Diachronic aspects of borrowing aspect Diachronic aspects of borrowing aspect : the role of Old French in the development of the be going to infinitive construction

change of state meanings Inanimate subject in (8): abstract change of state Aspectual reading in (9): author of the chronicle was conscient of the change (8) Also þe see and þe lond gan to wexe more bareyne þan þey were to forhonde. (PPCME2: polych) (9) . . . but he [the king Conradus] regnede sevene gere. In þis yere þe empere gan to passe from þe Frensche men to þe Almayns,. . . (PPCME2: polych) Mossé (1938b, §290): his example with the progressive form is a translation of a French text from 1513: (10) "’sir’, quod Gerames, ’we be frenchmen, pylgrymes, & are goyng to offre at ye holy sepulcre’” (Huon of Burdeux 1534: 191)


Introduction
This paper addresses the question of whether it could be argued for the assumption that the rise of the be going to + infinitive construction in English is the result of contact-induced change.More precisely, we would like to investigate if there is enough empirical proof for this construction to be borrowed from Old French (OF), during the Middle English (ME) period.We would like to point out that we are neither in favour of the contact-induced nor the language-internal point of view, rather we see this phenomenon as a primary example to deal with structural borrowing from a historical perspective since it shows that it poses methodological problems which have not been solved so far.
We are well aware of the fact that the contact-induced hypothesis is debatable but nevertheless it has been addressed by several authors.Mossé (1938), Pérez (1990), Danchev &Kytö (1994), andNunez Pertejo (1999) have all claimed that a transfer of this construction from French to English is plausible and probable since the contact situation between English and French in medieval times, more precisely between 1066 and 1400, has been claimed to have been most intense leading not only to massive lexical but also heavy structural borrowing (cf.Rothwell, 1968Rothwell, , 1975Rothwell, , 1980Rothwell, , 1993Rothwell, , 1996 on the sociolinguistic and culturual situation at that time).Recently, a number of authors have especially dealt with structural borrowing in this context and have come to the conclusion that it has been underestimated in the literature (cf.e.g.Ingham 2006Ingham , 2009Ingham , 2010Ingham , 2011a,b;,b;Haeberli 2010;Trotter 2000Trotter , 2010;;Wright 1997Wright , 1998Wright , 2003Wright , 2010)).
Direct evidence for intense contact, on the one hand, comes from a number of grammatical constructions which were demonstrably calqued from French (cf.Prins 1948), and, on the other hand, also from the fact that a multilingual situation (Anglo-Latin, Anglo-French, varieties of ME) prevailed.Additional support comes from the transfer of go-futures from French to Southern Dutch and Western German dialects (cf.Danchev & Kytö 1994).
According to Prins (1948) numerous phrases, proverbs and proverbial sayings were borrowed from French in ME times like e.g. the verbal phrase s'arrêter court > to stop short or faire la paix > to make peace (Old English (OE) friþ geniman).The phrases he lists must be analysed as calques which means that morphemic constituents of borrowed words or phrases are translated item by item into equivalent morphemes in the new language (also called loan translation).Since they are translations, however, it is very hard to define whether they were borrowed or not.
In a similar vein, Orr (1962) claims that the influence of OF on English was very strong and extends Prins's collection of words and phrases to syntactic constructions.As concerns the construction with a present participle he states: Much more apparent is the influence of French upon the conjugation of verbs, particularly upon the uses of the present participle (or gerund).To quote Einenkel: 'Despite the opinion com-monly encountered that modern he was fighting goes directly back to OE he waes feohtende, it is quite beyond belief that the modern durative participle has developed independently from the periphrastic participle of OE without determinating influence from the Romance gerund'.(Orr, 1962 : 15f) Moreover, he notes that OF as well as Modern English (ModE) show further parallels in exhibiting the combination of the participle and the verbs venir (come) and aler (go).He states that in OF the latter is abundant, one example of which is given in (1), (1) Asez savez le grant orgoill Reliant ... Pur un sul levre vait tutejor cornant.
Devant ses pers vait il ore gabant.Suz eel n'ad gent ki 1'osast querre en champ.Car chevalcez!Pur qu'alez arestant?(Roland, 1773ff. in Orr 1962 : 17) and that in both languages the verb can function both as a genuine verb of motion and as auxiliary (e.g.Don't go telling me that sort of stuff ).
In the following, we will support the language-contact hypothesis and claim that there is evidence for transfer not only of morphological but also of syntactic structures.In Trips & Stein (2008) it was shown that syntactic structures bound to the suffix -able had an influence on English word formation: The ME suffix had the "normal" passive sense 'capable to be Ved', but also the active sense 'capable of Ving' which occurred in Latin and OF.
We are aware of the fact that borrowing of an aspectual periphrase happens at a level which is different from suffix borrowing.Thus, Matras (2009 : 212) states that: "[i]nflectional morphology is applied at the sentence level, not at the word level [. . .]; it does not, by default, accompany individual words, since it is not an inseparable component of the meaning of those words, and hence not directly relevant to the goal for which the word is being borrowed in the first place (namely specificity of reference).This is the major difference between derivational and inflectional morphology, and the main reason why the borrowing of inflectional morphology is rare compared to that of derivational morphology." The assumed rarity of structural borrowings is a further reason to use text corpora in order to find more evidence for (or against) the language-contact hypothesis (this desideratum is also expressed by Hoffmann 2005 : 174, who is hesitant about considering the development of complex prepositions like by virtue of as structural borrowings from OF).We have conducted such a comparative study of OF and ME texts investigating a number of annotated and non-annotated corpora. 1 Thus, in this paper we focus on syntactic structures, and aim to answer the following questions: • What is the situation in OE like, i.e., do we find a construction that could be the predecessor of the ModE construction?
• What is the situation of the aspectual system in OF like (cf.Ménard 1973) and does it allow for a transfer of this construction to ME?
• What evidence do we find in ME texts translated from OF?
• From a more theoretical point of view: is there enough evidence for the language-contact hypothesis or is it more plausible to assume a standard grammaticalization process motivated by system-internal changes?
Moreover, we are aware of the fact that assumptions concerning structural borrowing also have to take into account the following three domains: • linguistic structures, • varieties and the socio-linguistic context, • extra-linguistic factors.
The paper is structured as follows: in the following section (2) we will start with a description of the be going to + INF construction in ModE and possible origins in OE and ME by rewieving the most promiment assumptions and by presenting and discussing findings from our own corpus-based studies.Section 3 provides socio-linguistic background information about the language contact situation between speakers of ME and OF from 1066 to 1400, and discusses a number of assumptions that have been put forward to support the language-contact hypothesis.In section 3.2 the properties of the French aler + INF constructions are introduced and discussed, again by reviewing the relevant literature and by looking at findings from Old French (OF) corpus data.Section 4 concludes by summarising our empirical findings and assumptions and by critically evaluating them.
2 English constructions with go and similar verbs

The be going to + INF construction in Modern English
In ModE (Greenbaum & Quirk, 1990 : 57), the construction under investigation expresses the meaning 'future fulfilment of the present' which includes two more specific meanings 'future fulfilment of a present intention', as in (2a) and 'future result of a present cause' as in (2b).
(2) a.When are you going to get married?b.It's going to rain/She's going to have a baby.
If we want to shed light on the origins of this construction, we will have to take a look at the earlier stages of English to see which constructions occurred at that time.More precisely, we must investigate the syntactic properties of all the elements part of the construction (be + go + -ing + to + INF), the (adjacent) order of these elements, the meaning/function of all the elements from OE to ModE, and possible semantic changes these elements have undergone.First we will turn to OE.

Old English constructions
In the literature, it has been noted that in OE the periphrastic construction with the verbs beon / wesan / (weorðan) and the present participle with the ending -ende might be the predecessor of the be going to + INF construction.In the following, the most prominent assumptions are given concerning the occurrence and the function of this construction.
According to Visser (1973), the OE construction beon / wesan / (weorþan) + V-ende may express ingressive, durative or terminative aspect.If the participle occurs in combination with a form of to be or with an adverb, the construction predominantly expresses futurity ( § § 1800-1889), as in the following examples 2 : (  Mitchell (1985) states that the participle in this construction can be adjectival, appositive or nominal, which, however, is hard to determine.He further notes that according to several authors the semantic potential of the construction is wide, since it may refer to a specific moment, a continuing process, a finished action, an action of limited duration or a habitual action.Mitchell himself denies aspect to be part of the OE system, and he further claims that the construction is far from expressing one of the mentioned semantic aspects in a systematic way, so only tendencies can be observed ( §685ff); further he observes that if the construction occurs, it is often found under Latin influence (but not consistently so, cf.§689).To these observations and assumptions, Nickel (1966 : 201ff) adds that the construction used to be a stylistic means and that variations occur from author to author ( §690).Denison (1993 : 380) notes that in OE the form of the progressive is be + Vende/ing, its frequency is unevenly distributed and it strikingly often occurs in the Orosius.Further, it has been noted that the construction was often used to translate perfect deponent verbs in Latin texts (cf.e.g.Wülfing 1901, cited in Nickel 1966), but as Jespersen (1982 : IV,166)  Although this construction obviously already existed in OE times, in the following we will assume that the pattern auxiliary + participle is not directly linked to the rise of the be going to + INF construction.Rather, the aspectual function is due to the co-occurrence of go (to) and the infinitive form of a following lexical verb (INF), whereas the development of the continuous is an independent process which occurs later and is not related to the OE "continuous form".As a consequence, for our analysis this means that semantically and structurally, we need to take into account the syntagmas go + to + INF and go + INF 4 , and further the change of verb-final to verb-medial (Infl-final to Infl-medial) structure because it is a prerequisite for the adjacent order of the modern be going to + INF construction 5 .
First, we will take a closer look at to which in the OE gan + to + Vanne 6 construction adds the purposive function to the originally directional construction and thus, with the verb of movement leads to an inference of futurity (cf.Hopper & Traugott, 2003).In OE, the preposition to in the ordinary use, i.e. before a noun, expresses the following functions relevant for us here (see the Oxford English Dictionary online (OED online)): 1. Expressing motion directed towards and reaching.In all these cases, the simple dative was reinforced by the preposition to, and this is also the reason why it started to become used in front of the verbal infinitive: we know that in OE, apart from the simple infinitive form the so-called "dative infinitive" with the -anne-ending existed which was consistently governed by the preposition to.It was gradually levelled with the simple infinitive form and reduced to the non-finite verb stem (i.e.without any inflectional ending).As with nouns, the meaning of to in combination with the infinitive was "motion, direction, inclination, purpose, etc., toward the act or condition expressed by the infinitive" (OED online) and it occurred with verbs such as onginnan 7 'to begin', ondraedan 'to dread', bebeodan 'to bid', etc.Thus, it was construed either with the simple infinitive (e.g.ic onginne + INF) or with the dative infinitive (e.g.ic onginne + to + Vanne).For our study, this means that apart from the constructions listed above we would also have to include the Vanne-construction into our study.
Our corpus study reveals that the construction with the elements adjacent to each other occurs, but it is quite rare (41 instances).9 of these occurrences show the verb beginnan, 4 the verb onginnan 'begin' and 1 cuman 'come, go, happen'; one of these examples is given below: Sentences including this construction show that due to the contiguity of beginnan/onginnan/cuman + to + dative infinitive an aspectual meaning arises since all three elements share the feature 'direction'.As concerns the occurrence of this construction with non-adjacency between the finite verb and the non-finite clauses introduced by to (i.e. in main clauses), we see that only few examples occur, and that most of them exhibit a non-complement infinitive (no examples of that type with the bare infinitive were found): worship (AECHomII 428.17 from Mitchell 1985:403) Concerning the other construction relevant here, gan + INF, we gain a similar picture: In the corpus, we found instances of this construction (see example ( 10)), but all in all only very few examples occur, as However, the first of these must be excluded since it is not directly linked to the ModE construction expressing futurity (as we will show in the following).The other three constructions, which are all relevant for us, were only rarely found in the corpus, and most examples showed a non-complement infinitive.Next, we will take a look at the structure and semantics of the construction in ME.

Changes from Old to Middle English
Since some of the drastic structural changes taking place in the transitional period from OE to ME bear on the development of the be going to + INF construction, we briefly deal with them here.
First, from the end of the 12th c. there was a growing tendency to confuse -inde, phonetically or scribally, with -inge.According to the OED online this confusion is especially noticeable in manuscripts written by Anglo-Norman scribes in the 13th c.The final result was the predominance of the form -inge, and its general substitution for -inde in the 14th c. (OED online).As Mossé states ... there is dialectal variation as concerns the forms -ande, -ende, -inde; the verbal -ing form develops from nouns that in OE were built with -ung/-ing suffix.The development from -ende/inde to -ing has to be explained phonologically, morphologically, and syntactically: ... (Mossé, 1938 : 145) Second, the OE preterit form of gan, eode, was replaced by the ME form went. Mossé describes the situation as follows: As in OE the verb gon had no preterit based on the root used in the present system.In ME the South (and the Midlands in the 12th century) used the suppletive forms eode, yede, yode (from the OE suppletive eode, ge-eode) for the preterit.The North and the Midlands used a new suppletive form went(e) from the current verb wenden 'to go'.This form appeared first in the North and moved gradually through the Midlands to the South.Mossé (1991 : 86) Third, it is a well-known fact (cf.Pintzuk 1999, Trips 2002, Pintzuk & Taylor 2006) that English underwent a change in the relative position of the full verb and its complement (from object-verb to verb-object) and of the finite verb and the full verb (from full verb-finite verb to finite verb-full verb).These changes also affect the be going to + INF construction since complements and finite verb no longer occur in postverbal position.
Obviously, these changes had an effect on the OE constructions discussed in the previous section: in ME we expect to find the new ending -ing on the stem of the verb gon, the occurrence of the new past tense form and a new ordering of the elements part of the construction under investigation.

Middle English corpus data
In the annotated corpus of ME prose, the PPCME2 (Kroch & Taylor, 2000), although the participle going (with its many spelling variants) occurs quite frequently as noun, it occurs only in five examples with the auxiliary be, and only twice with the preposition to, an example of which is given in ( 13): (12) and founde an ermytage and an ermyte therein whych was goyng unto masse.The example in (13) also shows that the element occurring after the preposition to is a DP, and not a verb.
In the corpus of the Middle English Dictionary (MED, 2001), we gained the same picture, but here we found three cases where the element after to is a verb 9 .However, in all three examples (15), the -ing form introduces a relative clause: (15) a.What is striking is that apart from these cases we found gan -the past tense of ME ginnen 'begin' (OE onginnan) -in these constructions (13 of the type gan + INF, 25 of the type gan + to + INF).This form is homophonous with the infinitive and plural form of ME gon 'go', and since both verbs share the feature 'direction', an influence of ginnen on the development of go + to + INF constructions seems likely. 10 In the corpus, the ambiguous form gan starts to emerge in period "m23" (composition date 1250-1350).Gan + INF and gan + to + INF occur only in 13 (out of 55) texts, but again it seems that the texts with more numerous occurrences are translations from French or Latin (The Book of Margery Kempe: 10, John of Trevisa's Polychronicon: 12), implying that the constructions with this form may have been borrowed and influenced the constructions with forms of 'go' in ME times.
Generally, it is difficult to distinguish clearly between local and abstract readings.In the PPCME2 some contexts are clearly local ( 16), others are ambiguous (17), but ( 18 The ambiguous form gan ('go' or 'begin') often appears with inanimate subjects indicating an abstract change-of-state meaning and allowing a progressive interpretation as shown in (19).In (20), taken again from the Polychronicon, the aspectual flavour is even more obvious, since the writer of the chronicle knew that the 'emperorship' was going to pass from France to Germany: , and what was striking here was that quite a number of them exhibited the ambiguous form gan which tends to have an aspectual flavour and predominantly occurs in texts which are based on French or Latin originals.Thus, it seems that the language-contact hypothesis may be tenable.To find further support for this hypothesis, it will be necessary to consider the special sociolinguistic situation during the ME period.
3 The language-contact hypothesis and OF influence

The socio-linguistic background: bilingualism
In the preface of the online version of the Anglo-Norman Dictionary (Rothwell & Trotter, 2005), Rothwell summarizes the situation in medieval England as follows: Anglo-Latin gradually lost ground to Anglo-French in its role as the official language of record at both national and local level, whilst Middle English emerged over time from being a predominantly spoken language to take over from the two others in the fifteenth century as the acknowledged national language, both spoken and written.This simple summary statement, however, hides a complex linguistic interplay brought about by the continuously evolving social situation in Britain and on the continent for many decades after the Conquest.
He also remarks that the complexity of the situation is increased by the absence of an adequate body of surviving recorded evidence before the fourteenth century, and the fact that Anglo-French, while the proportion of native French speakers diminished, had been adopted as the language of writing, whereas Middle English "was there all the time in the background as the spoken vernacular of the majority of the population".Rothwell stresses the persistence of French in the professional and commercial domain, but notes that "[i]n the private domain, the educated members of society exchanged letters in Anglo-French in abundance throughout this period."Numerous Anglo-Norman Texts have been translated to English: e.g. the Roman de Brut (by Wace, around 1150) or the Roman de la Rose (by Chaucer).We do not expect direct translation equivalents to prove the borrowing of aspectual periphrases, but a cursory glance at the French contexts translated by go + (to) + INF can give a first impression of the aspectual values covered by the English construction in Chaucer's The Romaunt of the Rose: the durative expressed by the French imperfect penoit in ( 21) and the inchoative expressed by commencer 'begin' in ( 22).Whereas lexical influence of French on English is treated in many publications, most scholars are sceptical about structural impact of French.The most courageous authors claim that ME is a French-based creole (e.g.Bailey & Maroldt 1977), but most others deny this, e.g.Görlach (1986) who states: Influence of French on inflections and, by and large, on syntactical structures cannot be proved, but appears unlikely from what we know about bilingualism in Middle English times.Middle English is a typical case of a language of low prestige, predominantly used in spoken form and split into a great number of dialects that had to assimilate the cultural (and in this case mainly lexical) impact of the 'high' language.To call such a process 'creolization' is misleading, especially if it is extended (as is done by Bailey) to the influences of French in the 15th century, these must have been purely from above and in written form.(Görlach 1986 : 338) It has to be noted, however, that although Bailey and Maroldt talk about creolisation, they define this term quite differently from the standard notion of the term in that they assume that ME is the result of a massive importation of English lexical items into the OF spoken at that time on the isle by the upper classes of England.According to Thomason & Kaufman (1988 : 306-315) this scenario is very unlikely unless we assume extremely unsuccessful learning of the English tongue by speakers of OF implying an urgent need for a new contact medium in a new multilingual contact situation.What clearly speaks against this assumption is first of all the fact that there were only two languages in contact and that there was "ample opportunity for bilingualism to develop" (Thomason & Kaufman, 1988 : 307).Moreover, according to these authors, the number of OF speakers in England was limited (mainly to the royalty and gentry), these speakers began giving up OF by 1235 at the latest, and there is no reason to suppose that any large proportion of native English speakers learned OF between 1066 and 1250; after that point they had no reason to do so.A further observation speaking against the language-contact hypothesis is that all dialects of ME which were in contact with OF did undergo simplifications which cannot be attributed to OF influence.Rather, having originated in the northern dialects, they emerged at that time in the East Midlands dialects and spread to the standard.Thomason and Kaufman therefore come to the conclusion that The massive French influence on English vocabulary, followed by the mild influence on English morphology and syntax, and the practically trivial influences on English phonology, took place at a time when there were practically no competent French speakers around for an Englishman to talk to.(Thomason & Kaufman 1988 : 308) We partly agree with Thomason and Kaufman in that the creolisation hypothesis is not tenable.However, it has been shown by a number of experts in the field, especially by Rothwell mentioned above, that the contact situation between ME and OF has been highly underestimated in the literature: The sheer size of the debt owed by English to Anglo-Norman French and -just as importantly -the period of time over which this debt was built up are still grossly under-estimated by specialists in the history of both the French and English languages.(Rothwell 1993 : 310) This is probably at least partly due to the fact that structural borrowings are very hard to find, and that only few texts that give clear information concerning the linguistic situation at that time are available (e.g. the so-called macaronic texts, see below).We claim that such borrowings can be found, if only we look hard enough for them.Here we follow authors who have extensively worked on ME and on the language-contact situation with OF, like e.g.Rothwell cited above, as well as some of the authors we will briefly mention in the following.
In his seminal work on ME, Mossé (1938) states that French influence is weak but may have sustained the English progressive in the period between 1200-1340, and that translations from French and Latin played an important role.One of the early examples Mossé (1938: §290) cites for the occurrence of the be going to INF construction is ( 23), which is indeed a translation of a French text from 1513: (23) "'sir', quod Gerames, 'we be frenchmen, pylgrymes, & are goyng to offre at ye holy sepulcre'" (Huon of Burdeux 1534: 191) In her study of the be going to + INF construction Pérez (1990) observes that in the world's languages two types of constructions occur: the andative based future expressing movement away from the speaker and using a form of a verb meaning 'go', and the venetive based future expressing movement towards the speaker and using a form of a verb meaning 'come'.She further notes that OE shows two forms of go, gan 'go, come, walk, happen' and gangan 'go, walk, turn out' (see B & T) with overlapping meanings.However, only the former is part of the construction and Pérez speculates that this is the case maybe because the notion of immediacy is inherent.Further, she finds it striking that the past tense forms eode etc. were replaced in ME times by the form went from wendan 'turn' with the result that the meaning of go is compatible with present tense, progressive, immediate (or prospective) future but not with the past (p.5).For her, this development is a standard process of grammaticalisation, and not due to external factors.
According to Haegeman (1983) the inherent properties of constructions containing a verb of movement have the potential to develop into expressions of futurity.For her, the verb go shows dynamic time-orientation in that the current orientation of the be going to construction relates a future event to the present and to the period of time immediately preceding the present.Hence, the volitional component in these constructions is a reflex of its origins, i.e., the contiguity of the lexical verb go + an adverbial of purpose.Danchev & Kytö (1994), on the other hand, support the language-contact hypothesis by saying that French influence is very likely since, first, a very similar andative periphrastic construction emerged somewhat earlier in French than in English and the development runs parallel to the English construction, and second, the go-future emerges in other French-Germanic contact areas (Southern Dutch, West-German dialects).In a similar vein, Nunez Pertejo (1999) assumes that the rise of the go-future might be due to French influence and puts forward the following arguments: first, calquing could be seen as a result of the loss of the OE prefixal system; second the go-future type of construction was somewhat earlier in use in French than in English and hence borrowing is possible and likely; third, some early examples are translations from French sources (she cites Mossé's example given in ( 23)).
This brief overview shows that the language-contact hypothesis has been discussed controversially, one main problem being that it seems to be very hard to find direct evidence for structural borrowing of the be going to + INF construction from OF.However, in our discussion it has also become clear that there are a number of clues that support this hypothesis.
The best proof for this kind of structural borrowing would be to catch the language user "red-handed", that is in the act of using the English structure to express its French equivalent.Asking for such kind of evidence in historical contact situations seems preposterous, also because it demands the purporters of the language-contact hypothesis to produce a kind of evidence which is out of reach for the purporters of the language-internal grammaticalisation hypothesis (who have to rely on appropriate contexts to postulate that the speaker has reanalysed a form).Nevertheless, some texts which are directly connected to the bilingual situation can provide findings which come quite close to these "ideal pieces of evidence".
Our first example are the so-called macaronic texts found in ME times.Laura Wright has shown in several papers that they provide clear evidence for direct contact between Anglo-Normans and English speakers (Wright 1995(Wright , 1997(Wright , 1998(Wright , 2003)).One example from Wright (1998:105)  Example ( 24) shows that the ordering of the elements in the nominal phrase haryng(e) blank(e) follow the syntactic rules of French where the colour adjective has to occur in postnominal position, whereas it has to precede the noun according to the syntactic rules of Germanic: Blanke heryng.There is no doubt that the postadjectival pattern was borrowed from French, and if such structural borrowings exist, the possibility of borrowing the be going to + INF construction cannot be excluded.Since agreement and word ordering are part of syntax there is no reason to exclude other syntactic borrowings from the language-contact situation between the Anglo-Normans and speakers of ME.
Our second example are glosses.Some glosses of OE texts show that there must have been direct contact between the two languages.AElfric's Latin Grammar (Zupitza, 1880) has been glossed by two Anglo-Norman hands.Especially two observations are important: First, AElfric uses periphrastic constructions to express the inchoative aspect: ic wearmige vs. ic onginne to wearmigenne (Zupitza, 1880 : 212); second, the two Anglo-Norman writers who glossed AElfric's Grammar used aler + INF periphrases for the future of the infinitive: "FVTVRO auditum ire VEL auditurum esse" (Zupitza, 1880 : 188) is glossed by "aler oir u estre alét oir" (Hunt 1991 : 110; analogous forms are quoted for the other conjugations, e.g.amare, docere, legere): uis amatum ire wylt dû faran lufjan; uenatum pergo ic fare huntjan; uis doctum ire wylt dû gân leornjan; lectum pergit hê g aed rêdan (Zupitza, 1880 : 134) Further examples, like direct translations from OF to ME (e.g.Somme le Roi and Ayenbite of Inwit) could be mentioned here, but for now it suffices to note that the analysis of these special text sorts can provide quite convincing corroborations of the language-contact hypothesis.For our analysis of the be going to + INF construction, however, we will focus on the general language corpora mentioned in the introduction, and survey eventual equivalents of this construction in Modern French and earlier stages of the language.

The diachrony of periphrases with aller
Today, the periphrase aller + INF expresses the immediate future, similar to ModE be going to.However, this meaning is fairly modern: in OF aler + INF had a different meaning.Latin did not have a morphological form to express the immediate future, and medieval French used modal verbs like devoir, also to translate the Latin simple future in contexts expressing immediateness (Gougenheim, 1929 : 85f): (25) Li jurs aprechad que li reis David dut murir.= Appropinquaverunt autem dies David ut moreretur.(Quatre Livres des Rois) With respect to the development of the periphrases with aller, most authors agree that it happened after the OF period: for Ménard (1973 : 131), the periphrastic future aller + INF developed only in the 15th c. to express an immediate future, Werner (1980 : 132) cites an example from the Cent nouvelles nouvelles (15th c.) as unambiguously abstract, for Wilmet (1970 : 191), quoting Gougenheim, its development accelerates in the 15th c., and according to Buridant (2000 : §213), "l'ancien français ignore la périphrase aller + infinitif signifiant au départ le futur proche".Other scholars consider periphrases with verbs other than aller, like Pountain (1982 : 147)

Aspectual meaning
Gougenheim attributes an aspectual meaning ("sens inchoatif") to sentences like Je le vois querre sans respit (Gougenheim, 1929 : 92ff).He notes that it occurs in collocations with certain verbs of movement or warfare, and considers it as a predecessor of the immediate future reading we mentioned above: "la périphrase française du futur prochain, qu'a précédée [. . .] une autre périphrase formée du verbe aller et de l'infinitif, avec un sens inchoatif très précis."(Gougenheim, 1929 : 92) He also notes that aller + INF and aller+GERUND have not been studied in sufficient detail.He tries to define more precisely the meaning of aler + INF by the paraphrases "il dit tout d'un coup" (sudden start of the action).More recent publications also argument in this vein and consider constructions with aler as aspectual periphrases: Les périphrases sont toujours restées en deçà d'une intégration pleine dans le système verbal.Ainsi, pour la phase durative, l'ancien français a surtout exploité dans une certaine mesure la périphrase aler+participe présent, mais il ne l'a pas grammaticalisée au point d'en faire l'équivalent de la forme progressive pour l'anglais, où s'opposent un présent d'action, purement narratif, et un présent de visualisation impliquant le rapport d'un témoin.(Buridant, 2000 : §283) Detges ( 2004) shows that constructions denoting the inchoative aspect also have a discourse-structuring function (foregrounding), like Catalan anar + INF and OF aler + INF.In the case of Catalan, this function can explain the development towards a periphrastic perfect, different from the usual metonymic paths suggested for 'go' verbs (movement > intention > posteriority).
Progressive periphrases in Romance languages combine either a copula verb ('be') or a movement verb ('go', 'come') with either the present participle or the infinitive.An optional preposition (mostly 'to') can precede the infinitive (Schøsler, 2005 : 116).Gougenheim (1929 : 2) quotes the participle constructions Qui mentiendo vadunt in Merovingian Latin (810), and Schøsler uses several tests to prove that even in the earliest OF texts (e.g. in Alexis: tut s'en vat declinant 'everything is getting worse'), these constructions are undoubtedly aspectual periphrases (Schøsler, 2005 : 119).Their frequency is low in the oldest texts, but increases after 1050.Schøsler quotes numerous examples where the fact that the second verb is a movement verb clearly proves the aspectual function (and hence the grammatical status) of the construction (Nunez Pertejo 1999 also uses this criterion).Like Gougenheim (1929 : 97), Schøsler claims that in the OF period these aspectual periphrases are more current in colloquial texts than in courteous verse (poésie courtoise), a tendency which is reversed in later periods.

Borrowing of periphrases
In his discussion of be going to, (Mossé, 1938 : Vol. II, §287-288) shows that French aspectual periphrases were borrowed, e.g.être sur le point de, and that they continued to exist as stylistic variants until English had found in the construction going to "la véritable forme qu'il cherchait": Although being of general interest, Mossé's statements refer to a period posterior to the Anglo-Norman influence on English.Earlier in his work, but still with reference to aspectual periphrases, Mossé (1938 : Vol.II, §97) claims that there is no difference between the Anglo-Norman and the other OF texts.
We therefore use examples taken from the Anglo-Norman Dictionary as well as from the Nouveau Corpus d'Amsterdam (NCA), in which 5.9% of the texts (about 176.000 words) are positively tagged as being of Anglo-Norman origin (with respect to the manuscript "lieu de manuscrit").

Old French corpus data
The relative frequencies for the OF and ME periphrases (adjacent forms) are compared in The construction with the direct object preceding the infinitive (aler + OBJ + INF, 55 occ.) is less frequent, and rather constrained with respect to the selection of the object: querre 'ask' is the most frequent verb (aler OBJ querre means 'go to ask for something', but also 'fetch something'), but occurs frequently in fixed expressions like congie querre 'to leave'.is significant that the second most frequent infinitive in aler constructions, ferir 'hit', does not appear with preceding objects at all. 11Also, the object clitics or pronouns, except for very few counter-examples, precede the auxiliary: OBJ-CLIT + aler + INF.We therefore conclude that the modern form, with adjacent verbs and postponed object, is less restricted and hence the relevant candidate for potential influence on English.
Corpus evidence also helps us to find out more about the aspectual value of the latter two constructions by an analysis of the tense forms of the auxiliary.The ModF past tense system distinguishes between perfective aspect, expressed by the perfect tense, and the imperfective aspect, expressed by the imparfait.This difference is well established in the 16th c. (cf. the explicit remarks of Palsgrave (1530) in the chapter Of the verbe, folio cxii-cxiii).Let us assume that this aspectual difference was already present in OF (for a detailed diachronic description of the tense system see Blumenthal 1986).Then the tense of the auxiliary aler could be an indicator for the aspectual value of the aler + INF vs. aler + GERUND periphrase.This seems to be true: the 3rd person singular and plural forms of the imparfait (aloit, aloient) occur in 9.7% of the aler + GERUND constructions (compared to 3.9% for the 3rd person perfect forms), but only in 3.2% of the aler + INF constructions (13% perfect).We therefore assume a closer affinity of aler + GERUND with the progressive, and of aler + INF with immediateness.

Conclusion
We are far from adopting an "anything goes" attitude (Matras, 1998 : 282) with respect to structural borrowing, we merely believe that methodologically, both hypotheses-language-internal grammaticalisation and structural borrowing-face the same problem of defining valid criteria for measuring the weight of their assumptions.In historical situations of intense language contact, it is impossible to exclude either of the two possibilities.Here we have shown what a corpus-based study can contribute to this discussion.
As we stated in the introduction, the plausibility of assumptions in favour of the language contact hypothesis relates not only to the analysis of linguistic structures, but also to the consideration of language varieties and the socio-linguistic context, as well as of extra-linguistic factors.In 3.1 we presented Rothwell's view of the bilingual situation in Anglo-Norman England, but the opinions on the socio-linguistic situation strongly diverge.Therefore, even if there is clear textual support for potential structural borrowing, one can always object that there is no certainty about the proportion of the language community which actually used a given structure.
This heuristic dead-lock can be alleviated by taking extra-linguistic factors into account.Without going into detail here, we suggest the use of weighted indicators, similar to the ones applied in Optimality Theory, to corroborate empirical findings.Meta-linguistic comments, for example, would be a very strong support (indicated by ++), some others are listed in Table 5.
Although we need to bear these indicators in mind, in this paper we focussed on the (structural) syntactic and semantic parallelism in the source and target languages, since they are a prerequisite for structural borrowing, even if form and meaning may be modified in this process.
Metalinguistic comment (in grammars etc.) ++ Higher frequencies with native SL authors ++ Higher frequencies in translations from SL + Higher frequencies with bilingual authors + First occurrences in or shortly after SL contact + . . .The meaning of the constructions is ambiguous in both languages between a movement of going with a subsequent event and an aspectual interpretation, which is inchoative.This inchoative aspect -and the sudden incidence of an event assumed by Gougenheim for OF (cf.section 3.2.2) -is also a plausible interpretation for many occurrences of ME go + INF.The relation between inchoative aspect, immediate (or prospective) future and the continuous form is quite straightforward: In the terminology of Reichenbach (1947), "inchoative" means that the point of reference time (R) is situated at the beginning of the event time (E).This temporal link is expressed by the combination of go + INF, where the infinitive refers to the event and go is an auxiliary.The tense of go denotes the speech time (R in the present tense, prior to R in the past tense, etc.).With agentive subjects, the construction can express intention (He is going to kill him), with non-agentive subjects it can express certainty (He is going to die).
The aspectual constellation also explains the fact that the Modern English construction only appears in the continuous form, which represents the event "as if we put ourselves within the event and viewed it in its development.[. . .] The use of the progressive form represents the activity as a state."(Haegeman & Guéron, 1998 : 534f).This is what Jespersen calls a temporal frame around a given time.The development go+(to) + INF > be going to INF therefore occurs for independent reasons and is bound to the obligation to use the continuous form in this particular aspectual context. 12  On the basis of the investigation presented here, we can make the following assertions about the development of the ModE be going to + INF construction: A. The aspectual meaning of go+(to) + INF is due to one or more of the following factors: a. language contact, i.e.OF (Anglo-Norman) aler + INF forms; b. grammaticalisation of a movement verb similar to OF processes, but independent of them; c. language-internal lexical influence: confusion of ME gan 'go' with 'begin'.B. The development of go+(to) + INF to the ModE be going to + INF is an independent process, posterior to the development of the ME construction.
tr.Baeda's Hist.I.vii.(1890)38)    3. Expressing the relation of purpose, destination, result, effect, resulting condition or status. the notions of 'purpose' and 'motion so as to reach a place' or 'contiguity'.K.AELFRED,Gregory's Past.C.xliv.328) ) a. and b. are good examples for abstract aspectual uses situation, we gained the following picture: only few examples with the relevant order of the elements part of the ModE be going to + INF construction occur, and in these constructions theing form introduces a relative clause, so these examples are not relevant.As concerns the constructions go + to + INF and go + INF we found more examples

Table 2 :
gan + INF in OE texts All the occurrences show a non-complement infinitive just as the examples with gan + to + INF.A number of examples also occur with faran as shown in (11), a verb that in OE coexists with gan and expresses similar meanings: 'go, proceed, travel, march, sail' (cf.B&T); in the present tense it became obsolete before the 14th century (cf.OED online) 8 : (11) Se He faerð went to to sceawienne see his his tun.dwelling (AECHom_II,_26:214.39.4734)To summarise our findings for OE, four constructions could be the predecessor of the ModE be going to + INF construction: beon/wesan/(weorþan) + V-ende, gan + INF, gan + to+Vanne and faran + to + INF.
And for he was a gere goyng and comyng to Rome, þerfor he made to sle all þe chyldre þat wern two ger olde or within two.

Table 3 :
[Plinius]goenge to serche the secrete causes of the grauelles[Higd.(2):gravel;Larenarum]whiche be in the region ... was ... pressede to dethe with hepes of that grauelle.As concerns the constructions go + INF and go + to + INF in the PPCME2, we gained the following results: For go + INF we found 14 cases, for go + to + INF we found 111 cases including the yode and went forms (and its variants), the latter of which have to be considered too since, as already mentioned, this form starts to replace the yede forms in the ME period.The construction occurs with a wide range of verbs, e.g.perception verbs (see, hear), psychological verbs (know, learn, despise), verbs of movement (search, find, spring, pass), verbs of utterance (weep, cry, laugh), and verbs of communication(teach, speak, preach,  write, bid).Token frequencies for gon/yede/ went+(to) + INF in ME texts (PPCME2) is given below: (24) It m vij lasĳde haryng e blank e a vij lī ... It' il doit pur ij last de Blanke heryng xiij lī ... 'And 7 lasts of white herring at 7 pounds sterling. . .And he owes for 2 lasts of white herring 13 pounds sterling' [emphasis added] (London, Gilbert Maghfield's Account Book, 1392) : "if Old French can be said to have a continuous form at all, it is constructed with estre".It is true that unlike other Romance languages, OF has not developed constructions based on Latin stare to express the progressive aspect (e.g.Span.estarcantando,It.starecantando) or the immediate future (Span.estarparacantar,It. stare per cantare).Nevertheless, OF constructions with aler 'go' existed:Gamillscheg (1957 : 461)notes that aler occurs with infinitival complements (with and without the preposition a 'to'), as in the examples in (26) (our emphasis):(26) a. Jo irai lassus a Loëi parler, (Alisc.2540).b.Il vat avant la maison aprester, (Alexius 323).c.Veit la Guillelme, va li cheeir al pié, (Cor.Looïs 1726)

Table 4 :
Table 4 and show that the constructions are well established in OF and just about to develop in Middle English: independently of the presence or absence of to, the relative frequencies in OF are at least ten times higher than in ME.Relative frequencies of adjacent go + INF constructions in OF and ME In the OF Nouveau Corpus d'Amsterdam, most occurrences of aler + PREP + INF have the preposition pour 'for' and express a purpose (this is similar to the OE findings).There are only four occurrences of aler + a + INF, none of which suggests an immediate future-interpretation, e.g.(27a).The construction of directly adjacent aler + INF, without preposition, is far more frequent (1441 occ., e.g.(27b)), and this also true for aler + GERUND (852 occ.).(27) a. vn iur li ala a demander de sa aleine si ele ert puiant (Marie de France, Fabliaux [fablesA]) b. devant l empereor lo vunt araisoneir (Vie de saint Alexis [alexo])

Table 5 :
Extra-linguistic support for data-based hypotheses (SL=source language) With respect to structure, the OF construction aler + INF can be considered as a potential source of the ME constructions go + INF and go + to + INF, regardless of the presence of to, which can already be considered as an infinitive marker as we have shown in detail.