4 DESARROLLO DE INGENIERÍA
4.1 DISEÑO CONCEPTUAL
4.1.10. Optimización
total n/10K total n/10K
Subjunctive 0 0.00 2 0.05
Should 5 1.34 11 0.26
Other modals 2 0.54 9 0.21
Indicative 1 0.27 45 1.07
Non-distinct forms 0 0.00 10 0.24
Total mandative 8 2.15 77 1.83
*Based on Table 2 in Klein (2009: 34).
In Klein’s subsample from the early LLC subcorpus, she finds no mandative clauses containing subjunctives, five containing should and one containing an indicative form. In Waller (2005: 49), the findings for LLC were five subjunctives, 20 should-clauses and seven indicatives. The difference is presumably at least partly due to the fact that Klein’s study was restricted to texts from 1958 to 1960, whereas Waller (2005) used all the data in the subcorpus. In the later ICE-GB subcorpus, Klein finds the same two subjunctive examples identified in Waller (2005). Curiously, however, there is evidence of some confusion when she discusses identification criteria for (151), her (2).
(151) When we got the request for help we suggested that we look not at general loans that wouldn’t help a country already in difficulty but to see if we could give specific help in certain spheres such as for example with food processing with transport with oil exploration.
<ICE-GB:S1B-053 #102:1:B>
She chooses to identify look in (151) as a subjunctive form on the basis of iNEG (2009: 34), but then expresses concern about that identification because not is not in the characteristic preverbal position. She is right to be concerned. This is not a case of iNEG, and the reason is a question of scope: not here has scope over the prepositional phrase that follows rather than the verb that precedes it. However, look is still a subjunctive form. As it follows the past-tense suggested, it can be positively identified because of the lack of backshifting, i.e. by iST.
Klein’s normalised figures for should variants indicate a significant decrease between the two subcorpora, though the danger of relying too much on such small numbers should be borne in mind.
The most remarkable finding is the large increase in the number of indicatives. She finds one example in her (reduced) LLC subcorpus and 45 in the ICE-GB subcorpus, whereas the comparable figures from
Waller (2005: 49) are seven in the (full) LLC subcorpus and just eight in the ICE-GB subcorpus. The likely explanation is that Klein has not disregarded non-mandative uses of triggers such as insist and suggest. As content clauses following non-mandative uses often contain indicatives, this is bound to skew the figures. These findings also affect her treatment of non-distinct forms: ‘It is generally agreed that ambiguous forms should be added to the numerically more important group, which in this case obviously are the indicatives’ (2009: 35). It is far from established that such an approach – though taken, for example, by Övergaard – is ‘generally agreed’, but basing it on flawed figures is doubly misleading. By combining the non-distinct and indicative figures, she comes up with an increase in the frequency of indicatives of 385.89 per cent, a figure that is not supported by similar increases in any other studies.
Despite this problem with the indicative figures, Klein’s results do support the findings in Waller (2005) that there is no evidence in DCPSE of a significant increase in the use of the mandative
subjunctive in spoken BrE between 1960 and 1990. That study also found a decrease in mandative contexts over that period, and Klein refers indirectly to the low number of such contexts when offering an explanation of why the apparent AmE influence on BrE use of the mandative subjunctive in written English is not also found in spoken BrE: ‘Probably, as it is still, even in American writing, a formal way of expressing commands or propositions, which is not typical of spontaneous speech’ (2009: 37).
The method for finding formulaic subjunctives within the corpora is not clearly set out, but it apparently includes searching for be it,58 far be it, so be it and if need be (2009: 35). Klein doesn’t expect formulaics to be common in speech, and finds just six examples in the earlier subcorpus and four in the later: in both cases, more than the number of mandative subjunctives found. Present subjunctives in conditional and concessive clauses are not common in written English, so it is not surprising that she finds no examples in the reduced LLC subcorpus, but she does find five examples in the later subcorpus (2009: 36). However, this demonstrates once again the difficulties involved in drawing conclusions about such low-frequency items from relatively small corpora.
Methods for identifying past subjunctives were also not made clear, but her findings revealed a ‘modest decline’ (2009: 36) between the two periods, though the overall figures were too low for it to be statistically significant. There are, however, two points that are of interest in relation to other studies regarding use of the past subjunctive. First, Klein’s results for the 1990–92 subcorpus reveal that indicative was was preferred to subjunctive were in relevant contexts in almost 80 per cent
58 Note that be it probably indicates what Huddleston & Pullum call the exhaustive conditional construction (2002:
1001), considered to be a productive construction and so arguably not ‘formulaic’.
of cases. The comparable figure for F-LOB, a corpus of written BrE from the same period, was found by Leech et al. (2009: 65) to be 48 per cent, showing a significant difference between preferences in spoken and written BrE.59 Second, in the 1990–92 subcorpus, a remarkably large proportion60 of the instances of the past subjunctive were in the set phrase as it were, a total of 35 instances (2009: 36).
Johansson & Norheim (1988: 34) had noted that as it were was more common in BrE than in AmE, with four instances in Brown and 17 in LOB. Klein points out that in the ICE-GB subcorpus, most of the instances of as it were were found in text category B (informal face-to-face conversation) and text category I (assorted spontaneous), where the expression ‘serves as a type of discourse marker’ (2009: 36).
5.2.16 Leech et al. (2009): ‘The subjunctive mood’
Recent change in BrE and AmE in the use of the subjunctive in a number of environments is investigated in a chapter of Change in Contemporary English, a book by Geoffrey Leech, Marianne Hundt, Christian Mair and Nicholas Smith that for the most part draws its data from the Brown family corpora but also uses other corpora when appropriate. In addition to thorough and clearly thought-out studies of both the mandative subjunctive and the most common use of the past subjunctive, the chapter includes a side investigation into reports of the increasing use of would in the protases of conditional sentences. The statistical results are also reviewed in the light of recent thinking about trends in language change, such as Americanisation and colloquialisation.
In the introduction, the authors neatly set out the two areas of modality with which the most common uses of the subjunctive in PDE are associated: ‘Just like some modal auxiliaries, the subjunctive in English can be used to express obligation or necessity (he demands that the evidence be / must
be / should be demolished). In if-clauses it can express “irrealis”, similar to the use of such modals as could and might’ (2009: 51). After citing reports in the literature about the demise of the subjunctive, they also make a point of distinguishing between the ‘paradigmatic poverty of the subjunctive on the one hand and its use on the other hand’ (2009: 51). The uses they recognise for the present subjunctive are (1) those that appear ‘in a few fossilized contexts’, such as if need be, be it that. . ., God save the
59 Care should be taken not to ascribe too much significance to these figures because of differences in identification techniques in the three studies.
60 The wording in the paper makes it difficult to be certain about the exact proportion: ‘as it were … is used almost twice as much as the genuine past subjunctive’ (2009: 36). If ‘the genuine past subjunctive’ is intended to mean all other – non-formulaic – uses of the past subjunctive, then the proportion with as it were is almost two-thirds.
Queen and after lest; (2) mandative subjunctives: ‘a development in which (written and spoken) AmE is leading world English in an essentially twentieth-century (post-colonial) revival’; (3) other low-frequency uses such as in conditionals: ‘The present subjunctive in if-clauses seems to be a fossilized feature of English, with only a sprinkling of examples in standard corpora such as Brown (5), LOB (9), Frown (4) and F-LOB (7)’. For the past subjunctive, their intention is to ‘answer the question whether the past subjunctive is on the wane’ (2009: 52).
5.2.16.1 The mandative subjunctive
The first part of the study of mandative subjunctives involves using the set of triggers in Johannsson &
Norheim’s (1988) analysis of LOB and Brown for their own investigation of F-LOB and Frown, then comparing their results with those from the earlier paper.61 This essentially replicates the study in Hundt (1998b) with a couple of differences. First, at the time of Hundt’s earlier paper, the Frown corpus was not quite complete; second, this time the adjective triggers from Johansson & Norheim’s list are included, not just the verb and noun triggers. Leech et al. state that the triggers are ‘the seventeen most common suasive verbs and related nouns and adjectives’ (2009: 53), though the justification for claiming them to be the most common, rather than just the ones chosen by Johansson & Norheim, is not supplied.62
For ease of comparison, they initially restrict the variants to content clauses containing subjunctives and should, leaving the discussion of indicatives to the second part of the study.
Identification of subjunctives is based on iNO-S, iBE, iNEG and iST, while non-distinct forms are not included in their figures. It has to be said that the explanation of iST is, on the face of it, rather
misleading: ‘With a past tense verb in the matrix clause, however, the unmarked form following a plural pronoun was interpreted as a subjunctive (e.g. He insisted that they go)’ (2009: 54). There is, of course, no reason to restrict iST to clauses featuring subjects realised by pronouns rather than other noun phrases, or to those containing plural pronouns rather than singular, and there is no evidence that this approach was taken in the study. The explanation seems to be that the statement is intended to refer to a specific example earlier in the paragraph – It is important that they leave on time – but it is easily read as a general statement, and is therefore potentially confusing.
61 As mentioned above (Section 5.2.8) in relation to Hundt (1998b), one problem with relying on the Johansson &
Norheim (1988) figures is the doubt about whether iST was used to identify subjunctives in all persons.
62 For example, as mentioned above (Section 4.3.2), Huddleston & Pullum (2002: 999) claim that mandative clauses in PDE ‘are hardly possible’ with the verb wish’, which is one of the Johansson & Norheim triggers.
The results, summarised in Table 5.14, confirm the general findings reported in Hundt (1998b), with AmE maintaining a strong preference for the mandative subjunctive and BrE increasing its use of the subjunctive in the thirty-year period, but still lagging some way behind AmE.
Table 5.14. Comparison of results from Leech et al. (2009) and Övergaard (1995):
absolute and relative frequencies of subjunctive and should variants in mandative clauses after verbal, nominal and adjectival triggers in BrE and AmE from the 1960s and 1990s.*
Leech et al.
* Leech et al. took LOB and Brown figures from Table 1 in Johansson & Norheim (1988: 29);
Leech et al. figures from Table A3.1 (2009: 281); Övergaard AmE figures from Tables 23 and 26 (1995: 56, 69), BrE figures from Tables 24 and 26 (1995: 56, 68).63
The authors point out that in their study the changes in BrE are not as extreme as in the findings in Övergaard (1995). To support this, they claim that in her 1990 BrE corpus Övergaard found 14 should variants and 44 subjunctives (2009: 54), but unfortunately these figures are misleading. First, they are taken from Table 1 in Övergaard (1995: 15–16), which is only concerned with mandative clauses after verbs, rather than all types of trigger. Second, as discussed in Section 5.2.5, Övergaard counts NDs as subjunctives in most of her figures, including those in Table 1. The more accurate figures for her 1990s BrE corpus, after all types of trigger and with NDs removed, are 35 should variants and 47 subjunctives, as shown in Table 5.14. Even with the correct figures, however, the reduction in the use of should in F-LOB compared with LOB in Leech et al.’s study is not as great as that found for the same period by Övergaard, nor is the corresponding increase in subjunctives. Both of these BrE changes in Leech et al.
(2009) are still statistically significant, however, but notably the should variant remains the preferred option in F-LOB. For AmE, the small differences between the results for the Brown and Frown corpora are not statistically significant. For Leech et al., the lack of significant change in AmE is to be expected:
63 Figures are not strictly comparable because of differences in approaches to the identification of subjunctive forms, as discussed in Section 5.2.5. See also Table A22 in the Appendix.
This hardly comes as a surprise if we consider that even back in the 1960s with almost 90 per cent of subjunctives in mandative contexts, a saturation point had practically been reached in