin influence might have been expected to encourage use of ne rather than na, so the choice of na here is still of interest.

The second remaining instance of na and past participle, (21), involves a coordinated construction parallel to the one in which Ælfric normally uses ne rather than na with bare infinitives. However, it is coordinated with a phrase headed by fæste, which is more likely to be an adjective than a participle, again highlighting the problem of deciding whether we are dealing with a non-finite verb or an adjective in the case of participles. Arguably, the construction is also strongly contrastive.

  1. (21)

For past participles, then, the data suggest that they may at least have preferred na, but there is little evidence to go on, and even less that is unproblematic and unambiguous.

The situation looks somewhat different for present participles. There are five instances where ne ‘not’ is found immediately preceding a present participle in constructions parsed as a participial phrase in the YCOE.Footnote 45 Latin influence is a distinct possibility for three of them, especially given that the two texts involved show signs of Latin-influenced syntax generally.Footnote 46 However, the remaining two look more persuasive. The first of these is given in (22).

  1. (22)

Footnote 47 The Benedictine Rule is a translation, but the construction does not appear to correspond to a similar construction in the Latin; caution is needed since we do not know precisely which version of the Latin text was used, but the translation appears to be very free at the crucial point.Footnote 48

The second example, (23a), is another interesting case. Being a verse from the Gospels, it is obviously a close translation of Latin. Yet the Latin version, given in (23b), is not mirrored exactly; if it had been, we should have got a ‘neither . . . nor’ construction with ne (or nawðer ne) introducing the whole non-finite clause. Instead ne is placed in the middle of the clause. So even though neque and ne are adjacent to the following present participle in Latin and OE respectively, the resulting construction in (23a) looks different from the one found in the Latin. The translator could have preserved the Latin word order, with the object following the participle. Or if he wished to alter the position of the object relative to the verb, there is no reason why it had to be placed before ne, ruling out the ‘neither . . . nor’ interpretation, yet the translator did just that. This would be hard to explain if using ne ‘not’ to negate present participles was ungrammatical.

  1. (23)

Footnote 49

The number of examples is very limited, and as said, Latin influence is a definite possibility in at least three out of five, but given just how rare it is for something that looks like the negative particle to be found anywhere other than immediately before a finite verb,Footnote 50 and taking into account that <ne> spellings of na are very infrequent, examples (22) and (23a) suggest that the use of ne with present participles may have been an option.

However, the very limited evidence also indicates that Ælfric at least normally used na rather than ne with present participles. Excluding instances of na . . . ac, there are four instances of na in participle phrases headed by a present participle in the YCOE, three of which are from Ælfric's Catholic Homilies. One (ÆCHom II, 11, 103.392) contains only na and the participle so that it could easily be regarded as involving an adjective. The example in (24), on the other hand, involves a close parallel of the type of context where Ælfric normally uses ne with bare infinitives.Footnote 51 A single example cannot prove that ne was unusual or ungrammatical in this context for Ælfric, of course, but it suggests that na was more likely than not the more usual choice at least, which would contrast with bare infinitives.

  1. (24)

We have seen, then, that ne is not found in the YCOE in combination with either inflected infinitives or past participles, whereas na is attested in the corpus with both of these forms. However, examples are very limited, and clear ones even more so, so we are not in a position to conclude safely that ne could not be used with either of these verb forms, even if the one clear example we have for inflected infinitives suggests that they probably did not have the same preference for ne as bare infinitives in the coordination context in Ælfric's language. A similar example with na may suggest the same for present participles, but we also have limited evidence indicating that use of ne with present participles may have been possible in at least some other varieties of OE, even if na is found too.

5 Implications

The existence of the ne + infinitive construction in OE is obviously important from a descriptive point of view and the main purpose of this article has been to cover the more descriptive aspects of the phenomenon. However, it also has consequences for the analysis of OE. A full discussion of these is beyond the scope of this article, but this section highlights the main issues.

The most obvious area where the findings would seem to have implications is in the formal analysis of OE negation. One proposed analysis of OE ne is to treat it as the head of a functional projection (NegP) which attaches to the finite verb as it moves through this position (e.g. Frisch Reference Frisch1997: 31; van Kemenade Reference Kemenade, Tieken-Boon van Ostade, Tottie and van der Wurff1999: 152; Fischer et al. Reference Fischer, van Kemenade, Koopman and van der Wurff2000: 126; van Gelderen Reference Gelderen2004: 81). Non-finite verbs are not subject to the same movement as finite verbs, so such an analysis would appear to predict that ne cannot attach to non-finite verbs. That was of course a desirable prediction while attachment of ne was believed to be restricted to finite verbs, but the data presented here show that it is actually possible for ne to be used with bare infinitives. Notice also that ne is consistently adjacent to the infinitive in the ne + infinitive construction, indicating that it attaches to the infinitive. The data are limited, but this pattern does not strike me as coincidental; compare the even more limited number of cases with na, where we do get non-adjacency (example (14b), plus (15) if regarded as relevant). The ne + infinitive construction appears to pose serious problems for the NegP plus movement analysis, then, as it simply cannot account for attachment of ne to infinitives, at least under current assumptions of OE verb movement.

Others regard OE ne as a prefix on the verb in V rather than as the head of NegP (e.g. van Kemenade Reference Kemenade, Pintzuk, Tsoulas and Warner2000: 65–6; Ingham Reference Ingham2007). At first sight, the ne + infinitive construction would seem to provide evidence in favour of such an approach, since it makes no direct prediction that attachment of ne must be restricted to finite verbs. However, that does not mean that it automatically solves all problems. A satisfactory analysis of OE negation should not predict free attachment of ne to non-finite forms; it must account both for the normal restriction to finite verbs and for the possibility of attachment to bare infinitives in the absence of an available finite verb. Finding a good mechanism to achieve both things at the same time could be a challenge.

Moreover, it is not certain that it has to be ne which is behaving abnormally in the absence of a finite verb. In principle, it could be the infinitive, and work by Koopman (Reference Koopman2005) on postverbal pronouns and particles suggests that this possibility needs to be considered. Koopman finds that placement of personal pronoun objects and/or particles after non-finite verbs is especially frequent in coordinate contexts (specifically those involving coordinated non-finite verbs), as well as in uton constructions and accusative-cum-infinitive (ACI) constructions. He likewise points to the ‘lack of a finite verb’ as a common factor in these contexts (Koopman Reference Koopman2005: 61). If we assume that pronouns and particles are not subject to rightward movement through extraposition (see Pintzuk Reference Pintzuk1991: 87, 95), then postverbal placement of these pronouns and particles should be the result of leftward movement of the non-finite verb, arguably of the kind normally restricted to finite verbs.Footnote 52 And if they can behave like finite verbs in terms of placement, it would not be too surprising if their syntactic behaviour were to show other traits normally associated with finite verbs, such as negation through attachment of ne. However, Koopman's data also suggest that the situation is complicated. Notably, there may be variation between contexts and texts/authors,Footnote 53 so that a satisfactory account of how movement of the infinitive in the absence of a finite verb would work in principle might not be sufficient to account for the data (at least not in full). More work is needed before we can decide how these data are best analysed.

Before a full analysis of the ne + infinitive phenomenon can be offered, then, it should first be established whether the construction is to do primarily with the behaviour of ne or that of the infinitive. In the former case, the analysis of OE negation would need revising (and an analysis of ne as head of NegP plus attachment through verb movement would probably be ruled out), whereas in the latter case, the implications would probably concern the analysis of OE verb movement (and the analysis of negation might remain essentially unaffected).

Finally, a full analysis of ne + infinitive constructions would also need to account for the apparently more frequent failure of negative concord in this environment as compared to contexts involving finite verbs. A possible way of dealing with this in a formal analysis might be to treat it as variation between two structures for infinitival conjuncts, one including functional projections such as NegP, triggering negative concord, and another conjoining at the VP level.

6 Conclusion

The data show that the OE negative particle ne can indeed be used with bare infinitives in both the uton construction and the coordination context. For uton, the evidence is all but limited to a single manuscript, but it looks robust for that variety of OE at least. There are more data for the coordination context. They show not only that the construction exists but also that ne is the norm in this context when negation is introduced in the conjunct clause concerned, certainly in the language of Ælfric, and possibly late West Saxon in general. To the extent that the requisite contexts occur, then, we are not dealing with a marginal option of using ne instead of na, so the pattern needs to be accounted for in analyses of OE. At present, it clearly is not.

I have suggested that the crucial property shared by the two constructions in which ne + infinitive occurs is the absence of an available finite verb, as there is both morphological and syntactic evidence that uton does not behave like a (normal) finite verb. It is less clear whether this context enables exceptional behaviour of the negative particle or of the infinitive; evidence from Koopman (Reference Koopman2005) on the placement of object pronouns and particles indicates that the latter possibility should be explored. This issue needs to be resolved before a full analysis of the phenomenon can be offered.

It has also been shown that negative concord appears to be less consistent in the context where ne + infinitive can be found than it is with finite verbs, suggesting that, although attaching ne to bare infinitives is certainly an option, there may be some reluctance to use it.

It has not been possible to provide a reliable answer to the question whether ne could attach to other non-finite verb forms under comparable circumstances or whether the phenomenon is restricted to bare infinitives. The limited data suggest that ne may to some extent be found with present participles at least, but if so, it is not clear that they follow exactly the same pattern as bare infinitives. Should the ne + infinitive construction be the result of infinitives behaving in ways more characteristic of finite verbs when the finite verb is absent, it might be possible to extract some additional relevant data by looking, for example, at pronoun placement in relation to various non-finite verb forms. Even in that case, however, the data may well be too limited again to settle the issue for all types of non-finite forms; some questions simply cannot be answered fully on the basis of the surviving OE material. On the other hand, sometimes even small data sets can provide convincing and important evidence, as this article has illustrated.

Appendix: Instances of ne + infinitive in the YCOE not given in the main text

Examples (A3) and (A7), like (9), involve dependency on uton. Examples (A1)–(A6) are from Ælfric, and (A9) has also been attributed to him.

  1. (A1)

  2. (A2)

  3. (A3)

  4. (A4)

  5. (A5)

  6. (A6)

  7. (A7)

  8. (A8)

  9. (A9)

Footnote 54Footnote 55Footnote 56Footnote 57

Footnotes

1

I would like to thank Rhona Alcorn, John Anderson, Fran Colman, Wim van der Wurff and two anonymous reviewers for feedback on a draft of this article. The beginnings of the research leading to this article date back to my time as a British Academy postdoctoral fellow; I gratefully acknowledge the British Academy for their financial support during that period.

2 The system of reference for the location of examples adopted throughout this article is based on the one used in Cameron et al. (Reference Cameron, Crandell Amos, Butler and DiPaolo Healey1981). For details, see Healey & Venezky (Reference Healey and Venezky1980 [1985]).

3 A partial exception is Wallage (Reference Wallage2005: 78), who identifies the construction with uton (without mention of Mitchell (Reference Mitchell1985) or the coordination context). I will come back briefly to Wallage (Reference Wallage2005) in section 4.

4 Compare Jäger's (Reference Jäger2008: 70ff.) observations on the cognate Old High German particle ni ‘not’. This particle is also normally restricted to a position immediately before the finite verb, but it occasionally occurs before non-finite forms in one particular translation from Latin, normally in the absence of a finite verb such as in absolute constructions that mirror the corresponding Latin. She concludes that these examples are ‘probably ungrammatical loan constructions’ (Reference Jäger2008: 71). The only non-translated instance she found in another text appears to involve scribal error (accidental repetition over a page turn; the phrase is deleted in the manuscript, as noted in Piper's (Reference Piper1883) edition of the text concerned). In the absence of examples without probable Latin influence or scribal error, Jäger's conclusion seems justified.

5 The first example does not involve coordination of clauses (na is coordinated with another adverb phrase) and the second occurs before an inflected infinitive rather than a bare infinitive.

6 In addition, Einenkel (Reference Einenkel1912: 209) gives an example which he describes as having an infinitive used as a subjunctive (the use of ne is still seen as erroneous): Ðaræfter ne slean man; ne unriht hæme; ne stele; ‘after that, do not kill anyone; do not commit adultery; do not steal’ (BenRW 23.14). There is no mention of such use of infinitives in Mitchell (Reference Mitchell1985), and the example occurs in an early thirteenth-century manuscript, so it is not reliable evidence for OE. Latin influence is also possible (it corresponds to non occidere).

7 Lat. Malum pro malo non reddere. Iniuriam non facere (D'Aronco Reference D'Aronco, Lendinara and Melazzo1983: 121–2).

8 The YCOE was searched using CorpusSearch, written by Beth Randall. All cited examples were checked against a printed edition, and information on manuscript variants usually derives from that source. (In a few cases, facsimile material was checked.) The editions consulted were normally those used in the YCOE; for details, see www-users.york.ac.uk/~lang22/YCOE/info/YcoeTextInfo.htm Alternative/additional editions used are Bazire & Cross (Reference Bazire and Cross1982), Cubbin (Reference Cubbin1996), D'Aronco (Reference D'Aronco, Lendinara and Melazzo1983), Irvine (Reference Irvine2004), Liuzza (Reference Liuzza1994, 2000), O'Brien O'Keefe (Reference Migne2001) and Schröer (Reference Schröer1888). Where the possibility of Latin influence needed to be checked for, Fontes Anglo-Saxonici was consulted to locate the source when not provided in the edition of the OE text.

9 There are also no instances of uton constructions with other negative phrases such as næfre ‘never’ in the YCOE, and I did not find any instances with placement of the negative particle ne before uton rather than the infinitive in either the YCOE or the DOE Corpus.

10 Latin Non sinam omnino concludi hunc libellum ‘I will not let this book be ended at all’.

11 There are four potential instances in the YCOE files of this text against 23 of <na> or <no>. At least two of those (SolilPref 1.8 and Solil 1, 16.5) look like straightforward cases. Solil 2, 61.16 may be an instance too. Solil 3, 69.9 and ne magon heom þeah na nane gode ne beon ‘and can nevertheless not be any good to them’ is probably another case since a second instance of <ne> is present before the finite verb, but the additional presence of <na> in the clause makes the example hard to interpret, plus beon is an abbreviated form in the manuscript (expanded to the finite form beoþ instead by one editor).

12 Thirty-seven instances of <ne> are parsed as the negative adverb na in the YCOE compared to 1,941 instances of <na>/<no>; at least five of the 37 should be excluded (apparent errors in the corpus or edition) but there are also five instances of <ne> parsed as the negative particle ne which are probably <ne> spellings of na (they involve reasonably clear cases of constituent negation and/or ne is also found before the finite verb).

13 Two examples found in a single passage (HomS 40.3 [ScraggVerc 10] 45) can easily be reinterpreted as involving the coordinator ne ‘nor’ if we assume a slightly different clause division than the one suggested by the editorial punctuation. On the evidence of the other witnesses, HomS 40.3 [ScraggVerc 10] 246 is probably a case of scribal corruption that also involves ‘nor’ (accidental omission of the word preceding ne). That leaves HomU 6 [ScraggVerc 15] 49 þæt nænig man ne wat to secganne ne nænigum eorðcyninge be ðam scipliðendum ‘that no one knows how to tell any king of land about the seamen’. Here the placement of <ne> following an inflected infinitive looks unusual even for na; there are no other witnesses to confirm, but there might be scribal corruption of some sort.

14 According to the YCOE data there is also one potential instance of ne separated from the finite verb in the Vercelli Homilies (HomS 40.3 [ScraggVerc 10] 181), but this example is more plausibly interpreted as involving a ne . . . ne ‘neither . . . nor’ construction. That is in fact the only possible interpretation in the other four witnesses, where another instance of ne is found in its expected position before the finite verb.

15 The data collection was restricted to examples involving bare infinitives. Definite cases of constituent negation were excluded automatically by the searches used. It is unlikely that any potentially relevant instances were missed as a result; the policy adopted by Taylor et al. (Reference Taylor, Warner, Pintzuk and Beths2003) was to parse na and other negative elements as sentential negators whenever such an interpretation was possible (even if unlikely).

16 If Mitchell's suggestion that analogy with the coordination context could be a factor in the occurrence of ne + infinitive with uton is on the right track, they do show how its use in the coordination context might have led to its appearance in clauses with uton.

17 See Pope (Reference Pope1967: 143) and Clemoes (Reference Clemoes and Clemoes1959) for information on which parts of the OE Heptateuch can be attributed to Ælfric.

18 The apparatus in Clemoes (Reference Clemoes1997) and Godden (Reference Godden1979) does not note spelling variants but variation between <ne> and <na> has apparently been treated as lexical variation, so it is noted.

19 In Ælfrician material not included in the YCOE I found another example of this type, as well as two addional instances of ne + infinitive in the coordination context: ÆCHom I, 17 (App) 537.80 don . . . ne geefenlæcan ‘do . . ., i.e. not match’ (two manuscripts have a finite form instead); ÆAbusMor 247 ne abitan ‘not consume’ and ÆAbusMor 240 ne gyman ‘not take heed’ (paralleled in ÆAbusWarn).

20 The YCOE codes ne as a coordinating conjunction here, but that does not fit the meaning. The other surviving manuscript has læton, and ne is a conjectural reading for that manuscript.

21 Warner (Reference Warner1982: 210–1) and Wallage (Reference Wallage2005: 178ff.) both observe that, even when not has become the normal form of negation in Middle English, pleonastic negation still takes the form of unsupported ne in this context; not is used for pleonastic negation in complements of ‘negative meaning’ verbs in Late Middle English, but not when the higher verb is negated. According to Wallage, pleonastic negation disappears from this particular context with the loss of ne.

22 The YCOE has . . . se weolocreada tælgh, þone ne mæg sunne blæcan ne ne regn wyrdan, deriving from the printed edition, which amalgamates versions found in different manuscripts and may not reflect what was found in any of them. The text given in (13c) is the version found in the alternative base manuscript normally used in the edition in the absence of the most authoritative witness. The other surviving Anglo-Saxon manuscript of the text has ðone nu sune blæcan ne mæg ne ren ‘which sun cannot bleach now, nor rain’ (reflecting the corresponding Latin text more closely than the version in (13c)). The form regn was taken from collated forms in an edition of the text made before Cotton Otho B. xi. was largely destroyed by fire, and it is not clear to me from that edition what this manuscript had precisely.

23 Two near-certain cases of constituent negation were excluded (ChronE 1087.65 and Nic (C) 95, both involving the phrase na mare ‘no more’), as was one instance involving a strongly contrastive construction where ne would probably not have been an alternative (. . . þæt we sceoldon ða hwilwendlican ðing to urum bricum habban. na on ure heortan lufe healdan ‘. . . that we should have the transitory things to our use, not hold them in our heart's love’ (ÆCHom II, 36.1, 268.24)).

24 The instance of na and the infinitive wanian found in (13a) could be regarded as a fifth case, and was in fact retrieved by the search of the YCOE data, but it has been excluded on the grounds that it is part of the first conjunct and straightforwardly dependent on the negated finite verb ne sceolan ‘must not’, even if it does start with a coordinating conjunction (nawþer ne ‘neither’).

25 There are occurrences of ute/uto/uta we in the DOE Corpus in texts not included in the YCOE, but not many, and usually in texts/manuscripts which also contain one or more instances of such ‘reduced’ forms of uton without a following subject pronoun (HomS 6, HomS 42, BenRW and Fates), so these reduced forms probably have nothing to do with the following pronoun. Only in the case of Lit 5.11.6 ute we is there no reduced form of uton in a different context, but no other instance of uton is found in that text at all, nor (as far as I can determine) in the manuscript.

26 As can be seen, the degree of consistency of this phenomenon is extremely high in the Catholic Homilies, for which (as said earlier) we have very reliable contemporary manuscript evidence. In addition, one of the three unreduced forms in this text is found in a passage that is omitted from the most authoritative manuscripts.

27 Six in Catholic Homilies, 1 in Lives of Saints and 3 in Supplementary Homilies.

28 Two instances of <uton> supplied from a different manuscript have been excluded.

29 Wulfstan's Homilies show something similar: whereas the form <-an> is found about as often as the form <-on> for the past indicative plural ending and about two-thirds of the time for the relevant present indicative plural forms, uton has the form <utan> in 47 out of 54 instances, and five out of the seven instances of <uton> are found in one short passage, taken from a manuscript which is not the usual base manuscript and which nearly always uses the form <uton> rather than <utan>. So <utan> is more frequent than would be expected if uton were a finite verb form with the inflectional ending -on. This is not a pattern that is found in all texts in the YCOE, however; in most, the regular form is <uton>.

30 The placement of uton resembles the typical placement of imperatives in OE, but unlike uton, imperatives can also occur elsewhere (especially in conjunct clauses), including clause-finally and/or following a subject pronoun, e.g. ÆCHom I, 34, 468.88 & þu ðær tomerigen mæssan gesing ‘and sing mass there tomorrow’.

31 A similar development can be seen in PDE adhortative let's, as discussed e.g. in Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 935).

32 There is also one instance that involves negative contraction (GD 1 (C) 9.61.17 oððe hwæt hi sceolan nyllan ‘or what they must not want’), but the few verbs allowing negative contraction in OE may well have lexicalised into negative verbs to some degree (in the case of willan and witan potentially encouraged by the existence of corresponding negative verbs in Latin, i.e. nolle ‘not want’ and nescire ‘not know’), in which case nyllan may not be subject to exactly the same constraints as forms with ‘normal’ attachment of the negative particle. Present participles with negative contraction are attested as well in the YCOE for both willan and witan; see fn. 45.

33 Mitchell (Reference Mitchell1985: §1602) mentions an apparent counterexample from poetry rather than prose. He comments that ‘it is presumably to be attributed to the demands of metre’, which seems plausible.

34 Another question that could be asked in relation to further possible constraints is whether ne + infinitive is restricted to the two contexts discussed, or whether ne can also be used to negate infinitival complement clauses of uncontroversial main verbs such as e.g. hatan ‘command’. Unfortunately, this question cannot be answered. There are no cases in the YCOE where negation is introduced in such a complement clause, as all complement clauses headed by a bare infinitive that contain a negative word or phrase involve negation of the higher verb (except for some in the coordination context, e.g. (A4) in the Appendix) – the only apparent exception (Bede 4, 28.364.1) involves editorial emendation. We can only say that ne does not appear to be repeated before the infinitive in such clauses for the purposes of negative concord (e.g. ÆCHom II, 12.2, 125.520 ðam men ne lyst nan ðing to gode gedon ‘that person does not desire to do anything for a good purpose’), but even for that, data are limited.

35 For practical reasons, I am ignoring any instances in the parts of the Heptateuch attributed to Ælfric: these parts have not been separated out in the relevant YCOE file from the remainder of the text. Cases with negative contraction have been included. The YCOE has four apparent instances without the negative particle, but two of these actually do have ne (ÆCHom II, 22, 194.125 involves an error in the corpus text, and ÆLet 3 (Wulfstan 2) 186 has a <næ> spelling of ne, as confirmed by the other witnesses). The third one was excluded – ÆLS (Martin) 63 probably has a negative particle rather than a negative conjunction before the verb – leaving just one clear case where the negative particle has been omitted (ÆCHom II, 21, 184.125), and even here three of the manuscripts supply ne before the finite verb.

36 The other two are ÆCHom I, 38, 511.122 nanum gebeodan ‘to command no one’ and ÆLet 6, 280 and nænne hatian ‘and to hate none’. The context of the latter is complicated and it might involve omission of nelle rather than wylle.

37 WHom 10c, 44 & næfre hyre derian wordes ne weorces ‘and never harm her in word or deed’.

38 The other two are BenR 4.17.5 with nan þing ‘nothing’ and 7.25.20 with to nahte ‘to nothing’.

39 Bede 3, 19.242.32 with næfre ‘never’ and Bede 5, 1.386.2 nænig hyht haelo ‘no hope of safety’. The remaining instances in the YCOE of infinitival conjuncts with a negative phrase other than na, without use of ne and not introduced by a negative conjunction, are the two instances with na mare already mentioned in fn. 23, ChronE 1093.3 with næfre ma ‘never more’, CP 63.461.5 nane wuht ‘nothing’, Med 3 (Grattan-Singer) 31.14 næs ‘not at all’ and HomU 9 (ScraggVerc 4) 235 næfre ‘never’.

40 In two instances (ÆCHom II, 38, 283.129 and GD 1 (H) 4.37.9) na looks heavily contrastive. In the case of Ch 1533, 7, parallel phrases with næfre instead of na are found twice in the same charter, suggesting the meaning ‘never’ rather than just ‘not’.

41 It survives in just one manuscript, but it is a near-contemporary witness: Ker (Reference Ker1957: 51) dates it to the beginning of the eleventh century. While it is a fairly close translation from a Bible passage – Matthew 23.23 Haec oportuit facere, et illa non omittere (Pope Reference Pope1967: 252) – it is difficult to see how Latin influence could have resulted in the choice of na rather than ne here. Note that (20) also shows that ‘independent’ negation of the type found in constructions like ‘They told him not to do this’ is attested with to-infinitives in OE, not just in purpose adjuncts as pointed out by Los (Reference Los2005: 21, fn. 3), but also in a complement.

42 Mk (WSCp) 2.26 þe him ne alyfede næron to etanne ‘which were not allowed to him to eat’. While an interpretation of <ne> as the negative particle is implausible here, it is striking that this spelling is found in four out of five manuscripts (even if two are from the twelfth century or later); the fifth omits it.

43 Sole other constituent: ÆLS (Christmas) 35, 63 and 76, ÆLS (Auguries) 17, ÆCHom I, 2, 196.196, ÆCHom I, 38, 513.190, ÆCHom II, 11, 97.185, and GDPref and 4(C), 41.328.21; na . . . ac ‘not . . . but (rather)’: ÆCHom I, 12, 277.62; næs na ‘not at all (not)’ (likely to be a fixed combination): ApT 12.14, HomS 10 (BlHom 3) 29.45; heavily contrastive constructions of the type illustrated in fn. 23: ÆCHom II, 41, 306.63, ÆLS (Book of Kings) 325. The example in ÆLet 2 (Wulfstan 1) 164 was excluded from the data altogether; it involves an abbreviation in the base manuscript.

44 Mk (WSCp) 12.20 na læfedum sæde ‘not having left any offspring’ (Latin: non relicto semine [Weber Reference Weber1975]). There is a second absolute construction containing na, also mirroring a Latin absolute (GDPref and 3 (C) 13.197.26 na þa gyt eallunga geendedum ‘not having completely ended yet’), but na þa gyt is probably a phrase here – it corresponds to necdum in the Latin [De Vogüé 1979: 3.13.9]).

45 In addition there are some forms of nellende ‘not wanting, unwilling’ and nytende ‘not knowing, ignorant, ignorantly’ in various contexts/uses in the YCOE, but as said in fn. 32, forms with negative contraction may well be a separate case, so these have been left out of consideration.

46 Bede 1, 16.86.10 ne weotende ‘not knowing, without knowing’ (Latin nesciens [Colgrave & Mynors Reference Colgrave and Mynors1969: 98.35]); Bede 4, 3.270.35 ne weotendumoððo ne gemændum þæm heordum þære stowe ‘without the guardians of that place knowing or telling’ (Latin nescientibussiue non curantibus loci custodibus [Colgrave & Mynors Reference Colgrave and Mynors1969: 344.36]; treated as a single example); LS 23 (MaryofEgypt) 185 þæt geswinc his syð-fætes ne under-standende ‘not perceiving the toil of his journey’ (Latin laborem non reputans itineris [Migne Reference Migne1844–55, 1862–5: 73, 677B]). For the examples from Bede, the other extant witnesses diverge from the base manuscript, which at least in the case of the second example means that ne is not combined with a present participle in those manuscripts.

47 The edition and corpus have the comma before na, but that makes little sense to me.

48 Latin Quartus humilitatis gradus est si . . . patientiam amplectatur et sustinens non lassescat vel discedat (Hanslik Reference Hanslik1960: 46.35) ‘The fourth step of humility is if . . . he should embrace patience and, enduring, not weaken or give up’.

49 One manuscript, Cambridge University Library ms. Ii.2.11, has na, but ne is found in five out of six manuscripts. (Two are very late, but if excluded, that still leaves three witnesses with <ne>.)

50 Instances in the YCOE where the negative particle is not left-adjacent to a finite verb outside the contexts and examples discussed in this article are very few and far between, and most involve probable or definite misparses in the corpus, plus a few misprints and likely editorial or scribal errors. Excluding those, just three remain (against ca. 12,000 that are adjacent to a finite verb): Alex 21.11 we hie ne onweg flegdon ne him lað dydon ‘we did not chase them away nor did them harm’, which may be a case of an awkwardly constructed translation of the ‘neither . . . nor’ construction in the Latin source; Bo 8.20.19 neon become ‘not come upon’, which is not easily accounted for as involving an inseparable prefix (Mitchell Reference Mitchell1985: §1073), but the crucial bit only survives in a twelfth-century manuscript; and Lch II (1) 38.1.4 fersc ne nyt biþ ‘fresh is not beneficial’, which is hard to explain; treating it as a rare <ne> spelling of na would mean failure of negative concord as well (not unattested in this text, but infrequent).

51 The remaining two examples are ÆCHom II, 2, 15.134 naswa ðeah cwacigende ‘however not trembling’ and Med 1.1 (de Vriend) 14.1.476 naþonne gyt geseondne ‘not yet seeing’. LS 23 (Mary of Egypt) 199 was excluded, as it probably involves nahwider ‘nowhere’.

52 In a double-base analysis in which both underlying OV and VO structures are available as proposed by Pintzuk (Reference Pintzuk1991), the alternative is a higher frequency of underlying VO, but it seems harder to connect that with the absence of a finite verb in any way.

53 According to Koopman (Reference Koopman2005: 55–7), postverbal pronouns are frequently found in the coordination context, whereas postverbal particles are particularly common in ACI constructions. He also states that, in the uton construction, Wulfstan ‘always puts a personal pronoun after the infinitive . . ., whereas Ælfric prefers the position before the infinitive’ (Reference Koopman2005: 55).

54 One witness has <na>, but it involves a twelfth-century manuscript and the other ten witnesses have <ne>.

55 Corresponds to Latin aliena non appetere (Godden Reference Godden2000: 469, quoting the Latin text from Migne Reference Migne1844–55, 1862–5: 76, 1150 BC]), so Latin influence is possible.

56 One of the seven witnesses (a mid-eleventh-century manuscript) omits ne.

57 Influence from Latin is possible: haec autem oportuit facere et illa non omittere (Lk 11.42 [Weber Reference Weber1975]).

References

Bazire, Joyce & Cross, James E. (eds.). 1982. Eleven Old English Rogationtide homilies. Toronto: University of Toronto Press.Google Scholar
Cameron, Angus, Crandell Amos, Ashley, Butler, Sharon & DiPaolo Healey, Antonette (eds). 1981. The Dictionary of Old English Corpus in electronic form. Toronto: Dictionary of Old English Project.Google Scholar
Campbell, Alistair. 1959. Old English grammar. Oxford: Clarendon Press.Google Scholar
Clemoes, Peter. 1959. The chronology of Ælfric's works. In Clemoes, Peter (ed.) The Anglo-Saxons: Studies in some aspects of their history and culture presented to Bruce Dickins, 212–47. London: Bowes & Bowes.Google Scholar
Clemoes, Peter (ed.). 1997. Ælfric's Catholic homilies: The first series (EETS SS 17). Oxford: Oxford University Press.Google Scholar
Colgrave, Bertram & Mynors, Roger A.B. (eds.). 1969. Bede's ecclesiastical history. Oxford: Clarendon Press.Google Scholar
Cubbin, Geoffrey P. (ed.). 1996. The Anglo-Saxon Chronicle: A collaborative edition, vol. 6: MS D. Cambridge: D. S. Brewer.Google Scholar
D'Aronco, Maria Amalia. 1983. Il IV capitolo della Regula Sancti Benedicti del ms. Londra, B.M., Cotton Tiberius A. iii. In Lendinara, Patrizia & Melazzo, Lucio (eds.), Feor ond neah: scritti di filologia germanica in memoria di Augusto Scaffidi Abbate, 105–28. Palermo: University of Palermo.Google Scholar
De Vogüé, Adalbert (ed.), with French translation by Paul Antin. 1978–80. Dialogues, 3 vols. (Sources chrétiennes 251, 260, 265). Paris: Cerf.Google Scholar
Einenkel, Eugen. 1912. Die englische Verbalnegation: ihre Entwickelung, ihre Gesetze und ihre zeitlichörtliche Verwendung. Anglia 35, 187–248.Google Scholar
Fischer, Olga, van Kemenade, Ans, Koopman, Willem & van der Wurff, Wim. 2000. The syntax of early English. Cambridge: Cambridge University Press.Google Scholar
Fontes Anglo-Saxonici Project (ed.). Fontes Anglo-Saxonici: World Wide Web register. http://fontes.english.ox.ac.ukGoogle Scholar
Frisch, Stefan. 1997. The change in negation in Middle English: A NEGP licensing account. Lingua 101, 21–64.Google Scholar
Gelderen, Elly van. 2004. Economy, innovation and prescriptivism: From spec to head and head to head. Journal of Comparative Germanic Linguistics 7, 59–98.Google Scholar
Godden, Malcolm (ed.). 1979. Ælfric's Catholic homilies: The second series (EETS SS 5). London: Oxford University Press.Google Scholar
Godden, Malcolm. 2000. Ælfric's Catholic homilies: Introduction, commentary and glossary (EETS SS 18). Oxford: Oxford University Press.Google Scholar
Hanslik, Rudolf (ed.). 1960. Benedicti regula (Corpus Scriptorum Ecclesiasticorum Latinorum 75). Vienna: Hoelder-Pichler-Tempsky.Google Scholar
Healey, Antonette DiPaolo & Venezky, Richard L.. 1980. A microfiche concordance to Old English: The list of texts and index of editions. [Reprint with corrections 1985.] Toronto: Dictionary of Old English Project.Google Scholar
Huddleston, Rodney & Pullum, Geoffrey K.et al. 2002. The Cambridge grammar of the English language. Cambridge: Cambridge University Press.Google Scholar
Ingham, Richard. 2006. On two negative concord dialects in early English. Language Variation and Change 18, 241–66.Google Scholar
Ingham, Richard. 2007. NegP and negated constituent movement in the history of English. Transactions of the Philological Society 105, 365–97.CrossRefGoogle Scholar
Irvine, Susan (ed.). 2004. The Anglo-Saxon Chronicle: A collaborative edition, vol. 7: MS E. Cambridge: D.S. Brewer.Google Scholar
Jäger, Agnes. 2008. History of German negation. Amsterdam and Philadelphia: John Benjamins.Google Scholar
Kemenade, Ans van. 1999. Sentential negation and clause structure in Old English. In Tieken-Boon van Ostade, Ingrid, Tottie, Gunnel & van der Wurff, Wim (eds.), Negation in the history of English, 147–65. Berlin: Mouton de Gruyter.Google Scholar
Kemenade, Ans van. 2000. Jespersen's cycle revisited: Formal properties of grammaticalization. In Pintzuk, Susan, Tsoulas, George & Warner, Anthony (eds.), Diachronic syntax: Models and mechanisms, 51–74. Oxford: Oxford University Press.Google Scholar
Ker, Neil R. 1957. Catalogue of manuscripts containing Anglo-Saxon. Oxford: Clarendon Press.Google Scholar
Koopman, Willem. 2005. Transitional syntax: Postverbal pronouns and particles in Old English. English Language and Linguistics 9, 47–62.Google Scholar
Liuzza, Roy M. (ed.). 1994, 2000. The Old English version of the gospels, 2 vols. (EETS OS 304, 314). Oxford: Oxford University Press.Google Scholar
Los, Bettelou. 2005. The rise of the to-infinitive. Oxford: Oxford University Press.Google Scholar
Mitchell, Bruce. 1985. Old English syntax, 2 vols. Oxford: Clarendon Press.CrossRefGoogle Scholar
Migne, Jacques-Paul (ed.). 1844–55, 1862–5. Patrologiae cursus completus. Series latina. Paris. Electronic version: Patrologia Latina Database. Cambridge: Chadwyck-Healey. http://pld.chadwyck.co.ukGoogle Scholar
O'Brien O'Keefe, Katherine (ed.). 2001. The Anglo-Saxon chronicle: A collaborative edition, vol. 5: Ms. C. Cambridge: D. S. Brewer.Google Scholar
Ogura, Michiko. 2000. ‘Gewat + infinitive’ and ‘uton + infinitive’. Neuphilologische Mitteilungen 101, 69–78.Google Scholar
Pintzuk, Susan. 1991. Phrase structures in competition: Variation and change in Old English word order. PhD dissertation, University of Pennsylvania.Google Scholar
Piper, Paul (ed.). 1883. Die Schriften Notkers und seiner Schule. Zweiter Band. Psalmen und Katechetische Denkmäler nach der St. Galler Handschriftengruppe. Freiburg and Tübingen: J. C. B. Mohr. Accessed online: www.scribd.com/doc/21238252/Die-Schriften-Notkers-und-seiner-Schule-Band-2Google Scholar
Pope, John C. (ed.). 1967. Homilies of Ælfric: A supplementary collection, vol. I (EETS OS 259). London: Oxford University Press.Google Scholar
Roberts, Ian & Roussou, Anna. 2003. Syntactic change: A minimalist approach to grammaticalization. Cambridge: Cambridge University Press.Google Scholar
Schröer, Arnold (ed.). 1888. Die Winteney-Version der Regula s. Benedicti. Tübingen: Max Niemeyer.Google Scholar
Scragg, Donald G. 1973. The compilation of the Vercelli book. Anglo-Saxon England 2, 189–207.CrossRefGoogle Scholar
Scragg, Donald G. (ed.). 1992. The Vercelli homilies and related texts (EETS OS 300). Oxford: Oxford University Press.Google Scholar
Taylor, Ann, Warner, Anthony R., Pintzuk, Susan & Beths, Frank (eds.). 2003. The York–Toronto–Helsinki Parsed Corpus of Old English Prose. York: Department of Language and Linguistic Science, University of York.Google Scholar
Visser, F. Th. 1966. An historical syntax of the English language, part II: Syntactical units with one verb (continued). Leiden: E. J. Brill.Google Scholar
Wallage, Phillip. 2005. Negation in Early English: Parametric variation and grammatical competition. PhD dissertation, University of York.Google Scholar
Warner, Anthony R. 1982. Complementation in Middle English and the methodology of historical syntax. London and Canberra: Croom Helm.Google Scholar
Warner, Anthony R. 1993. English auxiliaries: Structure and history. Cambridge: Cambridge University Press.Google Scholar
Weber, Robert (ed.). 1975. Biblia sacra iuxta vulgatam versionem, 2nd edn. Stuttgart: Württembergische Bibelanstalt.Google Scholar
Wenisch, Franz. 1979. Spezifisch anglisches Wortgut in den nordhumbrischen Interlinearglossierungen des Lukasevangeliums. Heidelberg: Carl Winter.Google Scholar
Zwicky, Arnold M. 1985. Clitics and particles. Language 61, 283–305.Google Scholar