Sorting the data
We have seen then that Ball and I adopt different analyses of the present-day it-cleft, developed from within different theoretical traditions. In this section, I show that this has a crucial effect on the way that we handle the historical data. By applying a theory-specific set of criteria, Ball (1991) explicitly discounts a collection of Old English examples which obtain fundamental importance in Patten (2012). This influences how Ball and I ascertain and explain the it-cleft's origin. In the section that follows, I go on to show that this, in turn, impacts in interesting ways on how Ball and I interpret the subsequent development of the English it-cleft.
Ball's (1991) it-cleft origin story
For Ball (1991), the it-cleft originated in Early Middle English. She provides the following example of a focus-first it-cleft from the late thirteenth century, which she identifies as one of the earliest it-clefts in her corpus.
(7) 'A-bidez,' quath pis holie man: 'ore louerd is guod and freo. pe deuel it is pat bringuthpis wedur...'
'Stay, said this holy man, our Lord is good and free. The devil it is that brings this weather...' (1280-90 South English Legendary [Ball 1991: 158])
Ball claims that this new sentence type developed out of an Old English NP BEON Rel-Clause configuration, an example of which is given here as (8).
(8) ...min fader is pe me wuldrad .. .my father is that me glorifies 'It is my father that glorifies me'
(EElfric, Catholic Homilies vol. II [Ball 1991: 27])
Ball (1991: 178-179) suggests that, perhaps due to a rise in dummy subjects in EME, this construction "was analyzed (or reanalyzed) as containing an empty unexpressed subject, with the initial NP being underlyingly the predicate complement", shown in (9). This empty subject could then be realized as it, providing us with the it-cleft configuration, as interpreted through the expletive account (see 3.1). On this analysis, the cleft clause is the syntactic sister of the postcopular NP, situated inside the VP.
(9) [S' COMP [S XP INFL [NP e] [VP beon NP S']]] (Ball 1991: 175)
Ball's account of the origin of the specificational it-cleft is successful in many ways. In particular, it provides support for and helps to explain the details of the generative, expletive analysis outlined in 3.1. If the it-cleft truly developed out of these Old English NP BEON Rel-Clause examples, then we have a reason as to why a predication relationship might exist between the postcopular NP and the cleft clause: because this NP was originally the subject, with the (then) postcopular clause functioning as the predicate complement. In addition, it explains why the PDE cleft clause is unique in having the internal structure of a relative clause, without modifying the antecedent noun. In Old English, relative clauses could occur without overt heads, especially when functioning as the complement of BEON, as in (8) (Ball 1991: 27). Finally, this diachronic account also provides an explanation as to why the it-cleft has a specificational meaning, since the Old English NP BEON Rel-Clause examples also had this function. For instance, in (8), both the original OE example and Ball's z'f-cleft translation of it identify the referent my father as the one that glorifies me.
On this basis, the Old English NP BEON Rel-Clause configuration seems to be a good choice as a potential source construction for the English it-cleft. However, when we examine the data, we find some fundamental problems with this diachronic account. First, as Ball (1991: 52) notes, the specificational NP BEON Rel-Clause configuration is very rare in Old English, occurring only in translations of Latin headless relatives. For example, (8) is an English translation of the Latin ...estPater meus, quiglorificat me. Second, while it is true that these OE tokens can be interpreted as PDE it-clefts (as shown by Ball's paraphrase in (8)), they are nevertheless formally akin to reverse pseudoclefts (see Ball 1991: 51). For example, (8) can be translated simply as my father is the one that glorifies me. For both the PDE reverse pseudocleft and the OE token, the focal NP is the grammatical subject; they differ only in that the predicate complement of the latter is a headless relative.
This begs the question of why this rare and restricted configuration, which looks very much like an early version of the reverse pseudocleft, would develop into the (relatively productive) it-cleft construction; without the prior existence of the focus-first it-cleft shown in (7) as an analogical model, it is doubtful whether a rise in dummy subjects would really be enough to motivate the reanalysis of an unambiguous surface subject as an underlying predicate complement. This suggests that Ball's account of the origin of the it-cleft is as much influenced by her subscription to the (then prominent) generative analysis of the PDE configuration as it is based on the diachronic facts. This becomes even more apparent when we observe, and examine Ball's treatment of, a small collection of Old English tokens, including (10).
(10) pa cwadon pa geleafullan, 'Nis hit na Petrus pat par cnucad,
Not-is it-n. not Peter-m. rel-нєЇspan>. there knocks
ac is his angel' but is his angel-m.
'Then the faithful said: It isn't Peter who is knocking there, but his angel.'
(EElfric, Catholic Homilies vol. I [Ball 1991: 39])
Despite the striking resemblance, for Ball, this Old English hit BEON NP Rel-Clause configuration is not an early version of the PDE it-cleft, nor is it instrumental in the development of the it-cleft proper. Instead, Ball explicitly discounts this set of examples from the main data set. She notes that in Old English, "there is no hit-cleft which corresponds to the stereotypical...it-cleft of the Linguistics literature" (Ball 1991: 45). However, as I go on to explain in 4.3, Ball's main reason for excluding these examples is that they are not amenable to, and therefore represent a problem for, the expletive analysis of it-clefts.