4 DESARROLLO DE INGENIERÍA
4.1 DISEÑO CONCEPTUAL
4.1.8 Estimación de los parámetros críticos de performance
Julia Schlüter’s chapter in Rohdenburg & Schlüter (2009) is notable for focusing on one particular conjunction, (Up)on (the) condition (that),57 which had previously not received individual attention in subjunctive-related studies. As a diachronic study, it also covers a much longer period in the history of English than most other studies, using corpora of material starting from the fifteenth century (for BrE) and the eighteenth century (for AmE), right up to the present day.
To produce worthwhile results, Schlüter draws on far larger corpora than in previous studies.
While most of her early corpora are from the Chadwyck-Healey prose collections, those for PDE are the BNC for BrE and the ANC for AmE, with the gaps filled by corpora put together in a University of Paderborn research project. Corpus size ranged from 4.9 million words up to 22.6 million for the ANC
57 Hereafter, for simplicity, just on condition, though all its variations were considered in the study.
and 100 million for the BNC. A large database of British and American newspapers from the 1990s was also used for the in-depth synchronic element of the study (2009: 283–285). Criteria for the identification of subjunctive forms were iNO-S, iBE, iST and iNEG, and four variants were considered: subjunctives, modal periphrases, non-distinct forms and indicatives.
In the diachronic study, it is clear that in the earliest BrE corpus (1460–1670) on condition is predominantly associated not with subjunctive forms but with modal periphrases, which feature in 82 per cent of complement clauses. This proportion falls steadily over time to a figure of 18 per cent in present-day BrE (2009: 287–288). There is a corresponding increase in indicatives, which rise from 8 per cent in the earliest corpus to 61 per cent in the most recent corpus. Subjunctives remain a very low-frequency option until a small but significant increase to 9 per cent in the twentieth century.
In the earliest AmE corpus (1728–1799) the situation is close to that in the BrE corpus for the same period, with modals at 60 per cent in AmE and 67 per cent in BrE, but thereafter the patterns for the two varieties diverge. Most strikingly, the reduction in modal periphrases in BrE in the nineteenth century does not take place in AmE. Then, in the twentieth century, the proportion of modal periphrases in AmE falls dramatically while that of subjunctives increases, with indicatives remaining a low-frequency option. In the most recent AmE corpus (1960-2003), subjunctives stand at 74 per cent, modals at 13 per cent and indicatives at only 9 per cent (2009: 289). Further data regarding the situation in AmE in the twentieth century was provided by samples from the Los Angeles Times and New York Times published between 1900 and 1960. These showed that use of the subjunctive in this environment in AmE was already established at the beginning of the century. What’s more, Schlüter’s assessment is that, when compared with the data from Övergaard (1995), it is ‘exactly as far advanced’ (2009: 290) as in mandative clauses at this date.
So what brought about this twentieth-century AmE adoption of the subjunctive in clauses following on condition, an environment in which it had apparently never been the most common choice (unlike mandative clauses)? And why did it not happen in BrE? Schlüter proposes two factors that may have played a part: a ‘certain disposition’ and a ‘triggering circumstance’ (2009: 291). The first concerns the point highlighted above, that in the nineteenth century AmE retained a preference for modals, whereas BrE increasingly favoured indicatives. The importance of this, Schlüter argues, is that:
both the subjunctive and modal periphrases ensure an explicit marking of the irrealis, while the indicative is indifferent to the realis/irrealis distinction. This means that forms marked for irrealis after on condition became ever more rare in the British homeland, while AmE preserved a grammatical marking of the irrealis. (Schlüter 2009: 291)
In other words, in AmE, changing from modals to subjunctives was acceptable because both were marked for irrealis. But what was the triggering circumstance that prompted the change in the twentieth century?
Schlüter suggests it was the increase in mandative subjunctives, and she draws attention to the semantic similarities between clauses following on condition and mandative clauses: ‘Its affinity with mandative interpretations accounts for the fact that it readily accommodates the subjunctive, and the choice of verbal syntagms consequently develops in parallel with the evolution of mandative contexts’ (2009: 292–293).
Both explanations are of interest but they raise further questions. Why did AmE retain modals in the nineteenth century and resist a change to indicatives? Why did modals then become less acceptable in the twentieth century in AmE? Was the pattern with mandatives actually similar in the nineteenth century, and are the supposed semantic similarities as strong as suggested? Crucially, however, the study shows that the different attitude of each variety towards indicatives is possibly of great importance, and that inter-varietal studies that do not consider the indicative option are possibly missing a vital piece of the overall picture.
The object of the synchronic investigation was to look more closely at the differences between BrE and AmE in clauses after on condition in the more recent history of English. The corpora contained texts from the 1990s from four British and three American newspapers (2009: 285). In addition to basic quantitative differences, differences connected to four areas were investigated: semi-formulaic uses, the special status of the verb be, the influence of negation and the choice of modal auxiliary.
One semi-formulaic use that came to light demonstrates the effect that different conventions can have in the different varieties. Overall in the synchronic corpora, the conjunction on condition (and all its variations) was found to be more common in BrE than in AmE: 2.99 occurrences per million words (pmw) as opposed to 2.54 pmw (2009: 293). But it was also discovered that more than half of the 501 examples from the AmE corpus were instances of what is clearly a semi-formulaic use, as in (150), her (12):
(150) One banker, speaking on condition he not be named, declined detailed comment on the plan, except to say, ‘They’ve got a way to go.’
(Los Angeles Times 1992)
In contrast, this pattern, which Schlüter schematises as (up)on (the) condition (that) NP not be Ved, was found only three times in the BrE corpus. Postulating that BrE might use the phrase (up)on (the) condition of anonymity instead, Schlüter was surprised to find that not only was the frequency of the proposed
alternative very low, but it was significantly more common in AmE. Further searches found several instances of both phrases in the ANC (22.6 million words), all in press sources, but only three instances of (up)on (the) condition of anonymity in the much bigger BNC (100 million words), all in an American context. Clearly, different conventions are at work in the press in AmE and BrE: ‘(up)on (the) condition (that) NP not be Ved and (up)on (the) condition of anonymity are part of American journalese but hardly extend to other text genres, and are not current across the Atlantic either’ (2009: 295). If this semi-formulaic use is omitted, the difference in frequency between varieties is even greater – BrE 2.98 pmw, AmE 1.25 pmw – leading Schlüter to conclude that ‘on the one hand, on condition is strongly associated with a stereotyped high-frequency type of clause in AmE journalistic styles, but is less often used elsewhere. On the other, BrE employs on condition more frequently overall and in a wider variety of contexts’ (2009: 295).
In the second part of the synchronic investigation, looking into the association of the subjunctive with the verb be (see Section 4.2.6), Schlüter aimed to establish whether the situation in clauses after on condition differed from that in mandative clauses. The most striking result was that while BrE shows a slightly higher proportion of subjunctives in clauses featuring be (13 per cent) than in clauses featuring other verbs (9 per cent), AmE shows a considerably lower proportion of subjunctives in clauses featuring be (58 per cent) than in clauses featuring other verbs (70 per cent). This Schlüter takes to indicate that ‘AmE clearly avoids the subjunctive in connection with be, falling back on modal periphrases instead’ (2009: 297). As she points out, this doesn’t tally with the suggestion that the distinctive nature of the be subjunctive makes it easier to deal with. Her explanation concerns the levels of association of the subjunctive with formality in the different varieties:
while subjunctives in general are a widespread feature in AmE, the particular form be may nevertheless be perceived as more formal than other, less distinctive subjunctives. This appears to be the crucial effect responsible for the avoidance of subjunctive be. (Schlüter 2009: 298)
In other words, the distinctiveness of subjunctive be seems to help its entrenchment in BrE but to mark it out in AmE, which is a very intriguing suggestion.
With regard to preverbal negation, the third element of the synchronic investigation, previous studies (e.g. Övergaard 1995: 70–74) have shown that in mandative clauses it remains a low-frequency option, particularly in BrE, where negated modals are preferred. One of the findings of Schlüter’s synchronic analysis of clauses following on condition in her press corpora indicates that in this environment both varieties also ‘show a remarkable avoidance of subjunctive forms in negated
subordinate clauses’ (2009: 299). This became particularly clear when the content clauses after on condition were split into negated and negated groups. In the BrE corpus, 12 per cent of the non-negated clauses featured subjunctives, while there were no non-negated subjunctives at all. In the AmE corpus, the share of subjunctives fell from 68 per cent in the non-negated group to 47 per cent in the negated. In both varieties, the number of modal periphrases increased accordingly (2009: 300).
To explain this, Schlüter refers to Horn’s ‘Embedded Negation Constraint’ (Horn 1978: 191–
205), which describes a cross-linguistic aversion to negation in non-finite embedded clauses, and also proposes a continuum of finiteness, in which subjunctives range closer to the non-finite end than indicatives. Horn offers the following motivation for the constraint:
The function of negation is to deny a proposition or claim, or to substitute an inverse act for the one under consideration. The less the dependent clause looks and acts like a sentence – the less it seems to express a complete proposition, thought, claim, or act – the less negation is admitted without corresponding discomfort, if it is admitted at all.’ (Horn 1978: 205)
If this in-built constraint exists, it suggests that negation is felt to be more acceptable with modal periphrases and indicatives because they are more clause-like, but is less acceptable with subjunctive clauses because they ‘are semantically more dependent and thus less fully-fledged sentential units than other finite clauses’ (Schlüter 2009: 301).
The fourth element of Schlüter’s synchronic study involves the choice of modal verb within those clauses featuring modal periphrases. In mandative clauses, previous studies (e.g. Övergaard 1995:
56) have shown that in the twentieth century BrE has a strong preference for should, while AmE is more varied in its choice of modal. Schlüter’s results indicate that the picture is very different in the case of on condition, with BrE exhibiting the greater variation. In the modal periphrases in the BrE press corpus, only 16 per cent feature should, while 38 per cent feature would, with could and will both frequent choices, and can, must and will less common options. In the AmE corpus, on the other hand, would is found in 83 per cent of examples (2009: 303). Schlüter sees this difference in terms of the bigger picture of differing modal use in the two varieties described, for example, by Leech (2003: 236), and also the volitional force of the different modals:
We may conclude that BrE manifests a considerably greater explicitness than AmE as far as the differentiation between different degrees of volitionality is concerned. Judging from the literature, this situation is contrary to the one obtaining for mandative subjunctives, where it is AmE that is less fixed in its use of modals. (Schlüter 2009: 304)
But what should also be made clear is that it is a reminder that, while it has semantic similarities, on condition is not strictly a mandative trigger, and so the same patterns of complementation should not be assumed.