Interpreting the data
In 4, I showed that our theoretical assumptions about present-day language structure influence the way that Ball and I organize and categorize the early historical it-cleft data. In this section, I examine how our theoretical perspectives on language change affect how we interpret the subsequent development of the specificational it-cleft. I explain how Ball (1994) and Patten (2010) choose to analyse what are essentially the same diachronic facts in ways that are consistent with divergent theoretical models. I conclude that while we both provide interpretations of the data which are to some extent manufactured by our theoretical standpoints, the diachronic evidence is more amenable to an explanation which invokes a usage-based, constructional model of change.
The diachronic development of the English it-cleft
The history of the English it-cleft involves two main developments which affect both the structure and function of the configuration. First, the it-cleft occurs with a greater variety of elements in postcopular position, changing from allowing only NPs into this focal position, shown in (21), to occurring with non-NP foci such as prepositional phrases and adverb phrases, as in (22).
(21) It's Howard that plays the bassoon [NP-focus it-cleft]
(22) It's at 2:30 that he starts playing [PP-focus it-cleft]
Second, the it-cleft's information structure changes so that the presupposed cleft clause, which was at one time associated with discourse-old information, is able to express brand-new information. For example, in (23) the previous discourse tells us that there is a bassoonist among them and so the open proposition of the cleft clause (somebody plays the bassoon) is given information. However, example (24) occurs in discourse-initial position. As a result, the proposition in the cleft clause (that someone said 'there are no facts, only interpretations') is not discourse-old and may in fact be entirely new (or unknown) information for the intended audience. Such examples are labelled informative-presupposition (IP) it-clefts by Prince (1978).
(23) A: Is he the bassoonist?
B: No, it's Howard that plays the bassoon
(24) (Start of lecture)
It was Friedrich Nietzsche who said 'there are no facts, only interpretations'
Ball (1994) and the mergers of the English it-cleft
In Ball's (1994) data, non-NP-focus it-clefts and IP it-clefts are not found until the Late Middle English period. She suggests that these new types of it-cleft result from a series of mergers between various configurations, including the existing NP-focus it-cleft, an impersonal/presentational construction and the reverse pseudocleft. Ball claims that the LME AdvP/PP-focus it-clefts evolved from an Old English impersonal construction with BEON/WESAN. These presentational sentences are often found with scene-setting adjuncts, such as not long afterward in (25). While such examples show a superficial similarity to PDE it-clefts with prepositional phrase and adverb phrase foci, the verb in the impersonal construction is not the copular BE of cleft sentences and is instead a lexical verb meaning 'happen' or 'come to pass' (Ball 1994: 611). So, for example, the meaning of (25) is something like, 'Not long after, it happened that he fell into a grave sickness'.
(25) Was hyt nat long aftyrward, He fyll yn a sykenes hard. 'Not long after that, he fell into a grave sickness'
(c1400 Mannyng, Handlyng Synne [Ball 1994: 612])
According to Ball (1994), in LME these impersonal examples underwent a partial merger with the existing NP-focus it-cleft, resulting in a new type of cleft construction: the AdvP/PP-focus it-cleft. This merger also explains the onset of the development of it-clefts with new information in the cleft clause. In the impersonal examples, the sentential complement (he fell into a grave sickness) comprises the main informational content of the construction. It therefore follows that this property is carried into the new AdvP/PP-focus it-cleft. However, in order to explain the development of IP it-clefts with nominal foci, Ball must propose another, subsequent merger between the new cleft construction and the existing NP-focus it-cleft. Since at this point the frequency of the new AdvP/PP-focus IP it-cleft is too small to have had a significant influence on the existing cleft construction, Ball suggests that other sentence-types must have contributed to this development, such as the impersonals again and/or the reverse pseudocleft.
Ball's rationale for proposing a merger between the NP-focus it-cleft and the it-impersonal examples is to some extent dependent upon her adherence to an expletive analysis of PDE it-clefts. As I explained in 3.1, on this account, the cleft clause shows a superficial similarity to restrictive relative clauses, but it does not modify its immediate antecedent. As a result, authors differ as to whether they analyse the cleft clause as having the same internal structure as a restrictive relative (Chomsky 1977; Williams 1980), or as a structurally unique type of that-clause, which appears with relative pronouns only as a result of analogy (Delahunty 1982; Rochemont 1986; Heggie 1988). Ball (1994) takes an intermediate position. She claims that, where the postcopular constituent is a noun phrase, the cleft clause is (in terms of its internal structure) a restrictive relative. However, for ¿t-clefts with non-NP foci, "there is little support for a restrictive relative analysis of the subordinate clause" (Ball 1994: 605); in these examples, the cleft clause often expresses a complete sentence (without a perceptible gap) and is very rarely introduced by elements other than that (see (22) above). Ball concludes that non-NP-focus ¿t-clefts contain a "sentential complement" rather than a relative clause, and so make up a separate kind of cleft configuration.
By adopting a synchronic perspective which makes classification of the cleft clause difficult, and which emphasizes an apparent difference in the form of NP-focus and non-NP-focus ¿t-clefts, Ball is forced to find an additional source construction for the LME AdvP/PP-focus ¿t-cleft beyond the NP-focus ¿t-cleft. The impersonal configuration is a reasonable choice; as well as sharing a superficial similarity to AdvP/PP-focus ¿t-clefts, these ¿t-impersonals function to present a complete sentence (with an optional adjunct). This explains why ¿t-clefts with non-NP foci possess sentential complements rather than relative clauses. Finally, for Ball, both the ¿t-impersonals and the ¿t-clefts contain an expletive (h)¿t.
Ball's (1994) account of the historical development of the English ¿t-cleft configuration can be understood in relation to the late twentieth century generative model of language change, outlined in 2.2. As we have seen, Ball proposes a merger between two structures which share a superficial similarity (including an expletive ¿t). She notes that "With a decline in the happen sense of be", there is nothing to prevent an ¿t-cleft reading of the impersonal, whereby the sentential complement is reanalyzed as containing an imperceptible gap (Ball 1991: 458). This is followed by a subsequent merger (between the new AdvP/PP-focus IP ¿t-cleft and the NP-focus ¿t-cleft), which is presented as a discrete change. For Ball, the AdvP/PP-focus IP ¿t-cleft and the NP-focus IP ¿t-cleft are the fully-formed products of these distinct mergers, which begin and end in the LME period.
In some ways then, these two changes are presented as abrupt and non-continuous. Ball focuses on the actuation of each change (reanalyzes brought about by ambiguity or opacity), rather than on their spread. For instance, the initial merger applies wholesale and uniquely to the syntactic categories PP and AdvP; it does not predict or explain the ¿t-cleft's subsequent development, as it goes on to permit a yet greater range of non-nominal foci. In order to speculate about the changes to the it-cleft beyond LME, Ball (1994: 614) has to invoke an increasing number of outside pressures; she notes, "From the LME period onwards, the it-cleft construction has taken in a greater variety of non-NP foci, possibly in response to the decline of some alternatives and functional change in others (e.g. preposing)". Once again, this fits more with a formal, generative account of language structure: the concept of structural gradualness, (that is, "an orderly progression across semantic/pragmatic, lexical and syntactic structures") is less consistent with this model (Traugott and Trousdale 2010b: 24).