Navigation – Plan du site

The grammar of later medieval French: an initial exploration of the Anglo Norman Dictionary textbase

Richard Ingham

Résumés

Dans cet article nous examinons la syntaxe de l’anglo-normand tardif, en confrontant l’hypothèse d’une « différence fondamentale » entre l’anglo-normand (AN) et le français du continent (Kibbee (1991), à celle de Trotter (2003), selon qui l’AN aurait participé au « continuum dialectal » francophone du Moyen Age. Il est proposé par la même occasion de démontrer la capacité de textes non-littéraires, comme le sont la plupart des textes AN tardifs, à nous renseigner quant à la datation d’évolutions en syntaxe. Une analyse de sources appartenant à la base de textes de l’Anglo-Norman Hub nous a permis de constater que la syntaxe du français insulaire s’est conformée à celle du français du continent sur les points étudiés, observant jusqu’en 1350 environ une distinction entre adverbes invertissants et non-invertissants, qui s’est ensuite perdue, et d’autre part en n’employant l’article partitif de qu’après point, plutôt qu’après mie ou pas. Ces résultats s’expliquent mieux si l’on suit l’hypothèse de l’adhésion de l’AN à l’aire francophone médiévale plutôt que celle d’une différence fondamentale entre le français insulaire et celui du continent. Il est proposé en outre que la base de textes de l’ Anglo-Norman Hub puisse servir de référence quant à la chronologie des changements syntaxiques survenus en français médiéval, en attendant d’être complétée par des sources continentales de genres comparables.

Haut de page

Texte intégral

Introduction

  • 1  As noted by Kabatek et al. (2005 : 4), the Romance research community experienced a relative delay (...)

1The diachronic study of language relies on the availability of at least approximately datable texts, and the study of diatopic variation in past states of language requires approximately localisable texts. At present the resources available under the heading of historical French language corpora are not ideally suited for either of these purposes. They consist predominantly of literary texts, which are often relatively difficult to date and localise. Research in the history of English and Spanish has been able to progress thanks to the availability of non-literary texts, allowing questions of the timing and geographical directionality of the processes of change to be pursued. Corpora now generally in electronic form have become available, permitting the exploration of these issues, especially regarding lexis and syntax, far more thoroughly than was possible heretofore1. The long association of the history of French with literary history almost inevitably meant that the sort of texts initially made available in searchable corpora would belong to literary genres. This had significant consequences as regards the earlier phases of the textual record. Datable and localisable literary texts are relatively uncommon during the mediaeval period, when many manuscripts were the product of rewriting, often in a dialect different from that of the original and at a time which may postdate the time of composition very substantially. It is unsurprising, therefore, that the study of medieval French has yet to produce research which considers the development across time and space of linguistic changes comparable with, e.g. the work of Herman on post-classical Latin (see especially Herman (1990)), of Kroch and others on medieval English (see e.g. Kroch & Taylor 2000; Kroch, Taylor & Ringe 2000), and Tuten (2005) and Harris-Northall (2005) on Old Spanish.

2In recent decades the research field has undergone a transformation as work on non-literary texts anchored to specific times and places has begun to bear fruit. Notable examples have been the work of Antonij Dees, and of Martin Gleßgen and his associates on Old French charters, research into Anglo-Norman by David Trotter, and Anthony Lodge’s historical sociolinguistic work on Parisian and other French varieties. However, until very recently this change in the evidential basis for descriptions of pre-modern French had not translated into available electronic versions of this kind of source material. In July 2007 the situation changed, when the textbase of the Anglo-Norman Dictionary project, directed by Trotter, was in large measure made available online in the form of browsable and searchable electronic texts, accessible on the Anglo-Norman Hub web site. The non-literary prose part of this textbase totals some 2 million words, not a large amount by the standards of today’s electronic corpora, but nevertheless sufficient to carry out investigation on the recurrent formal patterns and lexis of Anglo-Norman (also referred to below as insular French).

3The present study makes use of this textbase to explore changes in French syntax during the later 13th century and the 14th century, a period identified by earlier authors as the transition from old French to Middle French. Syntactic variation was found by Ingham (2007) using various published editions in a selection of 10 continental French non-literary texts of the late old French period, indicating the value of establishing a searchable corpus of such texts. Given the unavailability of a corpus of continental material of this type, we have explored how far the Anglo-Norman Hub textbase (henceforth ANHT) would provide a source of information on developments in late medieval French.

1. Source material

4The ANHT comprises about a hundred items from various periods and from various genres. It is organised into groups of texts according to genre and allows a selection to be made of the texts that the researcher desires to search. The section of the Anglo-Norman textbase which features legal, administrative and diplomatic texts was the main source of information for the purposes of the present enquiry. It was supplemented by a number of narrative prose texts notably chronicles. Only texts were used which could be unambiguously identified as composed in England: this determination had already been made by the compilers of the Anglo-Norman dictionary. We also needed texts with a fairly secure date as regards their composition: the advantage of non-literary text such as legal and administrative material is the frequent internal dating evidence. The texts that feature in the ANHT nonliterary sections, were mostly written in the 14th century, but cover a period stretching from the later 13th to the early 15th centuries. They are taken from a variety of published sources, which vary in length from the multivolume Foedera series (over 700,000 words) to a short correspondence collection such as Burton (c. 5,000 words). Our policy in this study was to use as many non-literary texts as possible, excluding only cases where the text cannot be dated to within a decade or two. We included all those in the ‘administrative, legal and correspondence’ subpart of the ANHT, in addition to prose chronicles whose approximate date of compilation is known, and a few miscellaneous prose works, the Mirour de seinte Eglise (c. 1250) the Livre de Seintz Medicinez (c. 1355) and the Manières de Langage of 1396 and 1415. The total word length of the material examined, contained in 36 texts, is slightly over 2 million words.

5Our selection from the textbase offers at least two unusual features by comparison with previous data-based investigations of medieval French. The first is that all the texts belong to the same regionally defined variety. Secondly, the texts can be identified with a particular time of composition and – in some cases – place of composition. These features are very attractive to historians of language, particularly those who need the potential to identify external factors of language change. The material therefore lends itself to testing linguistically and sociolinguistically derived hypotheses more readily than some other historical corpora where the texts that have been included in the corpus may well come from manuscripts that have been reworked by writers from more than one linguistic area and more than one period of time. Because it is possible to date so much of the material in the ANHT, we are in principle able to carry out quite precise analyses of language change within the overall time period covered by the available material, which in the case of our prose section ranges from the mid-13th century to around 1420.

6In common with certain other Old French electronic resources, e.g. the BFM and the LFA corpora, the ANHT is not grammatically tagged, having only lexical concordancing facilities. However, judicious use can be made of lexical search terms on a sampling basis in order to analyse syntax, since much of the time syntactic variation is lexically sensitive to the presence of lexical items, or classes of items. This is notably the case with negation and – in a verb-second language such as Old French – adverbial syntax, which are the domains to be studied here.

2. Rationale

  • 2  Though see Wilshere (1973).

7The particular area of linguistic enquiry we wished to investigate was that of syntax, which in Anglo-Norman studies has in the past received almost no attention2. Processes of syntactic change that were ongoing in continental French during this period of time have already received substantial attention from historical linguists (see e.g. Marchello-Nizia 1995, 1997; Vance 1993, 1997; Ingham in press), thus allowing us to interrogate the newly available textbase material as to whether corresponding changes took place in insular French. However, the evidential basis of much research into the transition from Old to Middle French, including all the publications just cited, consists heavily or even exclusively of literary texts, which are subject to the doubts to which we alluded above as to how well they represent the timing and geographical extension of language change. It is worthwhile, therefore, to carry out an investigation of non-literary texts in order to determine whether a different picture emerges of the sequences of syntactic change that one might discern. The ANHT offer one way of doing this, in the absence for the time being of a comparably rich continental database of non-literary material. The texts therein have the potential to inform us, for example, as to whether changes in syntactic form that are linked by theoretical linguistic analysis tended to cluster in time. Hypotheses made by Principles and Parameters syntactic theory that have already been tested on a number of other languages (see e.g. Kroch & Taylor 2000; Kroch, Taylor, & Ringe 2000) may be assessed for whether they provide an informative account of the sequencing of diachronic developments in French.

8Before any reliance can be placed on these texts as indicative of mainstream developments in the history of French, an important issue to address is the marginal status of Anglo-Norman (henceforth AN), as is often supposed. It is often assumed that AN texts present so many departures from Old French as used on the continent that it would be perilous to use them to draw any linguistic conclusions. The following citation typifies this view:

9It is clear from the kind of French that was being written in England in the thirteenth and, even more so, in the fourteenth century that the writers had less than total command of the language. What generally passed for French in England lacked any real roots in contemporary society and was indeed a language in an advanced state of decline. Grammatically, it was often little more than ‘bad French’ (Price 1984: 224).

10The ‘essential difference’ hypothesis of Kibbee (1991) claims that insular French, as a second language variety, departed in numerous ways from the mother-tongue French of France, notably in its syntactic constructions. Language contact researchers (e.g. Thomason & Kaufman 1988; Clyne 2003) draw attention to non-targetlike syntax where a language is acquired as an L2. The conventional assumption regarding AN is that by the C13 it was the mother-tongue of almost no-one in the population of England, and to that extent Kibbee’s hypothesis appears entirely plausible. Until very recently, very little research on the syntax of AN was available, so it was hard to tell whether it was empirically supported.

11In this study we have therefore sought to analyse the language of the ANHT texts from the point of view of conformity with continental syntactic norms, so as to assess how well the essential difference hypothesis performs on this data. We shall examine aspects of main clause syntax, the syntax of the partitive article in negative clauses, and syntactic patterns with particular adverbs. In these areas of grammar, continental Old French patterns in ways that were quite distinct from Middle English syntax. This makes it possible to identify of whether there is influence from English on the way Anglo-Norman constructed sentences using these elements.

12Trotter (2003) rejected the view that AN was simply ‘bad French’, pointing to the overwhelming evidence of its viability as a medium of communication within England and for purposes of international relations. He noted that the oft-quoted allusions to differences between AN and Central French do not tell us that AN was moribund, any more than do the extensive differences between central French and the French of Lorraine. He thus favoured an interpretation of AN whereby it formed a branch of the medieval French dialect continuum. Studies of pronoun syntax and word order by Ingham (2006a, b) seem to offer support for the hypothesis that AN was structurally well-integrated into that continuum, and we now seek to extend these findings by addressing the larger body of texts now made available thanks to the ANHT.

3. Research variables

13In order to assess the viability of the ANHT and to test the essential difference claim, we sought as research variables areas of medieval French grammar that drew formal distinctions lacking counterparts in Middle English. If the ANHT data showed that they were observed we would know that insular users were able to achieve a level of competence that could not be attributed to positive transfer from their mother tongue. However, the ANHT does not immediately lend itself to grammatical analysis, since it is not parsed and can therefore only serve as a research instrument if variables based on specific lexical items are selected. Accordingly, we chose a number of syntactic aspects amenable to a lexical search procedure, which are described in what follows.

14Wishing to investigate further some of the syntactic features of later medieval French examined by Zwanenburg (1978), Vance (1997) and Ingham (2005, 2006a), as regards Subject-Verb inversion in Old French, we focused on inverting and non-inverting adverbs. Foulet (1930) observed that V2 was not uniform after an initial adverb: with some, such as ore and encore it was the rule, whereas with others such as certes and neporquant it was not. According to Foulet, the adverb onques (‘never’) patterned with the latter pair, but as shown by Ingham (2005) the syntax of preverbal onques, while resembling them in that it did not permit inversion, differed from them to the extent that it did not permit the expression of a pronoun subject in main clauses, at least in the early C13 prose texts studied.

15Vance (1997) found that, as Old French gave way to Middle French, inversion began to be lost with certain time adverbs such as lors, and that around 1300 this process was beginning to affect other time adverbs, such as ore and encore.

16A further development identified by Vance concerned adverbs in preverbal position in an embedded clause, which could be accompanied by an overt subject pronoun in Old French. As noted by Ingham (2005), this applied in the case of onques, e.g.:

(1) Et certes se il onques le pensa, force d’amors li fist fere … (Artu, p. 5)

17Vance (1997) showed that in Middle French a change took place whereby the position of subject pronouns in embedded clauses shifted from immediately after the complementiser to a position immediately before the finite verb. Again, examples of this new trend can be found with onques, e.g.:

(2a) […] que onques je ne parlay a lui (Saintré 14, 7)

(2b) […] que onques il ne peurent eslongier Bristo plus hault de trois lieus (Frossart D 147)

18Our first two areas of enquiry, then, concerned how similar AN adverb syntax was to Old French in these various respects, and whether AN underwent change comparable to developments on the continent.

19The final research variable concerned negative particles and the use of the partitive article de. As discussed by Carlier (2007) the partitive article in medieval French negative clauses was limited to the context of the postverbal particle point, e.g.:

(3) Perceval, qui n’a point de cheval, car cil li avoient le suen ocis, le suit au plus tost (Queste p. 88)

20The negative particles mie and – at this time – pas apparently did not permit the partitive construction.

21As mentioned above, the ANHT allows for the time dimension to be included as an independent variable thanks to the dating evidence for the non-literary texts it comprises. We therefore included comparison across time periods in the design of the study. Of the six adverbs, two (ore and encore) fell into the category of the adverbs known to have undergone changes away from triggering Verb Second syntax in the C14. Likewise the use of an adverb separating a subject pronoun from a complementiser developed in the C14. For these features we therefore distinguished the period up to 1350 from that following 1350, and for consistency used the same division into two time periods for the investigation of the partitive article.

4. Methodology

22The same general procedure was used for each analysis. First, the ANHT was searched utilising the concordancer facility, for each of the target lexical items, using the alternative spelling forms for each item shown in the Anglo-Norman Dictionary. The resulting body of data was then searched by hand and divided into the appropriate analysis categories for that particular analysis. Hand-searching was required because, as mentioned above, no form of tagging is currently available within the textbase to guide the search. The dependent variable analysis categories were those determined by pre-existing linguistic analyses, referred to above for each separate subdomain of enquiry. In some cases a time variable was used, which required the texts to be assigned by date of composition to one or other time period. In most cases this was straightforward since we used only material whose date of composition had been independently established, and therefore all data from this source could be assigned to a given time period. The only texts which posed any practical difficulty in this respect were those such as the Foedera, which consist of compilations of many government treaty documents written over a relatively long time-span. Here we had to visit the full text version of each instance found by our concordance search, in order to identify the date of that particular case. The specifics of how data were obtained will be discussed more fully in the next section, in relation to the individual analysis concerned.

5. Analyses and results

Analysis 1: Subject-Verb inversion

23We begin with the results of the analysis of Subject-Verb syntax in clauses having an initial adverb. It will be recalled that in Old French these fell into two main types, inverting and non-inverting adverbs, together with a minor type that suppressed a pronoun subject when the adverb appeared preverbally in negative clauses. We selected two adverbs to represent each of these three cases. In each case, so as to ensure that our analysis identified the effect only of the particular adverb under study, we excluded all instances where another clause element also preceded the verb, e.g.:

(4) Ja par ceo ne sera oy de rien (Britton i 282)

24We did, however, include clauses introduced by the conjunctions et, ou and mais, as in Old Fench these items did not affect the syntax of verb-second with a preverbal adverb (Foulet 1930).

25To represent the inverting adverb type we selected ore and encore, as in Ingham (2005). Results on this type are displayed in Table 1, divided into whether the data were compliant with a verb-second grammar requiring inversion after an initial non-subject constituent (V2-compliant), or not compliant with it (non V2-compliant).

  • 3  Abbreviations: VSp = Verb-Subject pronoun order; VSn = Verb-Subject; VSo = Verb, no subjec

Table 1: Frequencies of syntactic orders3 after initial adverbs ore and encore

V2-compliant

non V2-compliant

VSp

VSo

VSn

Total

SpV

SnV

Total

Pre 1350

Ore

9

20

12

41

0

0

0

Encore

12

3

2

17

0

1

1

Post 1350

Ore

1

4

3

8

10

13

23

Encore

0

0

2

2

5

9

14

26The figures show a quite strict maintenance of inversion in texts up to 1350, e.g.

(5) Et encore promettoms nous a no dit frer (Foedera ii 1089 (1339))

27This was followed by a pronounced shift in the later C14 towards V3 order, as summarised in Table 2:

Table 2: Verb-Subject syntax following clause initial ore / encore

Pre-1350

V2 compliant

58

(98.3%)

V2 non-compliant

1

Total

59

Post 1350

V 2 compliant

10

(21.3%)

V2 non-compliant

37

Total

47

28The syntax of these time adverbs clearly displayed the main trend of the syntax of the later medieval French period, which was for XVS to give way to XSV.

29The four Old French non-inverting adverbs studied here showed a quite different outcome. Table 3 shows the syntax of main clauses with the four adverbs of this type, across the whole period of study, approximately 1250–1420. No temporal change was expected, following standard descriptions of Old French, and indeed none was found. As mentioned above, ja and onques were analysed only for their appearances in negative clauses.

Table 3: Syntax of main clause subjects with certes, nekedent, ja and onques (for key to abbreviations see Table 1)

VSp

S0V

SpV

Total

Certes

0

2

13

15

Nekedent

0

3

11

14

Ja

1

10

0

11

Onques

0

8

0

8

Total

1

23

24

48

30As predicted (Foulet 1930; Price 1973) there were almost no postfinite subject pronouns with any of these adverbs. The total absence of prefinite subject pronouns with ja and onques observed by Ingham (2005) for Old French is replicated in these findings.

31By restricting our search to two adverbs per category we have inevitably limited the amount of data that we were able to collect from the ANHT, so the figures here are probably no more than suggestive and would require replication on a larger body of texts for greater confidence to be obtained as to the validity of the conclusions. The results on clause syntax following initial adverbs are nevertheless promising insofar as they point in the direction of a distinction being maintained within AN syntax that accords with the findings of Ingham’s (2006a) study of all types of clause-initial phrasal constituents, not just particular lexical items. In that study, inversion or non-inversion in AN was also found not to be a random affair, even as late as the mid-C14, but instead depended on the status of the initial constituent.

Analysis 2: Subject pronouns with preverbal onques in subordinate clauses

32The syntax of onques in subordinate clauses was expected to show change across time, since in Middle French a Subject pronoun accompanying a preverbal adverb such as onques started to shift its position from immediately post-complementiser to immediately preverbal. The ANHT data for onques in subordinate clauses where no other preverbal element occurred beside a possible pronoun subject were accordingly divided into two time periods. We chose to make the same cut as in Analysis I, i.e. before and after 1350.

Table 4: the syntax of subject pronouns with preverbal onques in subordinate clauses (for key to abbreviations see Table 1)

Sp onques V

onques Sp V

onques S0V

T

Pre-1350

18

0

15

33

Post 1350

11

5

9

25

33Although the numbers are again not large, we see as before that a trend in continental French shows up in AN data too, in the right time frame. Between 1250–1350, AN writers avoided putting subject pronouns in subordinate clauses introduced by onques, whereas after that date this restriction was not observed.

Analysis 3: Use of the partitive article de following a negative particle

  • 4  Allowing for spelling variants and also for the common AN use of des as an alternative to de, as i (...)

34Our hypothesis was that AN respected the medieval French restriction on the partitive article de, i.e. that it should appear only following the negative particle point, rather than after pas or mie. We did so by conducting a lexical search in two stages for the phrases point de, pas de, and mie de4. First we established datasets for the three particles themselves by searching the ANHT source material. This gave the following overall frequencies: pas 1,011, point 335, mie 1,314, showing that point was in fact by far the rarest of the three. Then, within a Word document created as the output of these searches, we identified all particle + de combinations. This procedure gave the results shown in the first row of Table 5 below. These contexts were examined in the full texts themselves for whether the sense was that of a partitive or not. Partitive senses are shown in the second row of Table 5:

Table 5: Frequencies of combinations pas de, point de and mie de

pas de

point de

mie de

partitive senses

2

60

0

other senses

43

6

35

total

45

66

35

35Not far short of a fifth of the overall 335 occurrences of point, therefore, were accounted for by partitive uses, e.g.:

(6) […] les autres seignours nount point de franchises (Stats i 338)

36Other senses were represented by cases where the de phrase, rather than functioning as a partitive expression, modified a constituent preceding the preposition de, e.g.:

(7a) […] qe sa intencion nestoit pas de venir a Briggerak (St Sard 16)

(7b) […] & autres qui ne vivent point de gaynerie ne destore des […] (Stats i 290)

37It can be seen that the avoidance of partitive senses with the particles pas and mie was almost categorical. The only two exceptions both occurred after 1350:

(8a) […] et navons pas des boys ne des pierres (Lett & Pet 153)

(8b) […] par dedenz le lit n’avoyent pas de linthieux forsqe […] (Sz Med 204)

38Thus although the avoidance of a partitive after pas was not absolute in the AN data as a whole, until the mid 14th century it was categorical. Insular French users overwhelmingly, respected the limitation on the partitive to a context following point, without overgeneralising to the other contexts. This adherence to the syntax of continental French accords with the results of the preceding analyses, notably as regards the respective time-frames.

39In fact, even examples (8a) and (8b) need not necessarily have been non-native overgeneralisation errors: we cannot say for sure that equivalent French texts from the continent never exhibited such structures, since electronic searches on comparable continental material cannot at present be performed. They may conceivably illustrate the early stages of the spread of the French partitive article to the context following pas, and represent the precocity of AN, well known for example for its early loss of the Old French two-case system.

6. Discussion

40This study has sought to provide Anglo-Norman comparison data for certain attested continental developments in the late Old French and early Middle French periods. It has been shown that in the syntactic and lexico-grammatical respects studied the texts of the ANHT showed identical trends to those previously identified in continental texts. These findings give rise to two principal conclusions, the first bearing on the status of insular French, and the second on the viability of using insular textual material to investigate diachronic developments in the medieval francophone domain.

  • 5  This is not to exclude peripheral phenomena that often distinguish one dialect from another; the a (...)
  • 6  However, their non-native phonology would have seriously impaired the marking of various grammatic (...)

41The first is that, syntacticallyspeaking, as late as the mid-C14 AN was quite clearly not a deviant form of French5; an account of its survival across at least three centuries needs to acknowledge its close association with the norms of medieval French in this respect. The essential difference view taken by Kibbee and others does not provide a way of understanding the linguistic situation. To reiterate the argument put forward by Trotter (2003), later medieval England and France must have enjoyed enduring linguistic contact, for the features of medieval French syntax, both stable and innovative, to have been reliably adhered to. In the time-period studied here, insular French did not become a moribund sociolect barely understood by its users, as certain older accounts of AN would have us believe. On the basis of these results, insular French users clearly possessed a syntactic competence non-distinct from that of continental speakers6. As we saw, sentence adverbs were acquired with the tacit knowledge of their formal syntactic properties, specifically as regards inversion. When changes took place with respect to these properties in continental French, they did so in insular French too. Likewise, when in the C14 subject pronouns in subordinate clauses began to be used separately from the complementiser position, we find the same in the French used in England.

42Putting together the results of this study and those of Ingham (2006a, b) concerning other grammatical variables, we find that later AN users respected the syntax of Old French/Early Middle French in the following respects:

  • Complex rules for Object pronoun form and position;

  • Subject verb inversion after different types of clause-initial constituent (Direct Objects versus temporal adjuncts);

  • Subject-verb inversion with different types of sentence adverb;

  • Omissibility of subject pronouns with different types of adverb;

  • The use of the partitive de after negative particle point but not after pas and mie.

43On none of the formal grammatical contrasts involved in these domains did the grammar of Middle English make the same distinctions, so substrate English L1 influence can be ruled out as an explanation of the accuracy shown by insular users. It therefore seems to us that the status of AN at this time was not that of a learner variety less and less well acquired by its users, as certain conventional accounts claim. Our findings, replicating as they do on a larger database those of Ingham (2006a, b), which were obtained with different sets of AN texts, uphold the view advanced in those studies that AN formed part of the medieval French dialect continuum, at least as regards its syntax. By showing that the syntax of AN evolved in step with continental French, it has been demonstrated that a systemic level of the language also linked insular users to the wider French-speaking community at this time.

44Turning now to the viability of using insular texts for the purposes of investigating later medieval French, the implication of the analyses presented here is that if we ask how a given grammatical variable developed in medieval French of this period, we get the same answer whether we look at insular or at continental texts. The characteristics of the research variables in continental texts were already provided by prior research: in our investigation, the insular texts in each case showed the same characteristics. Since the ANHT texts can be accurately categorised by time period, we get a reliable picture of how and when changes took place. If we wish, for example, to pursue in detail the reanalysis of subject pronouns from being cliticised on to a subordinating conjunction to being cliticised onto the finite verb, the indication is that we can obtain reliable information on the issue by searching the ANHT more widely, rather than limiting the enquiry, as we did here for reasons of space, to clauses with the preverbal adverb onques. This was certainly not an expected outcome, given the long tradition of assuming that by the later C13 AN was just ‘bad French’, the product of imperfect learning by second-language users. But the samples we have taken leave no room for doubt: on no grammatical variable did AN diverge noticeably from the target-like norm, yet some divergence on subtle non-communicative syntactic aspects would surely have been clearly apparent if the texts had been composed by French users who had acquired the language under radically different circumstances from their counterparts in France. In fact the degree of parallelism went beyond adherence to the continental norms of Old French: where continental syntax was evolving in early Middle French, AN began to evolve in the same ways. Insular French users were thus clearly capable, as it were, not only of hitting the target in their acquisition of the language, but even of hitting a moving target. In short, we find no reason, on the basis of the outcomes reported here, to avoid the use of AN texts in studies of later medieval French syntax. AN syntax is not a separate domain in its own right: it is to all intents and purposes the syntax of medieval French. The corollary of this finding is that we can to a very large extent study the syntax of medieval French from AN sources: rules of constituent order, the development of the partitive, and the evolution of indefinites in negated clauses, for example, can be presented and understood on the basis of insular sources as well as they can using continental sources.

45A further issue to which our findings relate has already been raised in the introduction: to what extent are differing results found by analysing data of the administrative type used here, by comparison with literary texts? It would be imprudent to generalise too widely on the basis of the handful of syntactic constructions considered here, but as far as word order is concerned, they show the same trend in roughly the same time frame as the literary texts analysed by Zwanenburg (1978), Vance (1997) and Ingham (2005). The question remains the extent to which syntactic change is genre-neutral, or whether particular types of text are the locus of particular types of change. As an example of the latter, we may take the replacement of nul by aucun as an indefinite expression in non-affirmative contexts in legal texts well before this change takes place in literary texts (Buridant 2000: 180). Further research drawing on corpora of both literary and non-literary texts will no doubt be able to shed further light on whether types of grammatical change are genre-related in principled ways.

46In this connection, an important advantage of the AN material analysed here is that it is securely datable: hence it allows us to make reliable statements about the timing of the changes in French syntax. Now we know that the ANHT does indeed provide data capable of upholding a claim about the timing of a change, it offers encouraging opportunities for further claims of this nature to be investigated. These would not be seen as restricted to insular French but in principle would extend their applicability to the wider medieval French linguistic domain.

7. Prospects

47The ANHT has been shown to be a feasible research instrument for studying grammatical changes in later medieval French, though frequencies of occurrence will be relatively low if lexically-based searches have to be conducted. To address this issue, a parsed version of the data would of course be desirable, offering larger scope for the examination of grammatical variables, both in morphology and syntax. An alternative route to explore would be the formation of a searchable super-corpus of equivalent non-literary material dawn from various regions of the later medieval francophone domain, embracing both the ANHT material and comparable continental material. This kind of resource would in principle provide a numerically far more impressive basis for research into grammatical variation and change than anything we have now. For the time being, it can nevertheless be said that quantitative research into later medieval French using the ANHT is at least possible on a pilot basis, and allows hypotheses to be more precisely and confidently formulated as to the succession of changes that took place.

We gratefully acknowledge the assistance of David Trotter and Michael Beddow, who are however not responsible for any inaccuracies or omissions contained in it.

Haut de page

Bibliographie

Anglo-Norman Hub textbase (ANHT) : http://www.anglo-norman.net

Carlier A. (2007). « From preposition to article: The grammati-callization of the French partitive. »,Studies in Language 31, 1 : 1-49.

Clyne M. (2003). Dynamics of Language Contact : English and Immigrant Languages, CUP.

Foulet L. (1930). Syntaxe de l’ancien français, Paris.

Harris-Northall (2005). «The count / noncount distinction in Castilian : evidence for its place and function in the medieval language»,in P. Ricketts & R. Wright (eds.) Studies on Ibero-Romance linguistics. Newark : Delaware, 167-186.

Herman J. (1990). Du latin aux langues romanes : études de linguistique historique, réunies par S. Kiss.Tübingen : Niemeyer.

Ingham R. (2005). « Adverbs and the syntax of subjects in Old French »,Romania 123 : 99-122.

Ingham R. (2006a). « Syntactic change in Anglo-Norman and Continental French : was there a ‘Middle’ Anglo-Norman ? » Journal of French Language Studies 16, 1 : 25-49.

Ingham R. (2006b). « The status of French in medieval England : evidence from the use of object pronoun syntax », Vox Romanica 65 : 1-22.

Ingham R. (2007a). Variation in later Old French syntax, Ms., Birmingham City University.

Ingham R. (in press). « La syntaxe en transition entre l’ancien et le moyen français », in S. Prévost (éd.) Proceedings of Diachro III, September 2006.

Kabatek J., Pusch C. & Raible W. (2005). « Romance corpus linguistics and language change – an introduction », in C. Pusch, J. Kabatek & W. Raible (eds) Romanistische Korpuslinguistik II. Korpora und diachrone Sprachwissenschaft. Romance Corpus Linguistics II. Tübingen : Gunter Narr Verlag, 1-10.

Kibbee, D. (1991). For to speke Frenche trewely. The French language in England 1000-1600 : its status, description and instruction. Amsterdam.

Kroch A. & Taylor A. (2000). « Verb-object order in early Middle English. »in S. Pintzuk, G. Tsoulas & A. Warner (eds.) Diachronic syntax : mechanisms and models. OUP.

Kroch A., Taylor A. & Ringe D. (2000). « The Middle English verb-second constraint : a case study in language contact and language change. » in S. Herring, P. Van Reenan & L. Schøsler (eds.) Textual Parameters in Older Languages. Amsterdam : John Benjamins.

Marchello-Nizia C. (1995). L’évolution du français : ordre des mots, démonstratifs, accent tonique. Paris : Armand Colin.

Marchello-Nizia C. (19972). La langue française au XIVe et au XVe siècles. Paris : Nathan.

Price G. (1973). « Sur le pronom personnel sujet postposé en ancien français », Revue Romane 8 : 476-504.

Price G. (1984). « Anglo-Norman », in The languages of Britain. London : Edward Arnold.

Thomason S. & Kaufman T. (1988). Language contact, creolisation, and genetic linguistics. Berkeley : Univ. of California Press.

Trotter D. (2003). « L’Anglo-Normand : variété insulaire ou variété isolée? »,Grammaires du vulgaire. Médiévales 45 : 43-54.

Tuten D. (2005). « Reflections on dialect mixing and variation in the Alfonsine Texts », in P. Ricketts & R. Wright (eds.) Studies on Ibero-Romance linguistics. Newark : Delaware, 85-101.

Vance B. (1993). « Verb-first declaratives introduced by et and the position of pro in Old and Middle French »,Lingua 90 : 281-314.

Vance B. (1997). Syntactic change in medieval French. Dordrecht.

Wilshere A. (1993). « A plea for syntax », in I. Short (ed.) Anglo-Norman anniversary essays. London : Anglo-Norman Text Society, 395-404.

Zwanenburg W. (1978). « L’ordre des mots en français médiéval. », in R. Martin (éd.) Etudes de syntaxe du moyen français. Centre d’Analyse Syntaxique de l’Université de Metz, 153-171.

Haut de page

Notes

1  As noted by Kabatek et al. (2005 : 4), the Romance research community experienced a relative delay in gaining access to diachronic electronic corpora.

2  Though see Wilshere (1973).

3  Abbreviations: VSp = Verb-Subject pronoun order; VSn = Verb-Subject; VSo = Verb, no subjec

4  Allowing for spelling variants and also for the common AN use of des as an alternative to de, as in example (8a) in the text. This trait presumably arose from phonological confusion between the final vowels of de and des following the loss of schwa in later AN.

5  This is not to exclude peripheral phenomena that often distinguish one dialect from another; the argument here relates to core areas of grammar involving main word order features of the language.

6  However, their non-native phonology would have seriously impaired the marking of various grammatical notions, especially the exponence of gender.

Haut de page

Pour citer cet article

Référence électronique

Richard Ingham, « The grammar of later medieval French: an initial exploration of the Anglo Norman Dictionary textbase », Corpus [En ligne], 7 | 2008, mis en ligne le 13 novembre 2009, consulté le 28 avril 2017. URL : http://corpus.revues.org/1506

Haut de page

Auteur

Richard Ingham

Birmingham City University

Haut de page

Droits d’auteur

© Tous droits réservés

Haut de page
  • Logo Revues électroniques de l’université de Nice
  • Revues.org