1 Introduction
In recent work on linguistic contacts between English and other European languages, a great deal of attention has been paid to (morpho)syntactic features which English shares with its neighbouring Celtic languages but not with (all of) its Germanic sister languages, and especially German (or at least ‘mainstream’ Standard German). This has then been interpreted by many as one piece of indirect evidence showing that English has not been so impervious to syntactic influences from Celtic as has traditionally been claimed (see, e.g., Tristram Reference Tristram1999; Vennemann Reference Vennemann and Brinton2001, Reference Vennemann, Lucas and Lucas2002; van der Auwera & Genee Reference van der Auwera and Genee2002; White Reference White2002; Filppula, Klemola & Paulasto Reference Filppula, Klemola and Paulasto2008). Areal-typological distributions can, indeed, serve a useful purpose in our efforts to shed light on the linguistic outcomes of the early contacts between English and Celtic, as well as other western European languages. This method is by no means new in historical linguistic studies and has been successfully applied, for example, to ascertain Finno-Ugrian substrate influence on Russian by the Russian-Finnish Slavist Valentin Kiparsky (see Kiparsky Reference Kiparsky1969). More recently, this approach has received a new impetus from the recent advances in typological research which has been able to extrapolate from typological and geographical patterns even in those cases in which no historical records or other types of direct evidence are available. A good example and of particular interest in this connection is the extensive work done on so-called Standard Average European (SAE) and its historical origins (see, e.g., Haspelmath Reference Haspelmath1998).
This article focuses on the so-called it-cleft construction (or ‘clefting’ for short), which is one of the many syntactic features shared by English and Celtic but has no obvious, or only partial, parallels in German in terms of its degree of grammaticalisation. The traditional term in Celtic linguistic scholarship is the ‘copula construction’, which is a reflection of the fact that the Celtic construction does not have an introductory ‘dummy’ pronoun but starts off with copula ‘be’ or, alternatively, directly with the focus constituent (see, e.g., Stenson Reference Stenson1981; Ó Siadhail Reference Ó Siadhail1989). What makes the cleft construction particularly interesting from an areal-typological and contact-linguistic point of view is the fact that, although it does not belong to the defining features of SAE as outlined by Haspelmath (Reference Haspelmath1998), among others, it is a robust feature of French, which is in the ‘nucleus’ of languages forming SAE alongside Dutch, German, and (northern dialects of) Italian. This in itself prompts the question of some kind of a historical-linguistic connection between the Celtic languages (both Insular and Continental), English, and French (or, more widely, Romance languages).
In the following, I will first discuss some of the defining characteristics of cleft sentences (section 2) and then move on to examine the emergence of it-clefting in earlier English (section 3). The aim of this part of my study is to find out, first, whether it-clefting exists in Old (and Middle) English, and secondly, if it does, to what extent it is used in these periods. As will become evident from the discussion below, the very existence of clefting in OE has been called into question by some scholars. The main databases used in this study are The York–Toronto–Helsinki Corpus of Old English Prose (YCOE) and The Penn–Helsinki Parsed Corpus of Middle English, second edition (PPCME2). Both are syntactically-annotated using the same form of annotation and can be accessed by the same search engine, called CorpusSearch. YCOE consists of 1.5 million words, while PPCME2 has approximately 1.2 million words of running text. After tracing the development of it-clefting in English, I will turn to the areal spread of clefting especially among European languages and also discuss some typological issues associated with clefting (section 4). This will pave the way for an attempt to determine to what extent the rise of clefting in English can be explained through contacts with neighbouring languages and especially with Celtic (section 5). Finally, the concluding section (6) seeks to piece together the various strands of evidence speaking for such contact influences.
2 Clefting
By ‘clefting’ I refer here to what is in English syntax more specifically termed ‘it-clefting’ or ‘it-clefts’, as opposed to ‘WH-clefts’ or ‘pseudo-clefts’ (see, e.g., Prince Reference Prince1978; Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985; Ball Reference Ball1991). It-clefts have a structure consisting of an introductory it (or sometimes that) is/was clause followed by an NP or an adverbial phrase in the focus position and a relative clause. Semantically and pragmatically, one can distinguish between two major types of it-clefts. The first, illustrated in (1) below, is ‘specificational’ and ‘stressed-focus’ in the sense that in this type the introductory part and the focused constituent provides the value or ‘new information’ (here: a bike) for the variable or the ‘given information’ expressed by the relative clause. The latter can also be called a ‘presupposition’ (here: He bought something).
(1) It was a bike that he bought (not a car).
Stressed-focus clefts must be kept apart from the superficially similar ‘predicative’ structures such as (2), which predicate some property (here: very nice) of the object under discussion (here: a bike) and do not therefore belong to the category of clefts. That said, the discussion below will show that the borderline between specificational (cleft) and predicative (non-cleft) structures is sometimes hard to draw especially when dealing with some of the Old English putative clefts.
(2) It/That was a very nice bike (that) he bought.
The second major type of it-cleft is similarly specificational but differs from the stressed-focus clefts in that the main ‘new’ information is contained in the subordinate clause, while the focused constituent serves to specify the temporal or other setting for the event, action or other kind of state of affairs expressed by the that-clause. Prince (Reference Prince1978), who first made this distinction, calls this type ‘informative-presupposition’ clefts, and I will follow her terminology here. An example of an informative-presupposition cleft is given in (3), in which the information contained in the relative clause comes as new information to the hearer:
(3) It was last year that he decided to buy a new bike.
Since this article is particularly concerned with the possibility of linguistic contact effects between English and Celtic, it will be useful to consider briefly the corresponding Celtic constructions; their possible influence on English clefting will be discussed in section 5 below. In Celtic languages in general, as noted above, clefting is realised by means of the copula construction, but no introductory pronoun is used and even the copula can be, and is, omitted in some of the Celtic languages and/or in some situations or registers. The prominent use of clefting probably goes together with the general rigidity of the word order in these languages, which makes the copula construction a particularly apt means of placing emphasis on some constituent of a sentence. Gregor (Reference Gregor1980) explains that for emphasis, the subject, object or other part of the sentence is put first, while the rest of the sentence is cast in the form of a relative clause, with the fronted constituent felt as being dependent on an ‘understood “It is”’ (Reference Gregor1980: 147). Thus, a sentence like It is the woman who came is rendered by the following constructions in Irish (Ir.), Scottish Gaelic (Sc. G.), Cornish (C.), and Breton (B.), respectively (examples from Gregor Reference Gregor1980: 147)
(4) Ir.: (Is é) an bhean a thainig ‘It is the woman who came.’
(5) Sc.G.: ’S e am boireannach a thàinig.Footnote 2
(6) C.: An venen a dheth.
(7) B.: Ar wreg a zeuas.
The bracketing in (4) shows that, in Irish, the copula construction need not have an overtly expressed introductory phrase. Indeed, it is often omitted because of the stressless nature of the copula, a feature which goes back to as early as Old Irish (Thurneysen Reference Thurneysen1975: 494). From the descriptions of Scottish Gaelic by Gillies (Reference Gillies and Ball1993: 209–12) and Lamb (Reference Lamb2001: 87–91) one can gather that the form with the copula is used even today. In present-day Cornish and Breton the sentence always starts directly with the fronted item followed by the relative clause. The same is true for Welsh, although Gregor does not cite here the parallel Welsh construction Y wraig a ddaeth (lit. ‘the woman who came’).Footnote 3 A further well-known characteristic of clefting in these languages is its syntactic freedom: even the verb can be put in the focus position, as is seen in (8)–(10) from Irish, Welsh, and Cornish, respectively:
(8) Ir.: (Is) ag leigheamh atá sé ‘Reading he is’ (lit. (it is) reading that he is).
(9) W.: Darllen y mae ef.
(10)
C.: Ow redya yma ef.
(Gregor Reference Gregor1980: 148)
Before proceeding to the areal and typological aspects of clefting, the relationship between clefts and relative clauses in both English and Celtic calls for a comment. They form partially overlapping constructions, and at least in English, the former can be considered to contain as one of its parts a relative clause.Footnote 4 In a historical perspective, this is confirmed by Ball's (Reference Ball1994a) comparative study of the development of clefts and relative clauses, which shows that complementisers of it-clefts and relative pronouns of restrictive relative clauses have followed similar paths of change over time, and what is more, at the same rate. This Ball takes to demonstrate that the two clause types are not distinct (see, however, Ball (Reference Ball1994b: 604, fn. 2) for discussion of views that do not treat it as a relative structure). The same approach is, indeed, adopted in the syntactic annotation of the YCOE and the PPCME2 corpora: in both, it-clefts are explicitly analysed as having the same internal structure as relative clauses. As for the Celtic languages, the situation is less straightforward and depends on what syntactic status is assigned to the constituent introducing the subordinate clause: that of a ‘relative particle’ or of a ‘complementiser’ or of something else (see below). Both share, however, the same type of specificational reading and can thus be considered to be semantically equivalent to each other.
3 Clefting and ‘cleft-like’ structures in Old and Middle English
As said above, my discussion is based on, and follows, the annotation of the YCOE and the PPCME2 corpora, where it-clefts (or structures analysed as such by the annotators) have been analysed as one type of relative clauses. This means that the internal structure is the same as in relative clauses in general, while at the external level clefts are at the IP-level and are not contained within their antecedents or coindexed to their ‘expletive’ subjects, which may be overt like hit ‘it’ and þæt ‘that’ or left empty. Despite this variability in form, all are parsed as clefts and labelled as ‘CP-CLF’ to mark them off from other types of relative clauses. The paradigmatic examples of it-clefts given in the reference manuals of YCOE and PPCME2 include sentences like (11) and (12), respectively. Note that they are here (as in all the other examples further below) given in conventional orthography, and the clefted parts are italicised. The text identification codes are given in brackets after each example.
(11) . . . hit wæ;re Swyðun se ðe hine lærde mid þære halgan lare and þone ðe he geseah on ðære cyrcan swa fægerne
‘. . . it was Swithun who had provided him with the holy teaching and whom he had seen so beautiful in the church.’
(ID coaelive, ÆLS [Swithun]: 388.4472)(12) . . . for it was I myself that cam in the likeness
‘. . . for it was I myself that came in the likeness.’
(ID CMMALORY, 5.131)
The parsed output structures for these examples are given in the Appendix. I now move on to the results of my search for OE clefts in YCOE.
3.1 Old English it-clefts
The corpus search yielded a total of 89 instances of structures (tokens) parsed as clefts in accordance with the above-mentioned criteria, which gives a normalised frequency of 5.9 instances per 100,000 words. Clefts were found in 17 out of the 108 text samples in the corpus.Footnote 5 However, it is noteworthy that more than half of these (50 tokens) occurred in just one text, viz. that from Bede, who makes frequent use of what the corpus compilers term ‘time-clefts’ such as the one in (13):
(13) þa wæs æfter hire deaðe þæt þa broðor oðerra weorca swiður gemdon & þisse cirican timbro forlæton seofon gear.
‘Then [it] was after her death that the brethren were more occupied with other works, and for seven years neglected the erection of the church.’Footnote 6
(ID cobede, Bede_3:6.176.8.1722)
This type of cleft can be described as (Adv-Time) BE Prep Phrase Rel-clause (for ease of reference, the descriptions are given in ModE notation), and as the gloss indicates, the expletive subject position is empty. This variant is by far the most frequent in YCOE, accounting for well over half of all instances (in Bede's text, all but one instances are of this type, including the negative and other variants mentioned below). In some of the instances, there is no overt relative marker in the relative clause, but they have nevertheless been parsed as clefts. An example is given in (14):
(14) þa wæs æfter Æðelberhtes forðfore, Eadbald his sunu feng to ðam rice, . . .
‘Then it was after Æthelberht's death that his son Eadbald succeeded to the throne, . . .’
(cobede, Bede_2:5.110.22.1042)
Yet another subtype of time-clefts has no adverb of time in initial position, with the sentence starting off directly with the copula. The expletive subject is empty here, too, e.g.:
(15) Wæs on þyssum dæge þæt ure Drihten Hælend Crist þa menniscan gecynd þe he genam to his godcundnesse ahafen him sylfum ofor heofonas & ofor ealle engla þreatas
‘It was on this day that our Lord and Saviour Christ exalted the humanity that he united to his divine nature above the heavens and above all the hosts of angels, . . .’Footnote 7
(coblick, HomS_46_[BlHom_11]:115.3.1474)
From the semantic and pragmatic point of view, ‘time-clefts’ clearly belong to the class of ‘informative-presupposition’ clefts by virtue of the fact that the that-clause represents new information to the reader, with the introductory part defining the temporal setting within which the event or action described in the subordinate clause is said to take place.
The rest of the tokens in the database assume many different forms, which are briefly described in the following.
PrepPhrase BE that Rel-clause
This type has an adverbial expression (other than time) in initial position and no overtly expressed expletive subject. All 8 tokens were concentrated in one text, viz. Cura Pastoralis, King Alfred's West Saxon version of Gregory's Pastoral Care. An example is given in (16):
(16) For ðissum ðingum wæs ðætte Crist self, ðe swiðe hrædlice meahte getrymian ðone ðe he wolde, sealde bisene ðæm lareowum to ðæm ðæt ða unlæredan ne scoldon læran:
‘For this reason it was that Christ himself, who could very quickly strengthen whom he wished, gave an example to teachers to the effect that the uneducated should not teach.’Footnote 8
(cocura, CP:49.383.36.2599)
There were no instances in the corpus of the inverted version (BE PrepPhrase that Rel-clause).
It/That BE NP/PrepP Rel-clause
This pattern is closest to the ModE specificational it-clefts by virtue of having all the same elements in the same order. There were 10 instances, all of which had introductory hit. The instances were spread across 10 texts, which shows that, albeit rather infrequent, this pattern was relatively established in OE. The example in (17) has a noun phrase in focus position, while a prepositional phrase occurs in (18). The latter is also a clear example of an informative-presupposition cleft, as the that-clause contains new rather than presupposed information in this context. It is also noteworthy that both are from texts that are not translations from Latin.
(17) Þa cwæð Iohannes to Petre. þæt hit wære se hælend þe on ðam strande stod;
‘Then said John to Peter that it was the Saviour that stood on that strand.’
(cocathom2, ÆCHom_II,_17:164.118.3653)(18) Nu hit is for eahta and þryttiðan gearan þæt min dohtor me losode,
‘Now it is 38 years since I lost my daughter.’
(coeuphr, LS_7_[Euphr]:270.283)
Some instances with negation also occurred. An example is (19), which in turn is from a text translated from Latin:
(19) Ac hit ne sind na steorran þæt ðær feallað,
‘But it is not the stars that are falling there, . . .’
(cotempo, ÆTemp:9.1.286)
It/That NP BE Rel-clause
The pattern here is the same as the previous one, but has inversion in the introductory phrase, e.g.:
(20) . . . and þæt hus afylde mid ormætum stence, þæt man eaðe mihte witan þæt hit se deofol wæs þe hine dwelian wolde; . . .
‘. . . and that house filled with intense odour so that one could easily know that it was the devil who wanted to mislead him; . . .’
(coaelive, ÆLS_[Martin]:770.6456)
BE(Neg) It/That NP Rel-clause
This is yet another variant of the ‘modern’ type, with the introductory phrase starting with the (negated) copula, e.g.:
(21) . . . þa cwædon þa geleaffullan nis hit na Petrus þæt ðær cnucað,
‘. . . then said the faithful: it is not Peter that is knocking there, . . .’
(cocathom1, ÆCHom_I,_34:474.247.6867)
BE(Neg) NP Rel-clause
This could be considered a truncated variant of the previous type, with no overtly expressed expletive subject, e.g.:
(22) . . . ac is seo lyft þe we on lybbað on ðisum deadlicum life.
‘. . . but it is the air that we breathe to live in this mortal life.’
(cotempo, ÆTemp:10.7.318)
A negative version was also found, witness (23):
(23) Ne wæs þa elding, þætte Gode se leofa biscop Eadbryht wæs mid grimre adle ðread & gestanden;
‘It was without delay that bishop Eadberht, beloved of God, was seized with an attack of severe illness.’Footnote 9
(cobede, Bede_4:31.376.31.3771)
On the other hand, it is noteworthy that no instances were found of the inverted type NP/Pron BE Rel-clause, which is noted and discussed in Ball (Reference Ball1991: 45) as a pattern that resembles the ModE pseudo-cleft construction (see below).
Interrog. Pron/NP BE Rel-clause
This pattern can be considered a subtype of the previous one, with an interrogative pronoun (phrase) in the focus position. Consider (24)–(25):
(24) La hwilc burhscir wæs þe nolde bisceop geceosan of Martines mynstre, for his mærum gebysnungum?
‘Which city was it indeed that did not want to choose a bishop from Martin's monastery, because of the illustrious example he set?’
(coaelive, ÆLS_[Martin]:339.6176)(25) Hwylc is of us Drihten þæt hæbbe swa hwite saule swa þeos halige Marie?
‘Which of us, Lord, is it that has as white a soul as this holy Mary?’
(coblick, LS_20_[AssumptMor[BlHom_13]]:147.168.1817)
As this brief survey shows, OE could avail itself of a wide range of clefts displaying different kinds and orders of the key constituents but sharing the same basic structure. A notable exception is clefts with a personal pronoun in focus position, which were completely absent from YCOE. Ball (Reference Ball1991: 41) comments on this type by first noting that it emerges late in the OE period and is apparently rare. She finds only three examples in her database, all of which have the third singular pronoun (he) in focus position. One of these, given in (26) below, is from a (probably) late tenth-century text known as Passion of St Eustace and is introduced by hit, while the other two were found in the Blickling Homilies, also dating to the tenth century.Footnote 10 Though few in number and despite the inverted word order, these examples suffice to show that pronoun-focus clefts were available in OE. The possible contact-linguistic significance of this will be discussed in section 5.
(26)
‘and asked him whether it were he who formerly was the teacher of their soldiers; he then denied that it was he.’
(LS 8 (Eustace) 272; cited from Skeat 1890 in Ball Reference Ball1991: 41)
Semantically and pragmatically, the clefts found in YCOE can be classified in modern terms as representing either specificational stressed-focus or informative-presupposition clefts, hence functionally rather versatile. From the contact-linguistic point of view it is useful to know that, although the majority of instances were found in texts based on Latin originals (mainly because of the predominance of ‘time-clefts’ in Bede), as many as 15 occurred in non-translations and were spread over 6 different texts.
3.2 Middle English it-clefts
A search similar to the one made for OE clefts was run on the PPCME2 corpus. It yielded 96 tokens of clefts, which gives a normalised frequency of 8.0 per 10,000 words as against 5.9 for the OE clefts. This indicates a noticeable rise in the frequencies of clefts from OE to ME. However, more important than the absolute or normalised frequencies is the fact that in PPCME2 clefts were more evenly distributed between the text samples, appearing in 27 out of the 56 texts (17 vs 108 in YCOE). This in itself is a clear indication of a general increase in the rates of use of clefting in the ME period. Furthermore, there was no such concentration of any one type of cleft in just one text as was the case in YCOE. Some quantitative and qualitative differences also emerged between the clefts in YCOE and PPCME2 that suggest a more advanced stage in the grammaticalisation of this feature in the latter period. Thus, ME clefts seem to allow pronouns to occur in the focus position more readily than their OE counterparts, either without inversion in the matrix clause, as in (27), or with it, as in (28):
(27)
It was he þat counceled þe Senate þat Cartage schuld not be distroyed.
(CMCAPCHR, 44.388)(28)
And eftyr this techynge, hamelye, curtayse, and blysfulle and verray lyfe, ofte tymes oure lorde Ihesu sayde to me, ‘I it am that is hiaste . . .’
(CMJULNOR, 59.256)
There were several other instances from a number of texts, which further confirms the more established nature of this type of cleft, as compared with OE. Another noticeable difference concerns the relaxation of the constraints governing the syntactic type of focused constituent: objects and adverbial prepositional phrases now seem to occur more freely in this position than in OE. Examples of these are the following:
(29)
. . . for it is the develes hoord ther he hideth hym and resteth.
(CMCTPARS, 316.C2.1205)(30)
Me troweþ þat by þe prayers of þis holy mayde it is þat þat place was nevere ʒit destroyed.
(CMPOLYCH, VI, 125.869)
From the semantic and pragmatic point of view, most of the PrepP focus clefts are informative-presupposition clefts. Ball (Reference Ball1991: 45) considers them to be an innovation in ME but, as was seen above, a good few of them occurred in the OE texts, too, although they were there mostly ‘time-clefts’ of the type found in Bede.
Yet other new features of clefts include the use of WH-pronouns in introducing the relative clause, as in (31). On the other hand, þe shows a clear decline and disappears altogether from the later ME texts.
(31)
þis chastisyng schal enduryn xij days tyl þu wyl beleuyn þat it is God whech spekyth to þe & no deuyl.
(CMKEMPE, 146.3375)
In most other respects, the ME clefts found in PPCME2 exhibit the same essential characteristics as their OE counterparts. The dialectal distribution of the occurrences does not allow any straightforward conclusions, but it is of some interest to note that as many as 24 were found in texts from the West Midlands or the northern areas, with many of these dating to the early part of the thirteenth century (e.g. those from Ancrene Riwle and St Katherine).
3.3 Summary and discussion of OE and ME clefts
Although some of the structures parsed as clefts in YCOE, and especially the type of time-clefts found in Bede's texts, may be open to differing interpretations (see Ball Reference Ball1991 for discussion), there is enough ‘clear’ evidence to show that it-clefts, including even those with pronoun focus, have a certain presence already in OE grammar. On the basis of the present data, there is a noticeable rise in the frequencies of clefts in ME. Also, instances of clefts are at that stage more evenly distributed across different texts, syntactic constraints on the focus constituent are relaxed, and certain types (such as pronoun-focus clefts, in particular) become clearly more established than in OE. This I take to be a clear indication of the gradual grammaticalisation of clefting, starting already in OE and gaining momentum in ME, which is significant from the point of view of the possible contact influences to be discussed in section 5 of this article.
It is noteworthy that the evidence from YCOE and PPCME2 is in some contrast with some previous works in which it-clefts in OE, in particular, have been described as having been nonexistent or rare at best. For example, Mitchell (Reference Mitchell1985: §1486) states that there are no examples in OE of sentences with anticipatory it such as ModE It's food that I want. He does note Visser's (1963–73: § 63) examples with þæt as the introductory pronoun in cleft-type constructions but evidently does not consider them examples of ‘proper’ clefts. However, as has been seen, YCOE contains several clear examples of clefts with hit that can be considered specificational in meaning and hence genuine instances of it-clefts rather than other structures with referential it.
Visser himself adopts a more flexible stand than Mitchell. His paradigmatic (modern English) example of a clefted structure is ‘It is father who did it’, under which he lists examples of cleft constructions from OE onwards. He further notes that in OE introductory hit is sometimes omitted or, in some cases, replaced by þæt (Visser Reference Visser1963–73: § 63). Traugott (Reference Traugott and Hogg1992) mentions Mitchell's negative stand on the existence of hit-clefting in OE but draws attention to Visser's examples with anticipatory þæt. However, she stops short of passing judgment on whether they should be considered similar to the structures introduced by anticipatory it (Traugott Reference Traugott and Hogg1992: 280).
By far the most thorough treatment of clefts in OE and ME is Ball (Reference Ball1991), who presents a detailed analysis of ‘cleft-like’ constructions from their earliest beginnings. Her discussion confirms that the cleft construction (including some borderline ‘cleft-like’ structures) is an innovation in OE but that its frequencies remain very small in that period. In line with Mitchell (Reference Mitchell1985), she finds no hit-clefts which would perfectly match the stereotypical Modern English it-cleft with specificational reading, and especially of the ‘stressed-focus’ type (Ball Reference Ball1991: 45). She does, however, note one possible exception, which according to her could be considered an instance of it-clefting in this sense. This is the example cited, but discarded, by Mitchell (Reference Mitchell1985: § 2135) from Ælfric. It is, incidentally, also included in the data from YCOE:
(32) þa cwædon þa geleafullan,
‘Nis hit na Petrus þæt þær cnucað, ac is his ængel.’
not-is it-n. not Peter-m. REL-n. there knocks but is his angel-m.
‘Then the faithful said: It isn't Peter who is knocking there, but his angel.’
(ÆCHom I.517–18.1; here cited from Ball Reference Ball1991: 39–40)
Mitchell takes hit and not the focused noun, as in clefts, to be the antecedent of the relative pronoun here and therefore rejects this as an example of a cleft sentence (Mitchell Reference Mitchell1985: § 2135; Ball Reference Ball1991: 39–40).
Another type of OE cleft which Ball considers as a possible parallel to the Modern English specificational clefts is the pattern she labels as NP BEON REL-CLAUSE, as in the following example she cites from Visser (Reference Visser1963–73: 49):
(33) . . . min fæder is þe me wuldrað
my father is that me glorifies
(Jn 8.54: . . . est Pater meus, qui glorificat me . . .)
‘It is my father that glorifies me’ (Jn 8.54)
(ÆCHom ii. 234.3; cited in Visser Reference Visser1963–73: 49)
Unlike Visser, who treats (33) as a hit-cleft, with hit omitted, Ball is inclined to label it as a pseudo-cleft, although she acknowledges that it could also be analysed as a hit-cleft (Ball Reference Ball1991: 27–8). As was mentioned above, my YCOE data did not contain any examples of this type, but the one with no inversion did occur there (i.e. the pattern BE(Neg) NP Rel-clause). Ball also cites instances of the latter, e.g. from Ælfric's Lives of Saints (Reference Ball1991: 27). Her conclusion with regard to the NP BEON REL-CLAUSE construction is that, whether cleft or pseudo-cleft, it is rare in OE and, in her data, occurs only in translations of a Latin headless relative. However, she notes that Ælfric avoids it in some cases even in translations and uses a simple sentence instead of the Latin-type construction. This leads her to state that ‘in Late West Saxon, at least, NP/PRO BEON REL-CLAUSE was not the preferred construction for marking focus and open proposition’ (1991: 52).
As mentioned above, no instances of pronoun-focus clefts were found in YCOE, but a small number of examples occur in other OE texts. Ball (Reference Ball1991: 41) considers them pseudo-clefts rather than it-clefts and adds that pronoun-focus clefts identical in their surface-ordering of elements to the ModE ones are not attested until the early thirteenth century (Reference Ball1991: 45). In my view, the OE example in (26) above and the two others discussed in Ball (Reference Ball1991: 41f.) can be equally analysed as it-clefts, which – on the basis of the glosses at least – also appears to be the stand adopted by W. W. Skeat and Richard Morris, the editors of the texts in which these examples occur.
Another point on which my account seems to differ from Ball's concerns the so-called ‘informative-presupposition’ hit-clefts (in which the relative clause represents new rather than given information). As noted above, many of the examples from YCOE fulfil the criteria for this type of cleft, despite the fact that they are syntactically more restricted than their ModE counterparts, with most of them having an adverbial of time in the focus position. According to Ball, however, informative-presupposition clefts do not emerge until Late ME (Reference Ball1991: 46; the same position is repeated in Ball Reference Ball1994b). Yet she, too, cites an earlier example such as (34) from Whitelock (Reference Whitelock1930). Ball considers it a ‘predecessor’ of the Modern English adverbial/PP-focus it-clefts (Ball Reference Ball1991: 45):
(34) ##her swutelað on þison gewrite hu Æðelred kyning geuðe þ Æerices cwyde æt Boccinge standan moste. hit wæs manegon earon ær Æðeric forðferde þ ðam kincge wæs gesæd þ he wære on þam unræde þ man sceolde on Eastsexon Swegen underfon ða he ærest þyder mid flotan com . . .
‘It is shown here in this document how King Ethelred granted that the will of Æthelric of Bocking should stand. It was many years before Æthelric died that the King was told that he [ = Æ.] was concerned in the treacherous plan that Swegn should be received in Essex when first he came there with a fleet . . .’
(W 16(2); cited from Whitelock Reference Whitelock1930: 45 in Ball Reference Ball1991: 45; ‘##’ = discourse-initial)
Despite the aforementioned differences in interpretation and dating of the emergence of certain types of clefts, Ball's account and mine based on the data from YCOE corroborate each other. Though still relatively sporadic in OE, a variety of different types of clefts were used in OE, and these continued to be used and became more frequent and both functionally and syntactically more versatile in ME. Next, having established the presence of clefting in the grammar of English from early on, we can turn to cross-linguistic aspects of clefting to see if they could shed light on the rise of clefting in English.
4 Areal and typological aspects of clefting
A suitable starting point for our inquiry is the recently published World Atlas of Syntactic Structures (WALS), which does not, unfortunately, contain clefts of any sort among the features covered. However, some information about the areal distribution of clefting among European languages can be gleaned from individual studies, such as that by Heinrich Wagner (see Wagner Reference Wagner1959). He first notes that what he calls mise en relief, i.e. clefting in our terminology, is realised in exactly the same way in Celtic languages and Berber languages (which also belong to VSO languages). He further observes a clear ‘geolinguistic connection’ (ein sprachgeographischer Zusammenhang) between the French mise en relief construction, i.e. c'est-clefting, and its Insular Celtic parallels (Wagner Reference Wagner1959: 173ff.). This is interesting, given that French is considered to be part of the ‘nucleus’ of Standard Average European, while Celtic is firmly outside of it (Haspelmath Reference Haspelmath1998), and, as mentioned above, clefts (of any type) are not among the set of syntactic features characterising SAE. For some reason, Wagner stops short of commenting on possible implications of these observations for the rise of clefting in English but draws attention to the frequent use of clefts in the English of Ireland, Hiberno-English. This tendency can according to him be explained by the corresponding Irish usage.
An even more far-reaching ‘geolinguistic connection’ involving clefting as one of its central features is proposed by Wehr (Reference Wehr, Eichner, Mumm, Panagl and Winkler2001). She suggests that the Celtic and Romance languages, especially French and Portuguese, share a number of syntactic and other features which all result in some way or another from a tendency she describes as ‘weakening of the individual word’ (Schwächung des Einzelwortes). This involves several phonological processes such as sandhi phenomena, enchaînement (‘chaining’), liaison, elision, and fusion. The loss of autonomy of the individual word then explains the prominent status of clefting in all of the mentioned languages, which according to her form the core of what she terms a ‘Western Atlantic Sprachbund’ (ein westlich-atlantischer Sprachbund). Languages like English display mixed behaviour in this respect, as English, for example, preserves the possibility of ‘word accent’ as a means of expressing emphasis, with clefting being just an optional but not a necessary device for that purpose. German, Italian or even Latin are also languages where emphasis can be conveyed by prosodic means alone, although they too have clefting as an optional feature (Wehr Reference Wehr, Eichner, Mumm, Panagl and Winkler2001: 255). In German, however, clefting is ‘unidiomatic’ or at best rare in the spoken language according to Wehr (Reference Wehr, Eichner, Mumm, Panagl and Winkler2001: 272).
The marginal status of clefting in German is confirmed by Steinberger's (Reference Steinberger1994) comparative study of word-order variation in German and English. He notes that although the construction Es is X, der . . . ‘is possible in German, it seems to be less frequently used than in Romance languages and English’ (Steinberger Reference Steinberger1994: 87). According to him, the same holds even for the pseudo-cleft construction of the type Was mich ärgert, is der Larm (‘What annoys me is the noise’), for which there is ‘no need’ in German (Steinberger Reference Steinberger1994: 88; see also Miller Reference Miller1996: 115–17). It is true that Hermann Paul, in his Deutsche Grammatik (see Paul Reference Paul1919: 64), cites some examples of cleft-like structures (Umschreibungen in his terminology) which can be used to give prominence to what he calls the ‘psychological predicate’. However, some of his examples are predicative rather than specificational; as regards some others which are indeed specificational, Paul himself notes that they arose in the eighteenth century as a result of French influence and are thus fairly recent innovations.Footnote 11
The marginal status of clefting in German even as compared with English is also confirmed by a quantitative study by Ball (Reference Ball1990), whose comparison between the translations of the Gospels in several languages showed that where Modern English, French, and Danish used it/ce/det-clefts, Modern German mostly opted for a simple sentence without clefting (Ball Reference Ball1990; my source here Ball Reference Ball1991: 81). The contrast between German and other western and northern Germanic languages with respect to this feature is also noted by Ahlqvist (Reference Ahlqvist2002) in his survey of clefting in European and other languages. In Modern Swedish, as Ahlqvist points out, the cleft construction is quite common and has roots extending back to the late medieval period. Here he relies on Natanael Beckman's (Reference Beckman1934) study of early Swedish and other Nordic and western European languages where Beckman dates the first occurrence of clefting in Swedish to the fourteenth-century poem Erikskrönikan, so considerably later than in English.
To turn next to Romance languages, it has already become evident that clefting is a prominent part of the syntax of French and Portuguese. The French cleft construction bears a close resemblance to the English one, having an introductory ‘dummy’ pronoun (ce) similar to the English it. There is no such pronoun used in Portuguese or Spanish, which also has clefting but in which its use is more restricted than in Portuguese (see, e.g., Pinedo Reference Pinedo2000). Similar if not even more severe restrictions apply to clefts in Italian, which rather resorts to what Belletti (Reference Belletti2006) has called ‘subject inversion’ in situations where French would use ce-clefting. As regards the ancestral language, Latin, Löfstedt (Reference Löfstedt1966) shows that it had clefting as part of its syntactic repertoire. However, he notes its relatively rare use in Latin texts (Reference Löfstedt1966: 259) and relatively unestablished nature, as compared with French ce-clefting:
Es handelt sich höchstens um sporadische Ansätze, nicht um einen voll ausgebildeten Sprachgebrauch wie im Französischen.
(Löfstedt Reference Löfstedt1966: 260–1)Löfstedt further points out possible ambiguities between cleft and ‘determinative pronoun’ readings of some examples (Reference Löfstedt1966: 261f.) and that cleft-like structures are mostly used with demonstrative and interrogative pronouns rather than nouns or personal pronouns (Reference Löfstedt1966: 276). What is particularly interesting from the areal-linguistic point of view is Löfstedt's observation that no similar construction exists in Romanian (Reference Löfstedt1966: 256, fn. 5).
So far, then, our survey lends support to Wehr's suggestion about a western European areal distribution of clefting. This is also backed up by the fact that, further east, Slavic languages generally make little or no use of clefting. For Russian, for example, Miller (Reference Miller1996: 120) states that it ‘does not have a construction parallel to the English it-cleft’, but uses other means such as combinations of word order and deictic particles.Footnote 12 The same is true of such non-Indo-European languages as Finnish, Estonian, and Hungarian where thematic prominence is expressed by word-order shifts, particles, and prosodic means (on Hungarian, see also Miller Reference Miller1996). Ahlqvist (Reference Ahlqvist2002) notes its presence in Finnish, but points out that it is probably due to Swedish influence. Outside Europe, it is found, for example, in Maori (see Ahlqvist Reference Ahlqvist2002) and some African languages (see, e.g., Holm Reference Holm1988: 212; Sebba Reference Sebba1997: 187–8). Interestingly, it is also a feature of several English-based creoles (Sebba Reference Sebba1997: 187–8). In the context of Europe, however, there is a clear western-European bias in the geographical distribution of clefting (for a similar result, see Miller Reference Miller1996).
Many linguists have wanted to see a typological connection between rigidity of word order and clefting and sought to explain the spread of the cleft construction as a reflex of increasing constraints on word-order arrangements. For example, Jespersen (Reference Jespersen1937: 86) elaborates on this idea as follows:
In some, though not in all cases, this construction may be considered one of the means by which the disadvantages of having a comparatively rigid grammatical word-order (SVO) can be obviated. This explains why it is that similar constructions are not found, or are not used extensively, in languages in which the word-order is considerably less rigid than in English, French, or the Scandinavian languages, thus German, Spanish and Slavic.
In a similar vein, Ahlqvist (Reference Ahlqvist2002) considers clefting to be a particularly suitable means of expressing emphasis in strict VSO languages such as the Celtic languages. However, for English the same kind of connection between word order and clefting has not been universally accepted. For example, in her account of the rise of it-clefting in English Ball (Reference Ball1991) writes that the relationship between rigidity of word order and increasing frequencies of clefts from the EModE period onwards ‘is unlikely to be straightforward’ (Reference Ball1991: 497), because, as she puts it, ‘alternatives involving movement have long since disappeared’ (1991: 518). As more probable factors explaining the spread of clefts in later English she singles out the invention of printing, the rise of literacy, and the development of the written language; parallel developments in French and Scandinavian languages are also potentially relevant (ibid.). While Ball may be right insofar as the modern-period expansion of the use of clefting is concerned, word-order considerations must be considered to be one of the factors accounting for the emergence of clefting in the earlier stages of the language.
5 Contact-linguistic aspects of clefting
An areal connection is in itself indicative of contact effects. Ahlqvist (Reference Ahlqvist and Hopper1977) takes up the idea of an areal and early contact connection between those western European languages that have grammaticalised the cleft construction, i.e. the Celtic languages, English, and French, and suggests that in these languages it may ultimately derive from Celtic where it was attested earlier than in any other of these languages.
Chronological precedence must, indeed, be considered a major criterion for ascertaining contact influences, if other social and linguistic conditions such as those discussed by, for example, Thomason and Kaufman (Reference Thomason and Kaufman1988) are also met. In the case at hand, there are sufficient grounds for arguing that the Celtic copula constructions clearly antedate the English clefts and are both structurally and functionally close enough to their English counterparts to be transferable in a contact situation – as has, indeed, been the case with the modern-period ‘Celtic Englishes’ (see, e.g., Filppula Reference Filppula1999). Clefting has a long history in the Celtic languages, reaching most probably further back than their recorded history. In his description of Irish syntax, Mac Eoin (Reference Mac Eoin and Ball1993: 137) states that clefting is ‘a very common construction . . . at all periods’. Thurneysen (Reference Thurneysen1975: 492–4) cites examples like is hé día as éola indium-sa ‘it is God who is knowing in me’ from the mid-eighth-century Würzburg Glosses, and similar constructions are found in the somewhat later Milan Glosses. Further documentation is provided by Pedersen (Reference Pedersen1913: § 538, p. 223), who discusses early Irish cleft constructions under the heading of ‘Relativsätze’. In more recent research, Mac Cone (Reference Mac Cone1979: 13f.) presents evidence showing that the use of cleft sentences for marking a constituent arose even before the generalisation of the initial verbal complex. In the same vein, Mac Coisdealbha's (Reference Mac Coisdealbha1998) detailed study of Old Irish syntax leaves no doubt as to the prominent role of the cleft sentence there. According to him, the number of instances of clefts in the Würzburg material alone is ‘an imposing ca. 300’ (Mac Coisdealbha Reference Mac Coisdealbha1998: 143), divided between several formal classes depending on the type of constituent put in the focus position: nominal or pronominal phrase (both in subject or object position), prepositional phrase, adjective, adverb, subordinate clause, etc. Yet another work dealing with clefts in Irish (and other languages) is Ahlqvist (Reference Ahlqvist2002), who also confirms that clefting has a robust presence in O.Ir. texts.Footnote 13
Our sources for Old Welsh syntax and word-order patterns are much scantier than those for Old Irish, as there is a shortage of prose texts especially, and many of the extant prose works are either glosses on Latin texts or otherwise rely heavily on Latin models (see, e.g., Willis Reference Willis1998: 7–11). Yet there is enough evidence to show that clefting has been a characteristic feature of the Brythonic branch of Celtic from very early on. Thus, Pedersen (Reference Pedersen1913 § 547: 239) writes that both Irish and British (i.e. the ancestor of Welsh) use what he calls ‘relative inversion’ (G. Umschreibung) for the purposes of emphasis:
Soll ein betontes Wort des Nachdruckes wegen an der Spitze des Satzes stehen, so tritt eine relative Umschreibung ein: das hervorzuhebende Wort wird Prädikatsnomen der (oft nicht ausgredrückten) Kopula und Beziehungswort eines Relativsatzes, der die eigentliche Aussage enthält.
(Pedersen Reference Pedersen1913 § 547: 239; emphasis original)[‘Should a stressed word stand at the beginning of a sentence because of emphasis, a relative inversion is used: the word to be emphasised becomes the predicate of the (often unexpressed) copula and the antecedent of a relative sentence, which contains the actual proposition.’]
Pedersen goes on to state that in Irish the division of a sentence into two parts is particularly clear, whereas in British it is less noticeable because the copula is mostly omitted and the degree of emphasis on the fronted item is often very small. Indeed, with fronted subjects and objects it is only the presence of a before the verb which reveals the relative structure of the sentence, e.g. seith meib a oed idaw ‘sieben Söhne waren ihm’ [lit. seven sons REL were to-him, i.e. ‘he had seven sons’] (Mabinogion/The Red Book of Hergest, 193; cited in Pedersen Reference Pedersen1913 § 547: 239). While Pedersen treats the subordinate part of the sentence as a relative structure introduced by the relative particle a, in most of the more recent works it is simply labelled as a ‘particle’ (see, e.g., Willis Reference Willis1998). ‘Relative inversion’ has also disappeared from current terminology, which prefers either the term ‘abnormal sentence’ (see, e.g., Mac Cana Reference Mac Cana1973) or, in some of the more recent works, ‘cleft sentence’ or ‘clefting’ (Watkins Reference Watkins, Fife and Poppe1991; Tristram Reference Tristram2002). For Willis (Reference Willis1998), however, the abnormal sentence represents a ‘pseudo-cleft mixed sentence’ rather than clefting proper (Reference Willis1998: 8).Footnote 14 Whichever term is used does not change the fact that clear examples of clefting are found in Welsh from very early on. In fact, Tristram (Reference Tristram2002: 132, 134) cites even earlier examples of clefting than Pedersen from the tenth-century prose work entitled the Computus Fragment, e.g. is did ciman ha c(e)i (Comp. 2) ‘it is a full day that thou wilt get’, Is Aries isid in arcimeir [.e.] (Comp. 9) ‘It is Aries that is opposite [e]’.Footnote 15
Further documentation of the early occurrences of clefts in the Brythonic branch is provided by Evans (Reference Evans1964/1994), who gives examples of what he terms ‘the mixed [word] order’ in a number of Middle Welsh texts, the earliest of which date from the eleventh century. For instance, in the collection of tales collectively known as The White Book Mabinogion there are sentences like ys mi a'e heirch ‘it is I who seek her’ (479.29),Footnote 16 and in another early text, Ymddiddan Myrddin a Thaliesin, examples like Oed maelgun a uelun in imuan ‘It was Maelgwn that I could see fighting’ (57.5) (Evans Reference Evans1964/1994: 140–1). As Evans notes, the copula which still occurs in these examples was eventually dropped before the word or phrase to be emphasised, leading to the later Middle Welsh and Modern Welsh constructions without the copula, as exemplified above (Evans Reference Evans1964/1994: 141; see also Watkins Reference Watkins and Martin1993: 336–7). Evans’ first example is particularly important in this context, as it shows that personal pronouns could occur in the focus position of early Welsh clefts.
What is interesting from an areal-linguistic point of view is that the Celtic roots of clefting in western European languages may, in fact, go even as far back as Continental Celtic. As mentioned above, Wagner (Reference Wagner1959) sees a connection between Celtic and French in this respect, and more recently, Lambert (Reference Lambert2003) has mooted the possibility of Gaulish influence on French clefting:
La formule française ‘c'est . . . qui (que)’ en [of the ‘phrase clivée’] est l'équivalent exact: irl. mod. Is é a rinne é ‘c'est lui qui l'a fait’. On s'est demandé si la formule française était due à un substrat celtique: jusque ici aucun texte gaulois n'en a apporté le témoignage.
(Lambert Reference Lambert2003: 70–1)Although no agreement has yet been reached on this question, there seems to be some evidence from Gaulish inscriptions suggesting that Continental Celtic did, indeed, make at least some use of clefting and fronting (or, more generally, ‘topicalisation’) devices (for discussion, see, e.g., Evans Reference Evans, Bammesberger and Wollmann1990: 168). Perhaps the most vocal exponent of this view is the American Celticist John T. Koch (see, e.g., Koch Reference Koch, John1985). On the basis of his analysis of the so-called Alise-Sainte-Reine inscription as well as of the Chamalières tablet, he concludes that Gaulish must have had clefting and topicalisation patterns which were constrained ‘movements’ leaving the main structures of the sentence unaffected. On his account, then, Gaulish could place an NP in initial position either by means of the copula or without it (or even dislocate it to the rightmost position in the sentence); in the former case, this might or might not be accompanied by relative morphology of the verb, whereas this did not happen in the latter case (Koch Reference Koch, John1985: 34). Koch's interpretation of some of the crucial forms in the Chamalières tablet has not, however, been generally accepted and the question thus remains open (see esp. Lambert Reference Lambert1987).
Apart from Celtic influence on clefting in western European languages, which looks more than likely despite the lack of ‘hard’ evidence from Continental Celtic, one has to reckon with the possibility of Latin influence on both medieval French and English, as well as possible French influence on English. Indeed, Löfstedt's (Reference Löfstedt1966) study mentioned above seeks to show that the French c'est lui qui l'a fait construction type is based on the model of Latin because it can be traced back to the Old French types c'est celui-ci qui and qui est-ce qui, which in turn derive from the Latin hic est qui and quis est qui (Löfstedt Reference Löfstedt1966: 258, 267). Despite its relatively rare and sporadic use in Latin texts (Löfstedt Reference Löfstedt1966: 259), possible ambiguities between cleft and ‘determinative pronoun’ readings of some examples (Reference Löfstedt1966: 261f.), and the rare occurrence of nouns and personal pronouns in the focus position of Latin clefts (1966: 271), Löfstedt concludes that the Romance constructions are based on the Latin models and gradually developed into grammaticalised usages (1966: 276). He notes, though, that this holds more for French than for the other Romance languages, and explains it through the more rigid word order of French as compared with its sister languages (1966: 276 fn. 44).
Interestingly, Löfstedt goes on to suggest adstratal influence between French and Germanic languages (especially German), and argues that clefting belongs to ‘European syntactic characteristics’, just like the definite and indefinite articles, the pronoun ‘man’ and the periphrastic perfect (das umgeschriebene Perfekt in his terminology). He considers the contact influences to have been adstratal in nature as it is difficult to prove the direction of influence (Reference Löfstedt1966: 276–7). It seems to me, however, that in positing an adstratal relationship between French and Germanic with regard to clefting, Löfstedt overlooks the fact that clefting can hardly be said to be a grammaticalised feature of German – nor a ‘European syntactic characteristic’ in the light of recent work on Standard Average European. What also undermines his Latin hypothesis is his own observation that no similar construction exists in Romanian, a Romance language similarly derived from Latin, which could be expected to have developed clefting as the other Romance languages have (Reference Löfstedt1966: 256, fn. 5). Another factor weakening the Latin hypothesis is the relatively marginal status of clefting in Italian and, to some extent, also in Spanish. As regards Latin influence on clefting in English, one could add to these Ball's observation that Ælfric and other OE authors sometimes seem to avoid clefting in their translations of Latin texts (see Ball Reference Ball1991: 52 for examples). As yet another piece of evidence, my own searches through YCOE, reported above, revealed that a sizeable proportion of the clefts occurred in texts that are not based on Latin originals. This casts further doubt on (though does not necessarily rule out) the likelihood of significant Latin influence on English clefts.
The obvious formal and functional similarities between the English and the French constructions raise the question of possible contact influence from French especially in the centuries following the Norman Conquest. As we have seen, Löfstedt assumes a certain amount of mutual influence between French and Germanic, but he does not specifically comment on the rise of English clefting in that connection. Of the other writers on the subject, Ball (Reference Ball1991) notes that, in late ME texts, cleft constructions with pronoun foci (e.g. It is he/she that . . .) are ‘particularly evident in translations from French’ (Ball Reference Ball1991: 231), which could be taken to indicate some influence from French. However, she concludes that, despite similar developments, French influence is hardly likely because of the earlier attestation of the English pronoun-foci clefts. In Ball's words,
[t]he two languages are moving along the same lines, but English appears to have had a head start of at least a century. We can, therefore, dismiss the possibility of French influence.
(Ball Reference Ball1991: 280)Ball's main source for early French is Hatcher (Reference Hatcher1948: 1076), who presents a chronology for the French clefts with pronoun foci. According to Hatcher (Reference Hatcher1948), this type of cleft in French is, indeed, attested clearly later than in English: the first instance of c'est lui qui . . . ‘it is he who..’ dates back to no earlier than 1350, while c'est moi qui . . . ‘it is I who . . .’ follows even later, viz. in a text from 1371–2 (1948: 1074–5).
Ball also discusses the possible influence of French on English Adv/PP-focus it-clefts, which according to her do not emerge until the late fourteenth century in more literary English works, i.e. in a period which witnessed the greatest influx of French loanwords (see, however, my discussion of OE Adv/PP-focus clefts above). Here, too, Ball compares the chronology of the English and French clefts on the basis of Hatcher's (Reference Hatcher1948) study on the latter, and concludes that since the French PP-focus ce-cleft is also late and infrequent, it could hardly have provided the model for the English clefts. However, she leaves the question of French influence open ‘pending more precise information about the history of the French cleft’ (Ball Reference Ball1991: 461–2). Such information is to be found in Wehr's (Reference Wehr, Skaffari, Peikola, Carroll, Hiltunen and Wårvik2005) comparative study of focusing strategies in Old French and Old Irish. Wehr's findings confirm the overall picture given in Hatcher (Reference Hatcher1948): the cleft construction is relatively rare in Old French, which for purposes of focus marking could use what Wehr calls ‘free word accent’, i.e. prosodic means only or by placing focused constituents into sentence-initial position. She finds very few or no instances of clefting in Old French texts such as Chanson de Roland, Life of St Leger, or Life of St Alexis (Wehr Reference Wehr, Skaffari, Peikola, Carroll, Hiltunen and Wårvik2005: 365). According to Wehr, it was not until Middle French that the cleft construction became more frequent and eventually grammaticalised, first in combination with interrogative pronouns and gradually generalising to other syntactic contexts as part of the more general process of loss of autonomy of the individual word (Wehr Reference Wehr, Skaffari, Peikola, Carroll, Hiltunen and Wårvik2005: 366 ff.).
German (Reference German and Tristram2003), who relies on Ball's (Reference Ball1991) findings on the relatively late attestation of clefts in ME, is another writer suggesting ‘a possible French adstratal or even a Celtic substratal origin’ for the English cleft construction. Of these two potential sources, French influence on ME was already suggested in an early study by Leon Kellner (Reference Kellner1892), as noted by Ball (Reference Ball1991: 14), who does not consider it likely because of the chronological precedence of at least some types of English clefts (see the discussion above). German, however, draws attention to the possible significance of the fact that the English cleft construction bears a closer resemblance to the corresponding Brythonic construction than the French one: c'est normally remains in the present tense (in standard French); it also shows singular or plural concord. In English, by contrast, the verb is either is or was (cf. Middle W. ys/oed), and it remains in the singular.Footnote 17 There is also a difference in the behaviour of pronouns following the copula: it is in the nominative in ME and in Brythonic, but in the accusative in French. As yet another factor which indirectly supports Celtic influence on English clefting, German discusses the case of Breton French. According to his observations, this variety makes more extensive use of c'est-clefting than standard French, and even allows structures like c'est aller qu'il fait (lit. It is go he does) or c'est à travailler avec son père qu'il est, which echo the Celtic patterns, and are also closely reminiscent of the types of structures found in some ‘Celtic Englishes’. German's hypothesis is that these structures are the result of an unconscious semantic reconstruction and structural blending of the French and Breton patterns in the minds of bilingual speakers (Reference German and Tristram2003: 401–2). Finally, what also enhances the value of Breton French as a useful point of comparison is the fact that this variety displays another syntactic feature familiar from the Celtic Englishes, viz. periphrastic do, realised by the verb faire ‘do’. This is commented on by van der Auwera & Genee (Reference van der Auwera and Genee2002), who discuss the possible Celtic background of periphrastic do in English and various other European languages. They cite Trépos's (Reference Trépos1980) study of the periphrastic use of faire in Breton French (or ‘Franco-Breton’ in their terminology), which concludes that the use of faire in this function is a Bretonism, with no parallel in French (van der Auwera & Genee Reference van der Auwera and Genee2002: 299).
Breton French may, indeed, in this respect be compared with vernacular Hiberno-English and Hebridean English. In both of these varieties clefting is a frequent means of expressing prominence and it also displays qualitative features that can best be explained in terms of substrate influence from Irish and Scottish Gaelic (see, e.g., Odlin Reference Odlin and Kallen1997; Filppula Reference Filppula1986, Reference Filppula1999). Welsh English is yet another variety of English which has been shown to have been influenced by the focusing devices of Welsh, although in the Welsh setting this influence assumes more often the form of simple ‘predicate fronting’ or ‘focus fronting’ than clefting, most probably because of the lack of the copula in the Modern Welsh cleft construction (see, e.g., Williams Reference Williams and Tristram2000; Paulasto Reference Paulasto2006). Taken together, the evidence from Breton French and the ‘Celtic Englishes’ provides strong indirect support for similar Celtic influence on the earlier forms of English.
6 Conclusion
In this article I have sought, first of all, to trace the emergence of it-clefting in English on the basis of two corpora, i.e. The York–Toronto–Helsinki Corpus of Old English Prose and The Penn–Helsinki Parsed Corpus of Middle English (second edition). The evidence obtained from them shows that, though rather rare and sometimes open to different interpretations, clefting had a greater presence in the grammar of OE than has been assumed in many previous works. In ME, it continued to diversify both syntactically and functionally, although the relatively low frequencies indicate that it still remained a rather marginal option as compared with the wide range of available word-order arrangements.
Secondly, the survey of the areal distribution of clefting produced a clear geographical division amongst the European languages, with clefting being most strongly associated with western European languages, including especially the Celtic languages, English, and some of the westernmost Romance languages, most notably French and Portuguese. Crucially, this division marks a difference between English and German, which prefers other means than clefting for the expression of prominence within an utterance or sentence. Typologically, there is some evidence to suggest that clefting is particularly favoured in languages that have a relatively rigid word-order system. Obvious examples are the Celtic languages, French, and, to some extent, English, which in this regard represents a kind of ‘half-way house’, as it uses prosodic means as well for marking prominence. The place of English in this divisioning is not surprising in the light of the notion of Standard Average European: clefting is not among the features on which it is based, and as was noted above, English does not belong to the ‘nucleus’ of SAE. This holds even more for the Celtic languages, which are completely outside SAE. It is rather the position of French – one of the core languages of SAE – in the same camp with the Celtic languages with respect to the feature at issue that calls for an explanation. I believe that the answer to this question can be most plausibly found in its early contact history with Continental Celtic.
Thirdly, I have discussed the contact-linguistic implications of the areal and typological facts, with particular reference to possible Celtic influence on the rise of English clefts. A major factor supporting such influence is the earlier attestation of cleft constructions in Celtic (both Brythonic and Goidelic) as compared with the corresponding English constructions. Clefting in the former has a robust presence in even the earliest recorded stages and most probably extends back to Continental Celtic (i.e. Gaulish). This could also explain the prominent status of clefting in French. Another indirect piece of evidence supporting Celtic influence is the frequent use of cleft constructions in present-day (and earlier) Celtic-influenced varieties of English and in Breton French, which underlines the susceptibility of this area of grammar to contact influences.
Though hardly conclusive because of the indirect nature of much of the evidence, the contact scenario described here receives further support from some of the recent research into a number of other syntactic features of English that can be best explained in terms of substratal or adstratal contact influences from Celtic. These include the so-called internal possessor construction, periphrastic do, the progressive or -ing form of verbs, and the self-forms of personal pronouns (for discussion, see Filppula, Klemola & Paulasto Reference Filppula, Klemola and Paulasto2008). Taken together, they provide strong evidence for substratal (and possibly adstratal) influences from Celtic on the rise of it-clefting in English.
Appendix
Structure of clefts in YCOE and PPCME2 (see www-users.york.ac.uk/~lang22/YCOE/doc/annotation/YcoeRef.htm#clefts, and www.ling.upenn.edu/hist-corpora/annotation/index.htm, respectively; accessed on 21 March 2009)
(11) [. . .] hit wære Swyðun se ðe hine lærde mid þære halgan lare and þone ðe he geseah on ðære cyrcan swa fægerne (ID coaelive, ÆLS_[Swithun]: 388.4472)
‘. . . it was Swithun who had provided him with the holy teaching and whom he had seen so beautiful in the church.’
(12) . . . for it was I myself that came in the likeness (ID CMMALORY,5.131))
‘. . . for it was I myself that came in the likeness.’
