Sie sind auf Seite 1von 28

English Language and Linguistics

http://journals.cambridge.org/ELL Additional services for English

Language and

Linguistics:
Email alerts: Click here Subscriptions: Click here Commercial reprints: Click here Terms of use : Click here

Middle English case loss and the creolization hypothesis


Cynthia Allen
English Language and Linguistics / Volume 1 / Issue 01 / May 1997, pp 63 - 89 DOI: 10.1017/S1360674300000368, Published online: 12 September 2008

Link to this article: http://journals.cambridge.org/abstract_S1360674300000368 How to cite this article: Cynthia Allen (1997). Middle English case loss and the creolization hypothesis. English Language and Linguistics, 1, pp 63-89 doi:10.1017/S1360674300000368 Request Permissions : Click here

Downloaded from http://journals.cambridge.org/ELL, IP address: 131.111.185.11 on 09 Feb 2014

ENG. LANG. LING. 1(1): 63-89 (1997)

Middle English case loss and the 'creolization' hypothesis1


CYNTHIA ALLEN Australian National University (Received 7 May 1996; revised 16 September 1996)

The apparent rapidity of the loss of case-marking distinctions in English has been used as evidence that Middle English was a Creole. However, an examination of the available facts indicates that the reduction of case marking was more gradual and orderly than has often been assumed. The impression of 'confused' usage in many texts disappears once a proper distinction between form and category is made. The reduction of case marking seems to have begun with syncretism of forms but retention of categories, with one form spreading into the territory of another. The facts do not support the idea that case marking was drastically reduced as part of the formation of a Creole. However, it is likely that language contact played a significant role in the quick acceptance of internally motivated changes.

1 Introduction As is well known, Middle English (ME), which we can arbitrarily date from about 1100 to about 1450, was a period of very substantial reduction in the inflectional systems of the language. Similar reductions in inflections have taken place in other Germanic languages, notably the mainland Scandinavian languages and Dutch, as pointed out by Thomason and Kaufman (1988: 315-21), but the apparent rapidity of the changes in English has led observers to look at the invasions which England suffered at the hands of the Scandinavians and the French as possible causes. One idea which has enjoyed a considerable degree of popularity is that ME is reminiscent in many ways of a Creole resulting from language contact. Many parallels have been drawn between features of ME and features found in Creole languages by writers such as Domingue (1977), Bailey and Maroldt (1977), Poussa (1982), and Milroy (1984). However, there is little agreement about the nature of this Creole. For example, Bailey and Maroldt suggest that ME was essentially relexified French while Poussa argues that it was the Scandinavian invasion which was decisive, with the creation first of an East Midland Creole and then a koine which later formed a hybrid with London English. In this paper, I will not attempt to argue comprehensively against specific creolization scenarios which these authors have proposed,2 but will instead focus on
1

Some material from this paper was presented as a seminar at the Department of Language and Linguistic Science at the University of York and the Department of English Language at the University of Edinburgh. I am grateful to the audience of these seminars as well as to two anonymous ELL reviewers for their helpful comments on an earlier draft of this paper. Any errors or infelicities are of course my own responsibility. For a summary of articles before 1988, see Wallmannsberger (1988). For extensive arguments againts the Creole status of ME, see Thomason and Kaufman (1988: section 9.8) and also Gorlach (1990 [1986]).

64

C Y N T H I A ALLEN

one particular change, namely the reduction of case-marking morphology and categories in Old English (OE) and Early Middle English (EME). This reduction is one of the features of ME that is always mentioned as an argument that English underwent creolization. One of the characteristics of languages which are universally acknowledged as 'Creoles' is their analytic rather than inflectional nature. The typical Creole does not have case inflection for nouns, although a subject-object distinction is not unusual for pronouns. Mufwene (1993: 140) notes that case distinctions are often lost even in a Creole where the input languages have similar case-marking systems, as in Bantu-based Creoles. Mufwene observes that case marking is different in this respect from person and number distinctions, which are more resistant to being dispensed with, and suggests that Creoles tend to eliminate redundancy and purely formal grammatical categories. The formal category of case marking is not essential for communication because word order and prepositions can do the job. Although Thomason and Kaufman (1988) present many cogent arguments against the idea that the morphological changes of ME should be seen as evidence of creolization with either Danish or French (rather than simply extensive borrowing), they do not discuss the loss of case-marking morphology or categories in any detail. Furthermore, although one might expect that the decline of English case marking would be well documented, this is not in fact the case. Discussion of ME case marking falls into two broad categories: the broad overviews found in handbooks like Mustanoja (1960) or Mosse (1952) and detailed studies of the inflections of individual texts. What we lack is an overview of the changes which took place to the case-marking systems with sufficient details to give an idea of how these changes were implemented. By their nature, the general handbooks could not possibly provide much in the way of detail about regions and periods, but we might hope to piece together a more detailed overview from the individual studies, and such studies certainly provide many important facts for such an overview. But many facts that a linguist needs to know are simply unavailable in the literature, to a large extent because of the difference in point of view and interests of the linguist and English Studies scholars trained in traditional grammar. For example, these scholars frequently use the term 'dative' for both a form and a category, which is harmless when the difference between the two is firmly kept in mind, but it is usually not. The result is that it is often not clear when an author refers to 'datives' in a text whether they are in fact claiming that a dative-accusative distinction is found in the text, or whether they are talking about a form which is historically dative. Therefore, a new examination of the loss of case-marking categories in English is needed. In this paper, I will present evidence concerning the loss of case-marking distinctions in EME which bears generally on the mechanisms involved in this reduction and will argue that although language contact probably accelerated the reduction of case marking, the changes proceeded in too orderly a fashion to be the result of any sort of creolization process which involved the rapid stripping away of inflections, a characteristic of indisputable Creoles. The picture of 'confusion' that

CASE LOSS AND THE ' C R E O L I Z A T I O N ' HYPOTHESIS

65

has frequently been drawn results from inadequate attention to the distinction between forms and categories. Furthermore, the appearance of very rapid change in EME is to a large extent illusory, as the conservative practices of OE scribes often masked substantial changes which had already taken place before the Norman invasion. In what follows I mention some facts which are not necessarily directly relevant to the question of creolization, but are nevertheless relevant to an understanding of either the mechanisms involved in the loss of case marking or the workings of particular systems in ME. 2 The OE situation Before considering the impact of contact we must look briefly at the case-marking system of OE. It is important to note here that although OE had a healthy casemarking system, syncretism of forms was already considerably advanced. The loss of case-marking distinctions in English is part of a process which started long before EME. In fact, when we compare Proto-Germanic with Proto-Indo-European (PIE), we see that the case marking of Germanic was already considerably reduced; eight cases are reconstructable for PIE (see, for example, Lehmann, 1993: 145) but only six can be reconstructed for Proto-Germanic, with the vocative enjoying only a tenuous existence, since with the exception of Gothic it had disappeared by the time any of the languages are attested (Prokosch, 1939: 236). Further loss of case categories took place between the Germanic stage and OE. OE had three genders and four cases, with a fifth (instrumental) having only a marginal existence in the early stages and disappearing before the period when either Scandinavian or French influence could have been involved. There were several inflectional classes of nouns; for details of these classes and their inflections the reader should consult a standard grammar such as Campbell (1959). Broadly, it can be said that each gender had a dominant 'strong' declension. For the masculine and neuter nouns, this was the a-stem declension while for feminine nouns the o-stems represented the dominant declension. Some processes resulting in reduction of inflectional classes are already evident in OE. Importantly, this is true even of dialects which could not yet be seriously affected by contact with Scandinavian. For example, Dahl (1938: 178) comments that all the masculine w-stem nouns with long stems3 have passed over into the a-stem declension, either partly or entirely, even in the early OE texts. Given this sort of regularizing trend already in early OE, it does not seem very surprising to find the w-stem class disappearing entirely in EME when even the short stems go over to the a-stem pattern, giving plural sunes (< sunas) 'sons' instead ofsuna. In these dominant declensions, the distinction between nominative and accusative nouns in OE was already greatly diminished because of widespread syncretism. For
3

The reason why the long stems of this conjugation were more likely to go over to the a-stem class is that the -u of the nominative and accusative was deleted by a regular phonological process, making these forms similar to a-stems.

66

CYNTHIA ALLEN

example, the nominative-accusative distinction was scarcely evident in the plural. Most nouns had identical plural accusative and nominative forms, either through phonological changes or through levelling. In the masculine a-stems the nominative plural ending -6s was extended to the accusative at the West Germanic stage, giving OE -as (Campbell 1959: section 570). In the feminine o-stems the nominative-accusative distinction was~still- marked formally in the plural to some extent in the early West Saxon dialect,4 where the accusative plural suffix -e ( < Gmc -6ns) is found occasionally, although it is usually replaced by -a, the reflex of the Gmc nominative plural form (-dz). It is interesting to note that Campbell (1959: section 585) says that although the accusative form was rare in the texts of this dialect, it was only found in its historically correct function. That is, the nominative form was extended to the accusative function but not vice versa. As discussed below, this is a pattern that we see repeatedly in the history of English: syncretism is not a matter of confusion of categories but rather of one form taking over the functions of another. In the singular, the nominative-accusative distinction was still evident in the feminine o-stems, since -u appeared as the nominative ending after a short stem and no marker occurred after a short stem, contrasting with the -e found as the syncretistic singular non-nominative marker. But the nominative-accusative distinction had disappeared for the masculine nouns, and it had never existed for the neuters. Each gender also had a 'weak' declension in which most case categories were not distinctively marked, the most common ending in the paradigm of all genders being the syncretistic -an. However, the weak masculine and feminine nouns did differentiate the nominative from the accusative in the singular, with the nominative ending in a vowel (-a for masculine, -e for feminine), contrasting with the nonnominative -an. Thus the nominal inflections were highly syncretistic. However, the modifiers of the nouns often made the gender, case and number of the noun unambiguous. The nominative-accusative distinction was marked in the singular in the masculine and feminine definite determiner by the oppositions se~pone (masculine) and seo~pa (feminine). But the plural declension was the same for all genders and the nominative and accusative forms were not distinguished. Adjectives were declined 'strong' or 'weak' according to such factors as whether they were preceded by a determiner. In the weak declension the inflections did not give much information about gender or case, since the ending was most often -an, but the nominative singular was more distinctive, ending in a vowel, as was the general genitive plural -ra (or -end) and the general dative plural -urn. Strongly declined adjectives were similar to the determiner in showing a nominative-accusative distinction for the masculines and feminines in the singular but (mostly) not in the plural. The pronouns differed from Modern English in retaining an accusative-dative
4

'Early West Saxon' refers to the West Saxon of around the end of the ninth century (and before), while 'late West Saxon' refers to the language of c. 1000.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

67

distinction to some extent in the third person: accusative hine (masculine singular), hit (neuter singular), and hie (feminine singular and general plural) vs. dative him (masculine/neuter singular and general plural) and hire (feminine singular). However, already by the late OE stage the dative form had usurped the functions of the accusative in the first and second person, becoming a general object marker. To summarize, syncretism of forms was already well advanced in OE before contact with the Scandinavians or the French could have been a significant factor. This sort of syncretism is of the utmost importance in the loss of category distinction because when syncretism of form accumulates to the point where two categories are distinguished only by a few forms, the category distinction is difficult to maintain. For example, the nominative-accusative distinction was already imperilled by the late OE stage and likely to be lost even without foreign contact. Furthermore, the case-marking inflections were so syncretistic that it would only take a little bit in the way of phonological or analogical changes to alter the system drastically. 3 Early changes and contact with Scandinavian Let us now look at the possible involvement of contact with Scandinavian in the loss of case-marking distinctions in English. There is little reason to doubt the intimacy of the contact between the Scandinavians and the English, as there was surely intermarriage, although we cannot be certain of the number of Scandinavian settlers. The hybrid nature of many place-names in the Danelaw and the nature of the Scandinavian loans both attest to a good deal of intermingling. To assess the impact of the Scandinavian invasions on the English case-marking system, it would be most useful to compare pre-invasion texts from the northern area with post-invasion ones. By 'pre-invasion' I mean texts that date before 900, since Scandinavian settlement began with the invasion of East Anglia in 865, rapidly followed by the establishment of a Danish kingdom in York and the settlement by Danes (and some Norwegians) of the rest of the Danelaw. Unfortunately, the preinvasion remains of early Northumbrian (eNbr) consist only of a few runic inscriptions and five slightly longer texts. Of these, the Leiden Riddle is the longest at fourteen lines. Given such limited data, the fact that these earliest texts do not show the syncretism of forms found in later Northumbrian (INbr) texts of the tenth century means very little. But it is important to note that these eNbr texts, short as they are, do show two developments generally agreed to have played an important role in the loss of case distinctions: the loss of final -n and the reduction of back vowels. Campbell (1959: section 369) notes that the neutralization of front vowels to e had already begun in eNbr and in section 396 states that the 'confusion' of back vowels began before 900 and was completed in Northumbrian and Kentish before it was in West Saxon. The fact that this vowel reduction took place as early in Kentish as in Nbr shows that this process is not due to contact with Scandinavians; the Kentish had to put up with invasions but not with settlements. The loss of final -n is also a characteristic of Scandinavian, and without these early

68

CYNTHIA ALLEN

records it would be tempting to attribute this change in the north to Scandinavian settlement. However, the fact that the change was clearly already in progress before the Scandinavian invasions could have affected the language shows that this conclusion would be false. It is likely that the Scandinavian settlers in the north reinforced this variant in northern English and contributed to its spread, but they did not introduce it into English. Because of the brevity of the eNbr texts, we cannot draw any confident conclusions about the existence of variants to the traditional case forms. But INbr records are much more extensive. Unfortunately, they are limited to interlinear glosses of Latin, which can only be used with cautious interpretation, as it is always possible that a given form has been affected by the Latin. Nevertheless, it is clear that by the mid or late tenth century the amount of syncretism of forms in the Northumbrian area had increased markedly. The most extensive study of INbr forms is Ross' (1937) Studies in the accidence of the Lindisfarne Gospels. Ross (1936) looks in more detail at the gender forms in the Lindisfarne Gospels (LG), and the interaction of case and gender in LG (and some other late OE and ME texts) is studied by Jones (1967a, 1967b, 1988). The question of how the 'unhistoricaF gender forms of LG (and other texts) are to be explained is a complex and controversial one,5 but fortunately the status of the case categories is quite clear. The picture that emerges from these studies is one of syncretism of forms, but retention of the OE case-marking categories. We can divide the changes to nominal inflection into two broad categories: reduction in the number of inflectional classes and case syncretism. Let us now consider the nature of these changes. 3.1 Reduction of inflectional classes The tendency to treat all inflectional classes of nouns in the same way was considerably more marked in LG than in eNbr or late West Saxon. Most strikingly, the genitive singular ending -es of the masculine and neuter -a stems was extended to nearly all classes of nouns, including the old weak nouns (Ross, 1936: 99). Ross comments that the -es form must be considered the normal genitive singular ending for most nouns, although there are exceptions and the older forms are also frequently found. It is not hard to find an explanation for this change: as has often been noted, -es was the only distinctively genitive ending and was furthermore the ending found in two large classes of nouns, and so it is unsurprising that it should have been extended to nouns of other classes. Indeed, this extension was already underway by the time of the earliest OE texts (in any dialect), with masculine and neuter strong nouns of all types tending to go over to the -a stem declension. The generalization of the genitive to -es of course took place in all dialects of
5

In all of the works listed above, Jones attacks traditional explanations for these unhistorical forms and argues that the apparently aberrant forms were in fact highly systematic. The idea is that certain forms now tended to be used to mark case alone, regardless of gender, whereas historically they marked both case and gender. For support of this basic idea, see Mitchell (1985: section 68) and Roberts (1970).

CASE LOSS AND THE ' C R E O L I Z A T I O N ' H Y P O T H E S I S

69

English at a later date, and it should be noted that up until at least the very end of the fourteenth century this genitive marker was a true inflection, rather than a clitic as it is in Modern English. It is not until the writings of Chaucer (d. 1400) that we find examples like The grete god of Loves name (Ch. House of fame 1489), indicating that the genitive is no longer an inflection of the noun, but rather a clitic on the NP. 6 This retention of the genitive inflection is not a characteristic that we expect of a Creole and will be discussed in greater detail in section 7. Syncretism is also found, but to a lesser degree, in the extension of the -as plural to noun classes which did not have it earlier. Ross (1937: 100) notes that this extension happened in nearly all classes, but here the extension into the 'wrong' classes was sporadic, not regular. What has happened here is that the case categories have been kept intact, but the forms which express these categories have been extended to nouns which historically used a different form for the same purpose. Such regularizations result in a blurring of distinctions among noun classes, which must result in even further regularization, eventually leading to the disappearance of inflectional classes altogether. In LG the old a-stems were clearly maintained as a separate inflectional class, because if they were not we would not find that their plural always ends in -as (or a phonological variant) rather than in an ending appropriate to another declension, which would indicate that speakers had lost sight of the inflectional classes. But the old inflectional classes were clearly on their way out. One phonological change which severely weakened the distinction between these classes was the neutralization of unstressed vowels. Ross (1937: 53) reached the conclusion that all these unstressed vowels had neutralized to one, and more recently Hogg (1992: section 6.62) expresses the opinion that merger of unstressed front and back vowels is demonstrated in LG.1 This change by itself created a good deal of syncretism but in LG this vowel reduction combines with another change, the loss of final -n, to have devastating effect on the declension of nouns. Hogg (1992: section 7.98) notes that this loss of -n was apparently restricted to certain morphosyntactic categories, but in the endings of the weak nouns it was completely regular. This meant that the old weak nouns were hard to distinguish from other nouns which had a vowel in the nominative/accusative singular, and Ross (1937: 100) points out that the scribe of
6

This change seems to have taken place in the following way. First, the genitive suffix gradually generalized to -{e)s for all word classes and to the plural as well as the singular. Until it became the nearly universal genitive suffix, it remained a suffix, but once it had ousted all its rivals it was reanalysed as a clitic. However, Ross argues in chapter 3 that the scribe clearly had a strong sense of the etymological value of these vowels. To some extent, this knowledge of what the historically correct vowel was probably came from the continued existence of these forms in the speech of older speakers, but it seems to me that some of the conservative spellings could only be the result of training in an orthographical tradition which survived the Scandinavian invasions. I know of no evidence to support Poussa's (1982: 78) comment that 'literacy spread into the Danelaw from Wessex', although it certainly seems plausible enough that people like the southern-trained Archbishops of York of this period brought with them their own scribes (see Whitelock, 1981 [1965]: 215).

70

CYNTHIA A L L E N

LG apparently spoke a variety of English in which there was no longer in fact a distinction between the two types of nouns. 3.2 Case syncretism So far, the changes which I have mentioned have primarily to do with the replacement of case forms by other forms historically marking the same case, rather than with the replacement of a case form by a form formerly marking a different case. But the latter type of change is also in evidence in LG. First, the nominativeaccusative distinction had nearly disappeared from the nominal system, in both the singular and the plural, the continuation of a trend already evident by the earliest OE. This trend would have been strengthened by the loss of final -, which obliterated, for example, the difference between the nominative and accusative form of the masculine weak nouns, even without considering the effects of vowel neutralization. It is clear, however, that not all syncretism of this type was due to phonological processes; analogy clearly plays an important role, as it did earlier. Ross (1937: 120-1) notes that it is entirely natural that speakers should have eliminated the nominative-accusative distinction in the few places where it remained. It is interesting to note that the accusative -ne ending of the strong masculine adjective is sometimes replaced by the endingless nominative form even though this ending was quite distinctive and not subject to phonological erosion; this phenomenon seems only explicable by a general tendency to eliminate the nominativeaccusative distinction. It is also interesting to note that in LG (as well as in other northern texts and in later East Midlands texts) this tendency was implemented in the feminine strong nouns by extending the unmarked nominative form to the accusative, which earlier had a vowel (identical to the vowel of the dative/genitive singular), while further south the tendency was to go in the other direction: the non-nominative vowel was extended to the nominative, making the feminine nouns essentially indeclinable in the singular. A difference with this geographical distribution is the sort of thing we might hope to explain in terms of Scandinavian influence, and one might look for such an explanation in the fact that the equivalent feminines in Old Norse were endingless in both the nominative and the accusative. But such an explanation would not account for everything, such as the fact that syncretism of these categories is also found in the definite determiner, where no direct Scandinavian influence could be the cause. Ross (1937: 120) notes that while in general the distinction between nominative and accusative is preserved in the definite determiner, there are some instances in which the form proper to the nominative is extended to the accusative. Before discussing the nature of this change, we must note that the nominative form of the masculine and feminine definite determiners had itself undergone analogical levelling, namely the replacement of s- by an interdental fricative (written either (p) or (d)), resulting in masculine de and feminine diu (or dio). Both the older forms and

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

71

the newer ones are used in LG.S As Ross (1937: 115) notes, this sort of replacement is similar to that found in other West Germanic languages, sometimes at a very early date. It is interesting to note that this change cannot be attributed, at least not directly, to Scandinavian influence, since the Old Norse forms were the same as the normal OE forms in having s-; the forms differed only in the vowel. Ross (1937: 120) states that the distinction between the nominative and the accusative definite article is generally well preserved, but 'there are a number of examples of the use of the nom. sg. for the ace. sg. and vice versa'. In fact, though, it seems9 that it is only with the feminines that there is a two-way interchange of forms; it appears that the masculine accusative form is not extended to nominative uses,10 and that the nature of the change with the masculines is that the nominative forms (both older and newer) are seen as an alternative for the accusative. In other words, it seems that de (and its alternate se) was a marker of either nominative or accusative case, while done marked only accusative case. With the feminines, on the other hand, the old nominative and accusative forms seem to be pretty much interchangeable, perhaps because the accusative feminine form had always been identical to the nominative and accusative plural for all genders, making its extension to the nominative singular easier. That is, deo and da both marked either nominative or accusative case. Note that the extension of pa to the nominative in fact contributes to the spread of pe as the general uninflected definite determiner,
8

10

Jones' (1988: 100) assertion that 'There are a mere half a dozen or so instances where we find (de)' in LG is at first difficult to understand, given that Cook (1894: 168) indicates that there are 205 examples of (fie) for the singular masculine nominative form of the determiner in the Gospel according to St Matthew alone. However, taken in context, it appears that what Jones' remark was meant to refer to was the scarcity of (de) in contexts where (se) would not have occurred etymologically. Given this interpretation, Jones' remark is essentially correct. Cook lists only three instances of (se) and twentyfour instances of (de) in the masculine accusative function in LG. My conclusion is based on Ross' examples and also on an examination of the entry for se in Cook (1894). Although Ross (1937: 120) notes the existence of twenty-seven examples of (fie) and three examples of (se) used as an accusative, he gives no examples of (done) in a nominative function. Cook notes twenty-four examples of (de) and three of (se) as masculine accusative forms but does not list (done) as a nominative form, compared with hundreds of examples of (variants of) this form in accusative functions (Ross gives the number as 451). These numbers are impossible to explain in terms of confusion. Jones (1988: 85) argues that pone and other historically masculine forms ending in -ne were starting to be used to signal an absolutive (i.e. nonagentive and nonpossessive) relationship, regardless of gender, but this is hardly satisfactory since the old accusative form seems never to be used systematically for intransitive subjects, either in LG or later in ME. However, it does appear that in some ME texts pone was used as a general object marker (see below), and Jones' suggestion that this is because it was more distinctive than other markers seems plausible. Jones also argues that the reason why pone was sometimes extended as a determiner to historically nonmasculine nouns in LG is that it distinctively signalled only one case, unlike the feminine form. This seems like a good explanation for some of the uses of the scribe, who had to contend with Latin word orders which were not usually (or ever) used in English and so had a particular need for unambiguous case marking, but it is very difficult to assess the extent to which the scribe's usage reflects his internalized system, and not simply a practice tailored to the needs of glossing Latin word for word. Unfortunately, as Jones himself is careful to point out, the nature of these glosses makes it impossible to be certain of what is happening with gender in the spoken language.

72

CYNTHIA ALLEN

because when pe and pa were unstressed they would have been pronounced the same because of vowel merger. The other case syncretism which we find in LG is between dative forms and nominative/accusative forms. Ross (1937) discusses this syncretism only briefly but Blakeley (1947) gives full details of the nominal forms found with five prepositions historically governing the dative case. Ross and Blakeley both conclude that syncretism was more advanced in the singular than in the plural, with Blakeley (1947: 30) noting that although the nominative/accusative form is found instead of the dative with nearly all classes of nouns in the singular, there is only one really convincing example in the plural. For our purposes the most important fact is that the case categories were clearly still quite distinct, although individual forms had begun to be used as variants to mark cases which would earlier not have been marked by those forms.
3.3 Evaluation: contact-induced simplification?

Many other examples of case syncretism and levelling of inflectional classes in LG could be given; for example Ross (1937: 105) notes the tendency to eliminate the weak forms of adjectives in favour of the (more distinctive) strong forms, and the entries for the possessives in Cook (1894) indicate that unmarked forms like din 'thy' are used in the dative (as well as the accusative) for all three genders, although historically the accusative and dative should have an ending for the masculine and feminine. Nevertheless, it is quite clear that the scribe was not 'confused' about the case categories of his language; the 'errors' that he made in this respect involve new deployment of forms but do not point to any uncertainty about the case categories of the language he was writing, unlike the errors of the scribe of the Peterborough Chronicle (see below). It is unfortunate that we do not have more texts from this area for centuries, and so are unable to trace the further development of syncretism in this dialect. However, it seems clear enough that the loss of case-marking distinctions in the north did not begin as an abrupt shift towards the use of uninflected forms, which is the sort of thing we expect of a Creole. Although the use of de in LG as an alternative to done at first looks like the introduction of an uninflected determiner, in fact it turns out to be a manifestation of the tendency to use nominative forms for accusative uses. Of course, we cannot ignore the possibility that these glosses are a poor reflection of the status of case categories in the spoken language. But speculation about what was happening in speech is of little value when the hypothetical processes leave no trace in the written record. There are two basic possibilities here: (a) the scribe's own internalized case-marking system was significantly different from that reflected in his work, i.e. he had learned as part of his training a linguistic system which was not that of his first language; or (b) the scribe's work essentially reflects the spoken language as far as case categories go. If (a) is correct,

CASE LOSS AND THE ' C R E O L I Z A T I O N ' HYPOTHESIS

73

then the scribe learned the literary system very well, because he made no mistakes which suggest that he did not perfectly command the four-way case system of late OE. But if he learned the classical system so well as a second language, it is surprising that he did not adhere more closely to the classical forms. On the other hand, the use of the case markers in LG is easy to reconcile with assumption (b) because this use is so systematic. This is not to suggest that the scribe used the case forms in precisely the same way in speech and in writing; it is likely that the more innovative forms which occur only rarely in these glosses occurred more frequently in informal speech and that he used some innovative forms in speech which are not found in his writing. But this is a difference of frequency of choice of variants, not a difference of case-marking categories, and at any rate there is certainly no need to assume a big difference between the forms used in the written language and those used in formal and careful speech. Let us assume, then, that in terms of case marking LG is a reasonable representation of the system of a register of the spoken language. If so, then it seems clear that a century of contact with Danish in this region had not caused the extreme simplification of case categories typical of a Creole. On the other hand, it is probably no coincidence that case-marking reduction proceeded fastest in the area with the most contact with a closely related language. Haugen (1976: 285) notes that the Scandinavian dialects which show the greatest simplification of inflectional categories in the Middle Scandinavian period are generally those with the most contact with other languages (especially Low German). It seems that mere regular contact (not necessarily intimate in nature) is sufficient to lead to quite significant simplification. We can explain this phenomenon in the following way. There is always variation in language between more innovative and more conservative forms, and every generation of language learners attempts to make language more regular (e.g. English children attempt to make past tenses of all verbs regular). Under stable social conditions, the innovations usually do not gain ground rapidly because they are not accepted by older members of the community. But innovations spread faster in times of social upheaval generally. In particular, when a new group speaking a different language enters the community and begins to learn the local language, they introduce new variants into the mix. It does not seem unlikely that the mere existence of a large number of variants in the speech community would lead (under the right social conditions) to a greater expectation that language would be variable and a quicker acceptance of innovative forms. For example, once nominative-accusative syncretism became widespread (i.e. by the OE stage) children learning the language presumably would have had a tendency to replace the accusative form with the nominative form. But the fact that the community as a whole had only one norm for the accusative forms would have resulted in considerable pressure upon language learners to conform to the norms. But in a tenth-century community which included Danish speakers who had learned English as a second language there would not have been one set of norms that all speakers shared, and speakers would know that not everyone used the same forms.

74

CYNTHIA ALLEN

It seems plausible that one outcome of such a situation would be the more rapid acceptance of innovative forms. So although the case-marking facts lend no support for the idea that inflectional endings were simply stripped off when Danish met English or that the combination led to confusion, it does appear that contact with Scandinavian probably accelerated the propagation of internally motivated innovations. I have dwelt on the situation in LG at some length because it has important implications for our interpretation of contact as a cause of case-marking simplification in all English dialects. Nbr shows rather advanced simplification at an early date, and since the simplification seems to have proceeded from north to south, the question of whether the earliest simplification was the result of contact or not is important. If the evidence showed that the early simplification found in LG and other Nbr texts was best explained as the result of creolization, it could be argued that the later simplification in the southern dialects was originally due to this creolization, with the southern dialects affected by contact with Anglo-Danish Creole speakers. But given that the role of contact seems to have been merely to speed up the rate of change, there is no reason to assume that any Anglo-Danish Creole played any role in the development of English case marking. Before moving on to later developments in regions further south, it is worth emphasizing that a striking similarity between Danish and English is not in itself evidence of contact-induced change. Let us consider a feature outside the casemarking system which might seem like evidence of Scandinavian influence if we did not have evidence to the contrary, namely the use of'stranded' prepositions, e.g. The houses that we looked at were all expensive. If we did not have early records of English which showed this to be a regular feature of the oldest recorded English, we might assume Scandinavian influence here. But in fact preposition stranding is the rule in certain constructions in every OE text, most of which are in the West Saxon dialect, which was the dialect least likely to be affected by Scandinavian influence (for a discussion, see Allen, 1980 [1977], especially pp. 232-4). Such phenomena do not argue against creolization, but they show that caution is necessary when assessing the role of contact in the history of English. 4 The situation in the late eleventh century In contrast to the north, there are a reasonable number of texts from further south near the end of the eleventh century which tell us a good deal about what was happening to the case-marking system in some areas. These texts have to be interpreted cautiously because the political ascendancy of Wessex meant that West Saxon orthographic traditions held sway in non-West Saxon areas; furthermore, errors in the texts make it clear that certain spellings represented old pronunciations rather than current ones. As Moore (1928: 229) comments, it is unlikely that any of these texts reflect the speech of the scribes very accurately. Nevertheless, the occasional innovating forms of these texts give some important clues about what was happening in the spoken language.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

75

One change which is apparent is the shift /m/ > /n/ in unstressed final syllables,11 discussed in detail by Moore, who concludes that this change was completed by the end of the eleventh century, although conservative scribal practices obscure the situation. This change on its own caused considerable syncretism, especially when combined with vowel reduction, but this syncretism became massive when it was quickly followed by the loss of -n in unstressed final syllables.12 As we have already seen, the loss of -n had already taken place by the tenth century in certain morphological contexts in INbr. One of the many effects of this change was that it greatly reduced the evidence for any class of noun which had a different form in the nominative and accusative singular. In earlier OE, the distinction had been maintained in the weak masculine and feminine nouns by the final -n of the accusative versus the endingless nominative. After the loss of -n, this distinction was found only in the feminine strong nouns with long stems.13 This remnant of the nominative-accusative nominal distinction was lost when the -e of the non-nominative singular forms of these feminine nouns was extended to the nominative by the end of the eleventh century. Moore's suggestion that this analogical levelling succeeded because it was difficult to maintain the nominative-accusative distinction once such a small group of nouns signalled it seems reasonable. It should be noted that the loss offinal-n proceeded in a different way in different parts of the country and does not seem to have been a purely phonological change. For example, a striking difference between the Ormulum and INbr is that in the former the final -n of verbal forms was very resistant to loss. However, it appears that weak nouns and adjectives were subject to this change in all dialects, perhaps because the final -n in these paradigms already played such a limited role in OE in marking case. Whether the loss of final -n in a given dialect was a mainly phonological or mainly analogical change, the important fact is that once the change took place one of the few remaining markers of the nominative-accusative distinction had disappeared, and so the loss of final -n would have aided in the loss of the distinction elsewhere. The levelling found in the nominal system did not immediately result in the loss of the nominative-accusative opposition as a category distinction, since the determiners continued to be inflected for this distinction. However, an important change had taken place which was to combine with the phonological changes to result in an
11

12

13

In fact, -n was already spreading into the dative plural in early West Saxon texts (see Campbell, 1959: section 378, and Hogg, 1992: section 7.102, and the references cited in these works). Hogg suggests that the earlier change was an analogical spread of the weak inflection -an rather than the start of a phonological change. These two changes also had important syncretistic effects on adjectival inflection. Space does not permit the discussion of this reduction of inflection, which proceeded in rather different ways in different parts of the country. It is sufficient to note that after vowel reduction and loss of -n had taken place, only -ne remained an unambiguous marker of both case and gender, with the ending spelled -re still being somewhat distinctive, signalling either feminine genitive/dative or plural genitive. Presumably, this drastic reduction in contrasts resulted in a situation in which language learners tended to treat adjectives as unmarked for case. The distinction was no longer maintained in the short-stemmed feminines after vowel reduction.

76

C Y N T H I A ALLEN

uninflected determiner which was an option (if not the only possibility) in all dialects by the beginning of the thirteenth century, viz. the levelling of s- to p-, which Brunner (1963: section 56) says was accomplished 'by the end of the OE period' (i.e. by 1100). The conservatism of the standard orthography makes it difficult to be certain of just how advanced syncretism was in the determiner system in the eleventh century, but one thing which should be noted is that when unstressed, the reflexes of both the old pa (feminine accusative singular or general nominative/accusative plural) and pam (dative masculine/neuter singular and general dative plural) would have been indistinguishable in pronunciation from the reflex of the old se, due to vowel reduction and the loss of final -n (< m). A language learner who heard pe as the form used for so many case and gender combinations might well treat it as an uninflected form which could be used as an alternative to the unambiguously inflected forms which were still used.14 Thus a new variant (the uninflected determiner) would be introduced into the language, a variant which quickly gained ground in the next century. The rapidity of the loss of the inflected determiner is unsurprising when we consider that once a language is left with so few forms that unambiguously signal case and gender, these categories are likely to be abandoned. A case-marking system is little more than an encumbrance in a language when the forms no longer clearly signal any information which aids in understanding the message. Speakers started to rely more on word order to signal relationships like subject and object, and case inflection became optional. At this point it is useful to compare English with Modern (standard) German, which has lost nominal case inflection almost to the same extent as English, but has nevertheless maintained a system of four cases and three genders, which are mainly marked on the modifiers of the noun rather than on the noun itself. It would seem that the major difference in the development of the two languages has been that in English, the two processes discussed above which were at least partially phonologically motivated resulted in so much syncretism in the modifiers that it was no longer worthwhile to maintain the few remaining distinctive forms, such as pone. In German, however, enough distinctive markers remained to make it reasonably easy to maintain a sensible set of oppositions. This is not to suggest that phonological factors alone were responsible for this syncretism, but it seems clear enough that the phonological tendencies of English supported the analogical tendencies and speeded up the loss of some category distinctions.
14

The idea that uninflected determiners might have been introduced into the language by French speakers learning English incompletely and substituting the nominative masculine form for the others does not seem a priori implausible. A language learner would then learn pe as a variant which could be used in all situations (from French speakers) and the case-marking categories and inflected forms from English speakers. But although incomplete language learning on the part of foreigners could well have reinforced natural tendencies in the language, it would not explain either the pre-Norman-invasion extension of the nominative form to the accusative in Nbr (since the Scandinavian forms were too similar to allow incomplete language learning in this instance) or the similar development in the mainland Scandinavian languages, where there presumably was not much close contact between language learners and foreigners.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

77

To summarize, the case-marking system of English was highly syncretistic by the year 1100, and although the old category distinctions still remained, syncretism had reached such a level that it was probably inevitable that these distinctions would soon disappear. This syncretism was probably accelerated by the contact with Danish and French (and by the general social upheaval), but it cannot be directly attributed to this contact. Rather, the syncretism took place in a step-by-step fashion and was caused by phonological processes and familiar sorts of levelling.15 5 The twelfth century 5.7 The Peterborough Chronicle

It is unfortunate that we have so few trustworthy English texts from the twelfth century, the time when so many changes to the case-marking system of English become obvious. There is in fact no dearth of manuscripts written in English in this century, but most of them contain copies of OE texts, which cannot be considered accurate reflections of the language of any one period and which can be used only with extreme caution. It is also unfortunate that the one extensive original text that we have from this period, the Peterborough Chronicle (PC), comes from an area in which case category distinctions were lost early and from which we have no earlier texts which would give us some idea of how rapidly this loss took place and how it was implemented. However, the PC does give us some hints. This text has three basic divisions: copied material up to the year 1121, the 'First Continuation', covering the years 1122-31 and written by one scribe, probably at intervals (Clark, 1970: xxv, and Ker, 1957: article 346), and the 'Final Continuation' of 1132-54, written in block by a second scribe, presumably at about 1155. The handwriting of the copied annals shows that they were copied by the scribe of the First Continuation; in addition, this scribe seems to have written the Interpolations which are interspersed in the copied material (Clark, 1970: xvii). The First Continuation scribe seems to have been a very faithful copyist, because comparisons of the language of the copied annals with the language found in another copy of the AngloSaxon Chronicle which is based on the same 'Northern Recension' of the chronicle16 indicated that for the most part the morphological 'errors' found in the early part of the PC were not introduced by the PC scribe but must have been in the common archetype, since they are also found in D. Furthermore, there is a striking and consistent difference between the language of the copied annals and that of the First
15

16

The standard view is that phonetic weakening preceded analogical reshaping, but Kitson (1990), on the basis of an examination of spellings in various OE and EME texts, concludes (p. 82) that 'generally morphology led phonology' and also warns that we should not assume that the order of events was the same in all dialects. For our purposes, it does not really matter whether phonological weakening preceded levelling. MS Cotton Tiberius B. iv, known as MS D and edited by Classen and Harmer (1926). The PC is based on the archetype of D until 1031, at which point the ancestor of the PC went south and ceases to be a northern version (Whitelock, 1961: xvi).

78

CYNTHIA ALLEN

Continuations and the Interpolations which is best explained by assuming that the scribe stuck to his exemplar closely (Clark, 1970: xlii). This means that we can be confident in assuming that the morphological innovations of the copied annals of the late eleventh century in general represent forms that were in the exemplar which was used by the scribe, and were not introduced by him. Unfortunately, as Clark (1970: xlii) notes, it is impossible to be certain of the dialect of the copied annals, other than that it was 'southern'. But we can see in these annals evidence of certain changes that were taking place in the late eleventh and early twelfth centuries in the language of the scribes that compiled the exemplar. This evidence is important, because some of it is lacking in other late eleventh-century texts, which are mostly copies of earlier texts and are therefore more conservative. In particular, these copied annals make it quite clear that even in some southerly areas the changes to the determiner system went beyond the levelling to p-, since we find, for example, into pe mynstre in the entry for 1070 (Clark, 1970: 2.26) where we would expect pam. Since the inflected forms of the determiners appear (for the most part)17 only in their historically correct uses, and since this usage contrasts markedly with the use of the determiner forms in the annals composed by the scribe, we must assume that these uninflected determiners reflect the use of the exemplar, rather than a change by the scribe. That is, the uninflected determiner existed as a variant in the second half of the eleventh century even in some areas fairly far south. Turning to the First Continuation, wefindgreat changes to the case-marking (and gender) system, showing how much further advanced the Peterborough area was in these developments than areas further south. For example, in the copied annals the dativeaccusative distinction is still preserved in the pronouns, since we still find regular use of hi (feminine accusative singular and accusative plural) and hine (masculine accusative singular). But in the First Continuation, hire (the old dative/ genitive form) has completely ousted hi as an object form, while heom (an old dative form) is now the only form found for any plural object. Remnants of the dativeaccusative distinction are found only in the masculine singular; although the scribe frequently uses him in old accusative functions (i.e. as direct object or the object of certain prepositions) we find sixteen examples of hine in the First Continuation, and in all but two of these examples, it is found in the historically correct accusative function. Clearly, the scribe did not simply regard hine and him as variant forms for the same category. But the fact that the scribe made two errors18 here (in contrast to the lack of similar errors in the early thirteenth-century manuscripts of the southeast; see below) makes it clear that while he had a basic idea of the difference between him and hine, he probably learned the distinction as a conscious rule, rather than part of
17

18

A regular exception is the use of seo (historically feminine) instead of the masculine se in many places. It seems that the scribe considered the two simply variants of each other. These two examples are unetymological or 'mistakes' because hine refers to the indirect object of a ditransitive verb, e.g. & iafhine fane eorldom 'and gave him the earldom' (1127 entry). In contrast, the number of ditransitive examples like this in which him is (correctly) used in this Continuation is over seventy.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

79

his internalized grammar. That is, the scribe was probably in the same position as many Modern English speakers are where the who-whom distinction is concerned. So it seems likely that in the everyday language of the scribe the pronouns showed only a three-way distinction (subject-genitive-object). When we look at the determiners, we at first get the impression that case and gender inflection was still very much alive. But a closer look soon makes it apparent that whatever system the scribe was using, it was not the system of the annals he had copied. Lumsden (1987) argues that the determiner forms of the First Continuation represent oppositions which existed at the time, with a different alignment of forms and grammatical features from OE. I do not find Lumsden's arguments convincing; one major problem is that his proposed system does not account for a rather large number of counterexamples. Jones (1988: 143-56) also argues that the situation is more systematic than has usually been supposed, and it is certainly clear enough that the forms are not simply used randomly. It seems to me that the best explanation for the data is to assume that in the scribe's ordinary language the uninflected pe was a possible variant for all cases and of the inflected forms only pone and pes still had a place in the scribe's own speech. Pes retained its old value (genitive) but pone, the most distinctive of the object markers, had become a general object marker, although some deviations from this rule suggest that it was a rather archaic form which the scribe did not completely command. I think (along with Clark, 1970: lxi), that the deviations from this usage found in the First Continuation are most plausibly explained as the result of hypercorrection on the part of the scribe, who had copied the older annals and would almost certainly have considered the older language to be superior to his own (or else he would not have copied it so carefully). Hypercorrection can be invoked only with caution, and only when some reasonable sort of mechanism can be suggested, but in this case I believe that we can make an explanation work. Space does not allow me to go into the details of how hypercorrection could explain the distribution of forms here, but see Allen (1995: 172-7) for a fuller discussion. Essentially, we must assume that some older speakers in the community still used se(o) and the scribe was aware that older speakers frequently used this form where he would use only pe. Therefore he would sometimes use se(o) where pe would have in fact been a 'correct' form (e.g. for the reflex of pam). He would also extend pone occasionally to nonobjects because the scribe, in his awareness that the older language often had pone where pe would be most natural for him, would sometimes forget that this was an object marker and would extend it to other situations in which he would have used pe. Note that with the introduction of the uninflected determiner as an option, grammatical gender was doomed because this was the class of words which most clearly marked this category by the end of the OE period. A final change of interest concerns nouns. The not infrequent use of the suffix -e gives the impression that a dative-accusative distinction still exists for nouns, although marking this distinction was optional. But closer examination shows that this has become a marker of nouns which are the objects of prepositions (any

80

CYNTHIA ALLEN

preposition). Thus the possibility of having this marker is now predictable purely by structure, rather than something which was governed by particular verbs and prepositions. This marker of the new prepositional case is still found in the Final Continuation, but less frequently. The uninflected determiner has completely replaced all inflections, and the dative-accusative distinction has entirely disappeared, leaving this dialect with a case-marking system very similar to that of Modern English with its three-way system in the pronouns, the major remaining difference being that in the language of the Final Continuation nouns could still be inflected for genitive case, while in Modern English the possessive marker is no longer an inflection of the noun but a clitic which attaches to the end of the NP. To summarize, by 1122 the dative-accusative distinction had only a very tenuous hold in the language of Peterborough, and the fact that the uninflected determiner was always available as an alternative to the older inflected forms meant that the old case-marking system was doomed. By 1155, these changes were very nearly complete. Although the differences between the First and Final Continuations might give the impression of very rapid change in this period, to a great extent the changes have more to do with scribal practice than with the linguistic systems, and the systemic changes which took place are not in fact more than we would expect would be able to be accomplished by one generation of language learners. The fact that the breakdown of the old system was so advanced by 1121 gives the impression of rapid change, but unfortunately we cannot tell how quickly these changes took place because we do not know how advanced the syncretism of case marking was in (say) 1000. But it is reasonable to assume that it was intermediate between the advanced syncretism of Northumbria and the conservatism of the south; for example, Hogg (1992: section 7.98) notes that the (north) Mercian Rushworth glosses show final -n loss but not to the same extent as in Nbr. If we assume that this -n loss which was responsible for so much syncretism was also found in Peterborough at a fairly early stage, then the rate of change does not seem so rapid. It is not unlikely that the contact with Danish played a significant role here in creating a social situation in which variants would be adopted and spread quickly, but it is important to realize that origins for all these changes can be found in purely language-internal factors. It should also be realized that Danish itself has undergone similar changes, starting at a later date but being accomplished rather quickly. Gordon (1957: 265) states that until c. 1000 dialect differences within Scandinavian were slight. And although various (mainly phonological) differences were evident by that date, it does not seem that any loss of case distinctions was involved, yet Haugen (1976: 208-9) notes that by 1350 (among other significant changes) the genitive singular -s had generalized to all nouns in some dialects of Danish. Thus the same change as took place in English took place independently and apparently rather quickly. Contact may have played a role here too, but it was not the intimate contact of the Danelaw.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

81

5.2

Southerly manuscripts

Most of the manuscripts of the twelfth century written in non-Danelaw areas are either known or suspected copies of earlier work and are therefore of limited usefulness. Nevertheless we can clearly see some forms encroaching on the territory of others. For example, in all these manuscripts the dative -e is essentially optional. Adjectival inflection is also greatly reduced. Syncretism in the pronouns varies with dialect. For example, in MS Lambeth 487, dated 1185-1225 and written in a combination of the West Midlands dialect known as 'AB' and some other dialect,19 we find that historically dative forms are frequently used for the objects of highly transitive verbs, where we would expect an accusative. The accusative forms are also still used, always in the historically correct functions. It is interesting to compare these copied homilies with an original work composed perhaps c. 1175, the Poema Morale, contained in the same manuscript. This work contains similar usage of the dative and accusative pronominal forms, but in contrast to the copied homilies the determiners rarely show any inflection, indicating that syncretism was far advanced in the language of the scribe in the determiner system but not in the pronominal system. What is of most importance is that in the manuscripts20 which contain syncretism in the pronouns, the substitution only goes one way: for example, him is substituted for hine but not vice versa.21 In other words, these scribes (unlike the PC scribe) seem to have had the accusative-dative distinction as a category distinction in their own language or else they would have made some mistakes in the direction of hypercorrection. The difference with OE is that while the category dative masculine singular is represented only by him (as in the earlier language) the masculine accusative singular is represented by both hine and him. In MS Cotton Vespasian D. xiv we find little evidence of syncretism in the pronominal system, even in the plural or the accusative feminine, which show the most replacement by dative forms in the other texts. The determiners were also still fully inflected. In this manuscript there are a couple of short pieces which are not OE copies (Warner's (1917) texts 43 and 45-6) and were probably composed c. 1125. In these pieces we find the same use of pronominal and determiner inflection as in the copied ones, showing that the author of the pieces had internalized the old system. This is hardly surprising since the manuscript is thought to have originated in Kent, where case-marking distinctions lasted longest. In short, in the southern areas the use of case marking is not confused, but systematic, even where it differed significantly from OE use. The fact that substantial syncretism without loss of category distinctions was taking place in literary texts of the south, where creolization with Scandinavian could not have played a role, is evidence that such creolization is not necessary to explain the situation in the north.
19 20

21

This is the opinion of Thompson (1958: xi, liii). For a discussion of the works I have examined for this period, see Allen (1995: section 5.3.1 and 478-9). For a possible (but not certain) exception see Allen (1995: 169, n. 8).

82

CYNTHIA ALLEN

It seems most probable that the same sort of processes which are evident from the southern texts are responsible for the more advanced syncretism of the northern texts, although in the north the changes started earlier and proceeded more rapidly. 6 The early thirteenth century
6.1 General

By the beginning of the thirteenth century, original texts become reasonably abundant, and they show a good deal about dialect differences. In the northeast, it is no surprise to find that the few remnants of the old case-marking system which are still found in the PC are nearly completely gone in the Ormulum of c. 1200 or c. 1180, written (according to Parkes, 1983: 127) in or near Bourne (16 miles from Peterborough). Clark (1970: xl) notes that the language of the Ormulum can be seen as a regularization of the tendencies evident in the PC, i.e. some old variants have disappeared in favour of invariant forms. It is interesting to note that the decay of case marking is well advanced in some West Midland dialects which could not have been directly affected by substantial contact with Scandinavian. For example, in the texts of the Katherine Group and the Ancrene Wisse, all written in the dialect usually referred to as 'AB',22 we find that there are only remnants of the dative-accusative distinction. The formal distinction had completely disappeared from the plural, neuter, and feminine pronouns but was still somewhat in evidence in the masculine singular in MS A (the Corpus Christi manuscript, which contains the Ancrene Wisse) but not MS B (Bodley 34), where only him is used. D'Ardenne's (1961) extensive grammatical introduction to her edition of the life of St Juliana remains a basic reference for any discussion of the language of not only this text but also the other AB texts. The Ancrene Wisse contained in MS A is definitely a copy of a somewhat earlier text, and d'Ardenne (1961: 222) plausibly suggests that the scribe who copied the text did not have hine in his own speech, but was copying the work of someone who did. It is not possible to look at the casemarking system of this dialect in any detail here; for more discussion of the casemarking system of AB, see d'Ardenne and also Allen (1995: 181-5). However, it is of considerable interest to note that d'Ardenne points out (1961: 204) that 'The accidence of AB is marked by conservatism in the verbal system contrasted with advanced simplification of the declension of nouns, and still more of adjectives.' If we assume that the reduced case marking in this dialect was due to creolization, it is hard to explain why verbal morphology escaped similar effects.

22

Dobson (1976: 117) notes that it is generally accepted that this dialect belongs either to northern Herefordshire or to southern Shropshire and argues (pp. 130ff.) that the Ancrene Wisse originated at Wigmore Abbey in northwestern Herefordshire. D'Ardenne and Dobson (1981: xxxviiiff.) indicate that the two manuscripts under discussion here (Cambridge, Corpus Christi College 402 and Bodley 34) can be dated around 1230.

CASE LOSS AND THE ' C R E O L I Z A T I O N ' HYPOTHESIS

83

6.2 The Vices and Virtues It is in the southeast of the country that the old case-marking system is best preserved, and this section will concentrate on the system found in one text, the Vices and Virtues (henceforth VV) of MS Stowe 34, edited by Holthausen (1888, glossary published 1920). According to Utley (1972: 702) the manuscript dates from c. 1200-25 and the text was possibly composed c. 1175-1225. The dialect is uncertain, but it is likely to be that of Essex. In this text, there is a good deal of syncretism of forms, but the old categories are clearly maintained (including gender categories). The process of reduction of inflectional classes is evident; for example, we find pinges used as the nominative plural form of the neuter noun 'thing' as well as the etymological ping. Syncretism of dative forms with nominative/accusative forms was advanced in the nouns, particularly in the plural. For example, for the masculine and neuter a-stems, the distinctively dative form continues to be -e in the singular, and is either -e or -en in the plural,23 but the old nominative/accusative form is found replacing the dative form in both the singular and the plural; e.g. we find both kyng and kynge Icing' in the dative singular and both dai^en and da^as for the dative plural of'day'. This syncretism of forms has led to a good deal of confused terminology, which is one reason why it is often impossible to get an accurate picture of the case-marking system of a particular text from studies not informed by linguistic theory. For example, in his glossary to this text Holthausen treats forms like kyng and da^as inconsistently when they occur in a function which historically demanded a dative case. He lists kyng as dative in one example, but gast as 'nom.ac' for the example of de hali gast 'of the holy ghost' (61.9). The inconsistency arises from uncertainty whether 'dative' should refer to form or to category, and a failure to keep the two concepts apart. In fact, Holthausen does appear to have followed a system whereby he treats such forms as dative only if they are modified by an unambiguously dative form, but this method is no help in revealing the system behind the forms. In what follows, I will use the label 'dative' to refer to a grammatical category, rather than to the forms. Thus kyng is dative when it occurs in functions which would demand a dative form, i.e. where the corresponding pronoun would be him, never hine, but accusative when hine would be possible. It is clear that the dative-accusative category distinction was very much a living distinction for this author. Particular monotransitive verbs still governed the dative case, while others governed the accusative, and still others allowed either case. It is interesting to note that the suffix -e still represents a real dative case, rather than the prepositional case of the PC continuations and AB, since this form is found on the recipients of ditransitive verbs even when they are not the object of a preposition. The formal dative-accusative distinction is still well maintained in the pronouns in this text. There is no hint of syncretism in the feminine pronouns, but him is used as
23

Both plural endings are reflexes of the old -urn; the variation comes from the dropping of -n or not. The reflex of the nom./acc. form was much more common than the reflexes of the dative form in the plural.

84

CYNTHIA ALLEN

a variant of the masculine accusative singular and sometimes hem is used for the accusative plural in addition to its usual use for the dative. This is rather interesting because in most dialects it was the masculines which resisted syncretism the longest. One change to the case-marking system which should be noted is that all prepositions now seem to take dative objects.24 In the determiner system, we find that the uninflected determiner was always an alternative to inflected forms, with one interesting exception: it seems that genitive determiners were always inflected, never pe.25 We find a bewildering assortment of determiner forms in this text and most of the forms represent more than one combination of case, gender, and number, but when we find that certain forms are never found with nouns of certain combinations of grammatical categories, it becomes evident that the variety of forms is not evidence of confusion on the part of the writer concerning the system of categories, but rather of variation in the forms used for particular categories. One question of interest because it bears on morphological theory is whether uninflected forms could be combined with inflected forms. For example, we might assume that a noun phrase which was marked for a particular case in any of its parts must be marked for the same case in all of its declinable parts. Then we would get the combinations (a) inflected determiner + inflected noun and (b) pe + uninflected noun but not (c) inflected determiner + uninflected noun or (d) pe + inflected noun (where by 'inflected' I mean 'inflected for case'). But in fact all four combinations are found. We get (a) in into dan echefiere 'into the eternal fire' (25.30), (b) in ofde hali gast 'of the holy ghost' (61.9), (c) in for dan ilche hlauerd 'for the same lord' (47.8) and (d) in a^eanes pe kinge 'against the king' (47.8). Our analysis of such examples will depend on the morphological and syntactic theory adopted, but the facts lend themselves well to an analysis whereby forms like gast are entered into the lexicon as + SG, GEN and pe is entered as GEN, while pene is + MASC, + SG, + ACC. Such an analysis accounts well for the behaviour of verbs which governed the dative case, such as helpen 'to help'. Such verbs could appear with objects with no overt case marking or with overtly dative objects, but not with overtly accusative objects. If we treat a form such as gast, which is not positively specified for a case, as being compatible with any case feature for which it is not negatively specified, the attested patterns are accounted for.26 But whatever analysis we give, it is clear that we are
24

25

26

I a m n o t certain w h e t h e r a n y preposition still allowed accusative objects. T h e r e is at least one example of a p r e p o s i t i o n a l object with the determiner dene, normally an accusative form, b u t it is possible t h a t this form was s o m e t i m e s e x t e n d e d t o the dative case. T h e shorter form den is certainly used as a dative. I a m a w a r e of o n e possible c o u n t e r e x a m p l e : at 31.19 we find all pa werldes wele 'all the weal of the w o r l d ' (where pa m a y either be regarded as agreeing with wele or as uninflected, representing the same p r o n u n c i a t i o n as pe). P e r h a p s werldes wele is better regarded as a c o m p o u n d similar to M o d e r n English women's wear, lady's maid, etc. In t h a t case, the determiner w o u l d n o t agree with the genitive. W h a t e v e r the c o r r e c t analysis of such examples, it is i m p o r t a n t to note t h a t non-agreement of a d e t e r m i n e r with a genitive modifier is already f o u n d in O E , as noted by Mitchell (1985: section 1318), a n d s h o u l d n o t be r e g a r d e d as a n M E innovation which resulted from confusion. N o t e a l s o t h a t u n d e r usual a s s u m p t i o n s dative case is treated as a 'lexical' case, i.e. assigned by a verb's lexical entry, while accusative case is a 'structural' case assigned to objects by general syntactic rules if

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

85

dealing with a highly regular system despite the variation of forms. The appearance of confusion is superficial and the systematic nature of the variants gives no evidence for creolization. To sum up, VV supplies valuable evidence concerning the mechanisms involved in the loss of grammatical categories by offering a good look at the case-marking system in a dialect in which syncretism was well advanced but the old system of categories was nevertheless well maintained. It is clear that forms did not deteriorate because the categories were not learned by language learners; rather the categories were lost only when the forms had become too syncretistic to support them. If the syncretism of this dialect was due to contact, it would have been contact either with French (due to incomplete language learning) or with northern speakers who had only uninflected forms, although contact seems unnecessary as an explanation. But what is of most importance here is that the changes took place in a step-bystep and systematic fashion. They were not the result of abrupt simplification due to speakers of different languages or dialects attempting to communicate with each other. It should also be noted that none of the case-marking forms of this dialect is taken over directly from French,27 and that the retention of genitive case in particular goes strongly against the suggestion of Bailey & Maroldt (1977: 38) that ME was essentially relexified French. 7 Conclusion Only when we look at the history of English morphology in very broad outline does it seem attractive to compare the development of ME to the creation of a Creole. The impression of abrupt, drastic change is to a large extent due to changes in scribal practice in ME and to the fact that the earliest ME documents which are not copies of earlier compositions come from an area in which it is likely that syncretism was already well advanced in late OE, while our late OE documents come from more morphologically conservative areas. It is unfortunately not possible to trace the developments from late OE to late ME in any area through an unbroken succession of texts. But when we look at such details as are available for the systems of different periods, the picture we get is one of gradual and systematic (although sometimes rapid) change. There is no point at which we can say 'now there has been a radical break from the language of the preceding generation'. The
no lexical case has been assigned. This means that we cannot say that helpen only optionally called for a dative object; if it did, then when the object was not dative we would expect it to get assigned accusative case by the syntactic rules and so to allow overtly accusative objects as well as dative. Nevertheless, Bailey & Maroldt (1977: 49) claim that 'it is not unfair to say that the French forms prevailed' in the nominal declensions. This claim must be based mainly on the fact that the spread of -(e)s in the plural and of schwa make the ME declensions look more like the Old French declensions than the OE ones had. But as we have seen, these changes were the result of purely language-internal factors which had their roots in OE. They are also paralleled by changes in some Scandinavian languages, which did not have heavy French influence. Not a single nominal inflection is taken over directly from French.

27

86

CYNTHIA ALLEN

reduction of case-marking forms in English was not just a matter of substituting an uninflected form for the inflected ones. The evidence from the southerly texts of the early thirteenth century in particular shows that this reduction started out with syncretism of forms but retention of category distinctions. Particular forms began to be used as variants for categories which they did not historically represent. When the frequency of use of the new variant became high enough that the old variant disappeared, the category distinction disappeared also. What is of particular importance is that until the category distinction disappeared, the substitution of forms only went in one direction. We only get 'confusion' (i.e. forms being used as essentially interchangeable variants) when the category distinction has disappeared and writers hypercorrectly attempt to use a form which has disappeared from speech. Although creolization with Danish seems much more plausible than creolization with French, as far as the reduction of case categories > is concerned there is no evidence to suggest that the processes involved differed fundamentally in the north from those which are evident from the southern texts. The southern texts very clearly show that in this area category distinctions could be maintained despite substantial syncretism of forms, and that the loss of categories took place only after one form advanced into the area of another. The situation is less clear in the north, where texts are much more restricted. However, when due attention is paid to the distinction between form and category, the evidence suggests that the dialect of LG was just like the southern dialects in preserving systematic category distinctions represented by highly syncretized forms, although the details of syncretism differed considerably. There is therefore no need to appeal to creolization to explain the changes found in LG. All the changes can be seen as the result of natural languageinternal processes. We cannot call ME a Creole on the basis of its case-marking syncretism unless we are willing to call the mainland Scandinavian languages Creoles also. It must not be forgotten that ME contained a number of features which are not typical of a genuine Creole. One of these is an inflectional singular-plural distinction. Another particularly striking noncreole feature of ME was the retention of a genuine genitive case for nouns. Mufwene (1993) notes that pronominal possessive forms seem to be a common casualty of pidginization and creolization, although it is possible for a Creole to have possessive pronouns which are distinct in form from any other pronoun. While genitive pronouns are found in some Creoles, I am not aware that any creole has a possessive case for nouns. Instead, depending on the Creole, the possessor noun is normally either the object of a preposition, juxtaposed to the possessed noun without any inflection, or linked to the possessed noun by means of either an uninflected linker or a possessive pronoun, e.g. Pedro su amigu 'Pedro's friend' (example from Mufwene, 1993: 138). It could of course be suggested that the genitive inflection was lost from the spoken language during a period of 'creolization' and later reintroduced from the written language or as part of the hybridization of the East Midland creole and more southerly dialects which is proposed by Poussa (1982: 83), but there is not a shred of evidence that the genitive inflection was ever

CASE LOSS A N D T H E ' C R E O L I Z A T I O N ' H Y P O T H E S I S

87

endangered in any dialect. The retention of the genitive seems particularly incompatible with the idea of creolization with French,28 since in a French-based creole the genitive case would surely not have survived in the face of the prepositional genitive of French.29 It is notable that in the retention of a genitive case ME was less like a creole than is Modern Dutch, in which the use of the genitive inflection is extremely limited and a prepositional possessive is favoured. But to deny that ME was a creole is not to deny that contact played an important role in the reduction of case-marking categories. The findings of this study suggest that the role of this contact seems to have been mainly to hasten the acceptance and spread of naturally arising variants rather than to introduce new variants through faulty language learning. However, it is possible that incomplete language learning also introduced new variants. We cannot truly understand how languages can change their case-marking systems until we have good data about the stages involved. Such data can only be obtained by looking at a good deal of detail. But looking at details is not enough to compose an adequate picture of the system. Data-gathering must be informed by linguistic theory if we are to avoid the errors of the past, such as losing sight of the difference between form and category and thinking of variation as indicative of confusion. I hope in this paper to have provided some useful evidence bearing on the mechanics of how case-marking categories were lost in English, and so to have made a small contribution to our understanding of language change.
Author's address: Department of Linguistics Australian National University Arts Faculty Canberra ACT0200 Australia cindy.allen@anu.edu.au

References
Allen, C. L. (1980). Topics in diachronic English syntax. New York: Garland Press. (Slightly revised version of 1977 University of Massachusetts Ph.D. dissertation.) Allen, C. L. (1995). Case marking and reanalysis: grammatical relations from Old to Early Modern English. Oxford: Clarendon Press.
28

29

Bailey & Maroldt (1977: 45) note that 'the genitive singular remained for some time', but they offer n o explanation for how this could have been possible. Although Bailey & Maroldt (1977: 4 5 - 6 ) claim that the Modern English o/-'genitive' is 'certainly modelled in part on the French a dative', Mitchell (1985: section 1203) shows that o / a n d the genitive case already overlapped in some functions in O E , and that although there are n o really clear cases where o / m u s t be construed as denoting a relationship of pure possession in OE, nevertheless there are examples where of could have been interpreted in this way. Such examples form a natural model for the development of a true prepositional genitive (as has happened in other Germanic languages) and so the prepositional genitive in M E is no evidence for creolization, although French models may have assisted in its development in English.

88

CYNTHIA ALLEN

Aylmer, G. E. & R. Kant (1977). A history of York Minster. Oxford: Clarendon Press. Bailey, C.-J. & K. Maroldt (1977). The French lineage of English. In Meisel, J. (ed.), Langues en contact - pidgins - CreolesI Languages in contact. Tubingen: Narr. 21-53. Blakeley, L. (1947). Accusative-dative syncretism in the Lindisfarne Gospels. English and

Germanic Studies 1: 6-31. Brunner, K. (1963). An outline of Middle English grammar. Translation of the 5th edn of Abriss der mittelenglischen Grammatik (1962, Tubingen). Oxford: Blackwell. Campbell, A. (1959). Old English grammar. Oxford: Clarendon Press; repr. 1974. Clark, C. (1970). The Peterborough Chronicle 1070-1154. 2nd edn. Oxford: Clarendon Press. (1st edn 1957.) Classen, E. & F. E. Harmer (1926). An Anglo-Saxon chronicle. Manchester: Manchester University Press. Cook, A. S. (1894). A glossary of the Old Northumbrian gospels. Halle: Max Niemeyer; repr. Hildesheim: GeorgOlms, 1969. Dahl, I. (1938). Substantival inflexion in Early Old English. Lund: C. W. K. Gleerup. D'Ardenne, S. R. T. O. (1961). Pe liflade and te passiun ofSeinte Juliene. (Early English Text Society 248.) London: Oxford University Press. D'Ardenne, S. R. T. O. & E. J. Dobson (1981). Seinte Katerine (Early English Text Society supplementary series 7). London: Oxford University Press. Dobson, E. J. (1976). The origins of the 'Ancrene Wisse'. Oxford: Clarendon Press. Domingue, N. Z. (1977). Middle English: another Creole? Journal of Creole Studies 1: 89-100. Godden, M. (ed.) (1979). Ailfric's Catholic homilies: the second series. (Early English Text Society Supplementary Series 5.) London: Oxford University Press. Gordon, E. V. (1957). An introduction to Old Norse. 2nd edn, revised by A. R. Taylor. Oxford: Clarendon Press. Gorlach, M. (1990 [1986]). Middle English - a Creole? In Gorlach, M., Studies in the history of the English language. (Anglistische Forschungen 210.) Heidelberg: Winter. 65-78. Hart, C. (1992). The Danelaw. London: Hambledon Press. Haugen, E. (1976). The Scandinavian languages. London: Faber and Faber. Hogg, R. (1992). A grammar of Old English. Oxford: Blackwell. Holthausen, F. (ed.) (1888, 1920). Vices and Virtues. (Early English Text Society 89, 159.) London: N. Triibner. Jones, C. (1967a). The functional motivation of linguistic change: a study of the development of the grammatical category of gender in the Late Old English period. English Studies 48: 97-111. Jones, C. (1967b). The grammatical category of gender in Early Middle English. English Studies 48: 289-305. Jones, C. (1988). Grammatical gender in English: 950 to 1250. London: Croom Helm. Ker, N. R. (1957). Catalogue of manuscripts containing Anglo-Saxon. Oxford: Clarendon Press. King, A. (1986). The Ruthwell Cross - a linguistic monument (runes as evidence for Old English). Folia Linguistica Historica 7: 43-78. Kitson, P. (1990). Old English dialects and the stages of the transition to Middle English. Folia Linguistica Historica 11: 27-87. Lehmann, W. P. (1993). Theoretical bases of Indo-European linguistics. New York: Routledge. Lumsden, J. (1987). Syntactic features: parametric variation in the history of English. Unpublished Ph.D. thesis, MIT. Milroy, J. (1984). The history of English in the British Isles. In Trudgill, P. (ed.), Language in the British Isles. Cambridge: Cambridge University Press. 5-31. Mitchell, B. (1985). Old English syntax. Oxford: Clarendon Press.

CASE LOSS AND THE 'CREOLIZATION' HYPOTHESIS

89

Moore, S. (1928). Earliest morphological changes in Middle English. Language 4: 238-66. Mosse, F. (1952). Handbook of Middle English. Translated by J. A. Walker. Baltimore: Johns Hopkins Press. Mufwene, S. (1993). Are there possessive pronouns in Atlantic Creoles? In Byrne, F. & J. Holm (eds.), Atlantic meets Pacific: a global view of pidginization and creolization. Amsterdam: John Benjamins. 133-43 Mustanoja, T. (1960). A Middle English syntax. Helsinki: Societe Neophilologique. Parkes, M. B. (1983). On the presumed date and possible origin of the manuscript of the 'Orrmulum': Oxford, Bodleian Library, MS Junius 1. In Stanley, E. G. and D. Gray (eds.), Five hundred years of words and sounds: a festschrift for Eric Dobson. Cambridge: D. S. Brewer. 115-27. Poussa, P. (1982). The evolution of early standard English: the creolization hypothesis. Studia Anglica Posnaniensia 14: 69-85. Prokosch, E. (1939). A comparative Germanic grammar. Philadelphia: University of Pennsylvania; repr. 1960. Roberts, J. (1970). Traces of unhistorical gender congruence in a Late Old English manuscript. English Studies 51: 30-7. Ross, A. S. C. (1936). Sex and gender in the Lindisfarne Gospels. Journal of English and Germanic Philology 35: 321-30. Ross, A. S. C. (1937). Studies in the accidence of the Lindisfarne Gospels. Leeds: School of English Language, University of Leeds. Thomason, S. G. & T. Kaufman (1988). Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Thompson, W. M. (ed.) (1958). Pe wohunge ofure lauerd. (Early English Text Society, 241.) London: Oxford University Press. Utley, F. L. (1972). Dialogues, debates, and catechisms. In Hartung, A. E. (ed.), A manual of the writings in Middle English 1050-1500, vol. HI. New Haven: Connecticut Academy of Arts and Sciences. 669-745. Wallmannsberger, J. (1988). The 'creole hypothesis' in the history of English. In Markus, M. (ed.), Historical English: on the occasion of Karl Brunner's 100th birthday. (Innsbrucker Beitrage zur Kulturwissenschaft. Anglistische Reihe, 1.) Innsbruck: Institut fur Anglistik. 19-36. Warner, R. (ed.) (1917). Early English homilies from the twelfth century MS Vesp.D.xiv. (Early English Text Society, 152.) London: Kegan Paul, Trench, and Triibner. Whitelock, D. (1961). The Anglo-Saxon Chronicle: a revised translation. London: Eyre and Spottiswoode. Whitelock, D. (1981 [1965]). History, law and literature in lOth-llth century England. London: Variorum Reprints.

Das könnte Ihnen auch gefallen