If-insubordination in spoken British English: Syntactic and pragmatic properties

Language Sciences 66 (2018) 42–59

Contents lists available at ScienceDirect

Language Sciences journal homepage: www.elsevier.com/locate/langsci

If-insubordination in spoken British English: Syntactic and pragmatic properties Cristina Lastres-López Department of English and German, University of Santiago de Compostela, Avda. Castelao s/n, 15782, Santiago de Compostela, Spain

a r t i c l e i n f o

a b s t r a c t

Article history: Received 31 January 2017 Received in revised form 9 September 2017 Accepted 19 December 2017

This paper analyses insubordinate if-clauses in spoken British English (e.g. If you’ll just come next door) as independent from full conditional clauses and indirect interrogative complement clauses, using data extracted from the British component of the International Corpus of English and the British National Corpus. The study shows that such constructions occur most frequently in conversation and that they express a wide variety of functions in discourse. The polyfunctionality of insubordinate if-clauses is presented together with other, alternative grammatical patterns that may express the same function in discourse. The analysis also reveals a correlation between the discourse function of the clause and the type of verb (modal or lexical) used in it. Ó 2017 Elsevier Ltd. All rights reserved.

Keywords: Insubordination if-clauses Discourse functions Spoken discourse Syntax Pragmatics

1. Introduction This paper analyses insubordinate if-clauses in spoken British English, as exemplified in (1), as multifunctional constructions which can have a variety of uses in interaction: (1)

Uhm well uh if you could just remind me for example what age he was when he left (ICE-GB:S1A-072 #045:1:A)

While conditional subordination and clause complexity have been studied from many different perspectives (Greenbaum and Nelson, 1995, 1996; Bybee and Noonan, 2002; Verstraete, 2011; Kortmann, 2012; Ehmer and Barth-Weingarten, 2016; Traugott, 2017; among others), instances of conditional clauses in which the clause appears in isolation from the main clause have generally been regarded as exceptions to the norm and, as such, have received little attention in grammars and more specific studies (Quirk et al., 1985; Biber et al., 1999; Huddleston and Pullum et al., 2002). The aim of this paper, then, is to analyse the linguistic phenomenon of insubordination itself. In particular, the focus will be on clauses introduced by the conjunction if1 in English, paying special attention to the syntax and the pragmatics of such constructions and to the interconnection between these two areas with regard to insubordination, using a corpus-based approach. More specifically, the aim is to shed light on the diversity of functions these constructions may have in discourse, looking at their frequency in the language and at alternative grammatical patterns that may also be possible to

E-mail address: [email protected]. Following many mainstream grammars, the distinction between if conjunction and if preposition made by Huddleston and Pullum et al. (2002) is not contemplated here. 1

https://doi.org/10.1016/j.langsci.2017.12.009 0388-0001/Ó 2017 Elsevier Ltd. All rights reserved.

C. Lastres-López / Language Sciences 66 (2018) 42–59

43

express the same discourse function. Furthermore, the analysis will consider whether certain functions increase the likely use of specific sets of verbs, and whether these can therefore be considered predictors of the function of the clause. The methodology used will be corpus-based. The data will be extracted from the British component of the International Corpus of English (henceforth ICE-GB) (Nelson et al., 2002) and from the British National Corpus (henceforth BNC). After this brief introduction, this paper is structured as follows. Section 2 offers a preliminary characterization of insubordinated if-clauses. Section 3 reviews the literature on the topic, starting from the general notion of insubordination coined by Evans (2007),2 and exploring more specific studies that have extended and broadened the notion (Mithun, 2008; Sansiñena et al., 2015; Cristofaro, 2016; Evans and Watanabe, 2016a; Heine et al., 2016). The section defines the notion of insubordination and examines the formation of insubordinate clauses diachronically, considering both Evans’ (2007) ellipsis hypothesis and proposals made here by others (Mithun, 2008; Sansiñena et al., 2015; Cristofaro, 2016; Heine et al., 2016). Section 4 delves into prior research that has specifically addressed if-insubordination in English (Stirling, 1999; Mato-Míguez, 2014a, 2014b, 2016; D’Hertefelt, 2015; Kaltenböck, 2016; among others). Section 5.1 discusses methodology, corpora and data extraction, before providing a detailed quantitative and qualitative analysis of the corpus-based results in Section 5.2. In particular, Section 5.2.1 examines the discourse function of these constructions, and Section 5.2.2 discusses their distribution across the corpora, before addressing the correlation between discourse functions and verb types, in Section 5.2.3. Finally, Section 6 offers some concluding remarks based on the corpus-based data on insubordinate if-clauses. 2. Insubordinate if-clauses characterised Clauses introduced by if have generally been classified as instances of subordination, in which the subordinate clause depends on a matrix clause. As such, they are usually conditional clauses, as in (2), or less frequently, complement clauses of the indirect interrogative type, as in (3). (2) (3)

If it’s a really nice day we could walk (ICE-GB:S1A-006 #301:1:B)3 I don’t know if I can let you have that (ICE-GB:S1A-085 #302:1:A)

In both (2) and (3) above, the clause introduced by if, highlighted in bold in the examples, is subordinated to a main clause, shown in italics. In (2) the main clause, also known as apodosis (q), expresses a situation that “is contingent on that in the subordinate clause” (Quirk et al., 1985: 1088), known as protasis (p); whereas in the syntactic pattern in (3) this contingency of situations is not present in the construction and, contrary to what happens in conditional clauses, the verb in the matrix clause licenses the subordinate if-clause, which expresses an indirect interrogative. For some authors, such as Matthiessen and Thompson (1988) (c.f. Halliday and Matthiessen, 2014: 428–588), sentences like (2) and (3) are instances of two distinct types of subordination. While complement clauses, as in (3), are classified as cases of embedding, adverbial clauses, as in (2), are not embedded, but rather are instances of clause combining or hypotaxis, with the two clauses considered to be at the same level. Both types of subordination, in particular conditional clauses, have been studied extensively (Traugott et al., 1986; Athanasiadou and Dirven, 1997; Couper-Kuhlen and Kortmann, 2000; Declerck and Reed, 2001; Dancygier and Sweetser, 2005; among many others). However, instances in which the if-clause stands in isolation, as in (4a) below, with no matrix clause preceding or following the subordinate clause, have traditionally been neglected or treated in grammars as marginal cases or exceptions to the rule. For instance, Quirk et al. (1985: 841–842) refer to these structures as irregular sentences, and Huddleston and Pullum et al. (2002: 945) as a minor clause type, namely, conditional fragments. Major corpusbased grammars, such as Biber et al. (1999), also treat these structures as a minor type, under the label unembedded dependent clauses, but do note their frequency in conversation (Biber et al., 1999: 223, 1043). (4a)

Uhm <„> perhaps if you could tell me a little bit about your own father <„> (ICE-GB:S1A-072 #042:1:A)

Example (4a) above not only differs in formal terms from (2) and (3) – in that it does not contain a main clause preceded or followed by a subordinate clause – but also has a different function in discourse. While the if-clause in (2) expresses a conditional, its counterpart in (4a) issues a directive, thus being an alternative, pragmatically, to other constructions which may have the same illocutionary force, such as an interrogative, as in (4b), or an imperative, as in (4c). (4b) (4c)

Could you tell me a little bit about your own father? Tell me a bit about your own father

2 As pointed out by an anonymous reviewer, although the term insubordination was not coined until 2007, this linguistic phenomenon had been noted earlier. Ohori (1995) studies the pattern of adverbial clauses without main clauses in Japanese using the term “suspended clause constructions”, attributed to Fillmore et al. (1988). Likewise, studies on conditionality, such as Ford (1993), also mention that conditionals can be used without apodoses. However, with a couple of exceptions, the independent study of insubordination had largely been neglected prior to the publication of Evans’ study. 3 All the examples are extracted from the two corpora analysed, ICE-GB and the BNC, unless otherwise indicated; and they follow the transcription conventions used in the two corpora analysed. In ICE-GB, short pauses are indicated as <,> and long ones as <„>. In the BNC, pauses, laughs and other relevant details are marked in square brackets. Overlapping between speakers is indicated as <-j-> in both corpora.

44

C. Lastres-López / Language Sciences 66 (2018) 42–59

Constructions such as that in (4a) are instances of what Evans (2007) has called insubordination, as opposed to subordinate clauses, illustrated in (2) and (3). Although according to Evans (2007) insubordinate(d) is a prototypical label for the study of these constructions (Sansiñena et al., 2015; Evans and Watanabe, 2016a, 2016b; Heine et al., 2016; Kaltenböck, 2016), different terms have also been used by others: isolated if-clauses (Stirling, 1999; Mato-Míguez, 2014a, 2014b), covert-Q conditionals (Declerck and Reed, 2001), free conditionals (Lombardi Vallauri, 2010), and independent conditional clauses (D’Hertefelt, 2015). 3. Earlier research on insubordination As already noted, with a couple of exceptions, the study of insubordination was largely neglected until the publication “Insubordination and its uses” by Evans in 2007. Evans (2007: 367) defines insubordination as “the conventionalized main clause use of what, on prima facie grounds, appear to be formally subordinate clauses”. Thus, the assumption is that insubordinate clauses derive from complex sentences involving a main and a subordinate clause; the former being absent in the resulting construction, and the latter conventionalising to such an extent that it is understood as a main clause. At this stage, this means that the main clause is not elided, since it is not easily retrievable from the context, although it was so at a certain point in time. This is graphically illustrated in Table 1, which shows Evans’ (2007) four-stage model for the diachronic formation of insubordinate clauses. Although clauses are properly considered insubordinate when they reach the constructionalisation stage, previous studies on if-insubordination have considered insubordinate clauses at both stages 3 and 4 (Mato-Míguez, 2014a), given that the boundaries between these steps are sometimes blurred.

Table 1 The diachronic formation of insubordinate clauses (Evans, 2007: 370). Subordination (1)

Ellipsis (2)

Conventionalised ellipsis (3)

Reanalysis as main clause structure (4)

Subordinate Construction

Ellipsis of the main clause

Restriction of interpretation of ellipsed material

Conventionalised main clause use of formally subordinate clause (Constructionalisation)

In line with this, it is generally assumed that insubordinate clauses can be reconstructed. However, it is usually the case that insubordinate if-clauses, such as the one in (5a), can either be reconstructed as originally deriving from a full conditional clause or from an indirect interrogative complement clause, as illustrated in (5b) and (5c) respectively:4 (5a) (5b) (5c)

If he could speak to Gill (BNC: KBF 13475) If he could speak to Gill [I would be grateful/happy.] [I wonder] if he could speak to Gill

As mentioned, Evans (2007) proposes the origins of insubordinate constructions by means of ellipsis. However, this model has received some criticism and has recently been reformulated to accommodate other possibilities that may also explain the appearance of such structures in the language. Among others, Mithun (2008) rejects the possibility that these constructions emerge as a result of ellipsis, explaining their appearance in terms of “extension of markers to functions at levels beyond the sentence” (Mithun, 2008: 108), that is to say, insubordinate clauses extend from the sentence domain to the discourse situation. In a similar vein, Heine et al. (2016) reject the ellipsis hypothesis as exclusive, arguing that Evans’ (2007) and Mithun’s (2008) proposals may be regarded as complementary, although they reiterate the need for further research. On the other hand, Sansiñena et al.,’s (2015: 17–18) findings support Evans’ (2007) hypothesis, claiming that ellipsis provides an intermediate stage in the pathway from subordination to insubordination, contributing to the functional specialisation of the utterances. In addition, Cristofaro (2016) contemplates multiple pathways towards insubordination, in that structures which are different in origins frequently result in the same insubordinate construction. For example, she mentions that insubordinate if-requests may in some cases originate from ellipsis and in others from clausal disengagement, that is, from the reinterpretation of insubordinate clauses as self-standing units which, in the absence of a specific clause, elaborate on a general topic (Cristofaro, 2016: 402). In the light of these studies, it seems that more research is needed as a means of establishing solid grounds for any claim as to the diachronic development that has led to the formation of insubordinate clauses, although multiple or complementary pathways, such as those suggested by Evans (2007) and Mithun (2008), are by no means discarded.

4 Despite the fact that it is sometimes assumed that if-insubordination derives from a full conditional, there seems insufficient evidence to claim this on solid grounds. Indeed, Evans (2007: 380, 390) mentions this with a note of caution, although he is inclined towards the conditional hypothesis, as in (5b), since he argues that it is not possible to find parallel examples with insubordinate whether. However, the origin of these constructions, it seems, remains unclear (Cristofaro, 2016; Evans and Watanabe, 2016a; Heine et al., 2016).

C. Lastres-López / Language Sciences 66 (2018) 42–59

45

Insubordinate clauses display certain formal features that are generally regarded as distinctive of subordinate clauses. These characteristics are retained from the constructions from which they originally derive. For example, in the case of insubordinate if-clauses, the construction begins with the subordinating conjunction if. Other languages may also include other prototypically subordinate characteristics; for instance, insubordinate clauses in German maintain the verb in final position as is required in their subordinate counterparts. Evans (2007: 377) argues that the fewer formal features characteristic of a subordinate clause taken by the new insubordinate construction, the more independent is its use. Schwenter (1999) and Evans (2007) discuss some tests that reaffirm the status of insubordinate clauses as independent main clauses. In particular, two tests proposed by Schwenter (1999) for Spanish can be applied to insubordinate if-clauses, namely: (i) they do not generally allow a negative interpretation (e.g. *If he couldn’t speak to Jill5), whereas full conditionals can show both positive and negative polarity (e.g. If he could speak to Jill, I would be happy/ If he couldn’t speak to Jill, I would be happy); and (ii) they cannot occur with certain sentence-scope adverbs such as obviously (e.g. *Obviously, if he could speak to Jill), while their full conditional counterparts can (e.g. Obviously, I would be happy if he could speak to Jill). Furthermore, Mato-Míguez’s (2014b: 63–64) corpus study in English also raises the question of other syntactic features of insubordinate if-clauses. Like other main clauses, they can (i) coordinate with other sequences, and (ii) take dependent clauses, usually causal and temporal clauses. In addition to this, semantically, insubordinate if-clauses are more restricted than their subordinate counterparts. Evans and Watanabe (2016a: 21) mention that “the overall meaning of the construction does not precisely match what would be obtained from the corresponding full construction”. Thus, while conditionals express a hypothetical state of affairs, insubordinate if-clauses do not convey conditional meaning, but rather may show more certainty on the part of the speaker than full conditionals, since they usually expect an action that, most frequently, will indeed take place. For instance, in (6a) it is assumed that the addressee will bring what the speaker is asking for, while in (6b) this possibility seems to be more hypothetical. (6a) (6b)

If you could just bring it here (BNC, KBW 2791) If you could just bring it here it would be better

Hence, it can be observed that insubordinate if-clauses show restrictions compared with their subordinate counterparts, not only in terms of form – they do not allow conjunctions other than if, as opposed to conditional clauses, for example – but also with respect to their interpretation and distribution with certain elements. As mentioned, insubordinate structures are syntactically independent from their original main clauses and convey a number of different functions from their subordinate counterparts. Heine et al. (2016) discuss insubordination in relation to the notion of cooptation (Kaltenböck et al., 2011), arguing that insubordinate clauses can be regarded as subtypes of theticals or parentheticals. As such, insubordinate clauses are freed from syntactic constraints and, consequently, the meaning of the constructions “is no longer defined with reference to its syntactic function but is redefined by its new ‘environment’” (Heine et al., 2016: 43). Thus, by adopting new meanings and functions, their scope widens from syntax to the discourse situation, as also suggested by Mithun (2008). If-clauses have been studied by different authors with respect to insubordination, and equivalent constructions have also been analysed in other languages; among others, Finnish and Swedish (Lindström et al., 2016), Spanish (Schwenter, 1996, 1999, 2016a, 2016b; Montolío Durán, 1999), and a wide range of Germanic languages including Danish, Dutch, English, German, Icelandic, and Swedish (D’Hertefelt, 2015). Cross-linguistic studies evidence the multifunctionality of these structures, not only showing that insubordinate clauses can express a wide range of functions in discourse (Evans, 2007; Mithun, 2008; Sansiñena et al., 2015; Heine et al., 2016), but also revealing that these functions vary from language to language. If we take Spanish, for instance, originally-conditional clauses display different functions in their insubordinate counterparts in Spanish and English. While one of the main functions for if-insubordination in English has been attested to be the use of these constructions as directives (Stirling, 1999; Mato-Míguez, 2014a, 2016; Kaltenböck, 2016), their Spanish equivalents are often used to express disagreement or refutation, as in (7), in a use that is not possible for insubordinate ifclauses in English. (7)

A: Los primos van a llegar esta tarde. (Schwenter, 2016a: 90) ‘Our cousins are going to arrive this afternoon’ B: ¡Si ya han llegado esta mañana! *‘If they have already arrived this morning!’

In addition to this, prosody is often required to understand the functional classification of insubordinate if-clauses, and should thus be treated as a distinct characteristic here (Evans and Watanabe, 2016a: 25–26). Sometimes a specific prosody

5

As usual, ungrammatical examples are preceded by an asterisk.

46

C. Lastres-López / Language Sciences 66 (2018) 42–59

helps to determine the construction’s function in discourse, and indeed certain constructions are only permissible with a particular prosody. For instance, (8) can be understood as a request on the lines of May I just add one other thought?; yet a change in prosody, in the appropriate context, can also lead to the interpretation of the clause as a complaint. The importance of prosody has also been discussed in prior studies on insubordination, in particular in the case of Spanish insubordinated constructions (Gras, 2011; Elvira-García, 2015; Schwenter, 2016a), but has been largely overlooked in English, with the exception of Kaltenböck (2016).6 (8)

If I may just add one other thought (BNC, KB0 3508)

4. Revisiting insubordinate if-clauses in English Focusing our attention on English, the earliest study to deal specifically with the discourse functions of insubordinate if-clauses was Stirling (1999). Insubordinate if-clauses, it was argued, are distinguished from their subordinate counterparts in that they are main clauses in themselves. As such, they express functions different from their canonical full counterparts, as can be seen in (9), (10) and (11) for English. While example (9) indicates conditionality, the insubordinate if-clauses in (10) and (11) are used, respectively, as a directive (a request), and as an optative (expressing a wish), the two main functions of if-insubordination attested by Stirling (1999) in her analysis of Australian English: (9) (10) (11)

If you can demonstrate your ability to ride a bike safely on the road he will issue you with a certificate of validation (ICE-GB:S2A-054 #135:2:A) If you’ll just come next door (ICE-GB:S1A-089 #159:2:A) If only Denis Betts could have picked that ball up and got it out to Offiah (ICE-GB:S2A-004 #377:1:A)

Huddleston and Pullum et al. (2002: 945) prefer to use the term regret for optative insubordinate if-clauses, as in (11) above. These structures are usually introduced by if only and express a wish which often refers to a past time; as such, this wish indicates counterfactuality, hence the adequacy of the term regret as an implicature. Dancygier and Sweetser (2005: 217–219), in their extensive study on conditional clauses, refer to these constructions as monoclausal if-only wishes and argue that they should be considered as non-conditional constructions, deriving from a conditional use that leaves the consequent meaning to the addressee’s interpretation. In addition to the functions reported by Stirling (1999), Huddleston and Pullum et al. (2002: 945) identify an additional use of these constructions to express surprise, as illustrated in (12). Thus, in (12) the speaker conveys surprise at seeing his friend, despite the fact that the clause shows negative polarity. (12)

Well, if it isn’t my old friend Malcolm Duce! (Huddleston and Pullum et al., 2002: 945)

Mato-Míguez (2014a, 2014b) examines if-insubordination in English mainly departing from the two functions distinguished for these constructions by Stirling (1999), but focusing her attention on insubordinate directives exclusively and moving forward to compare these with other constructions which may also express directive meaning, namely full conditionals and imperatives (Mato-Míguez, 2016). Taking some of the functions already mentioned, plus certain others, D’Hertefelt (2015) provides one of the most extensive taxonomies for insubordinate if-clauses, distinguishing between deontic, evaluative, assertive, argumentative, reasoning and post-modifying types, and establishing subcategories within each one. Deontic is one of the richest categories in terms of function, comprising controlled and uncontrolled deontics; the former category is then subdivided into speaker-oriented utterances (requests and threats) and addressee-oriented ones (offers and suggestions). On the other hand, under the label uncontrolled deontics, only wishes are included. It can be observed, on the whole, that insubordinate if-clauses are polyfunctional constructions that allow speakers to convey a wide and rich range of functions. However, it should be noted that D’Hertefelt’s (2015) classification is not exclusively based on English data, but rather on several Germanic languages (Danish, Dutch, English, German, Icelandic, and Swedish). Consequently, the repertoire of functions for English insubordinate if-clauses is more restricted, and Kaltenböck (2016) proposes a twofold classification between performative and elaborative insubordinated if-clauses. The former category includes directives, optatives and exclamatives, grouping under a single umbrella term the functions proposed by Stirling (1999) and Huddleston and Pullum et al. (2002). The latter category, comprising D’Hertefelt’s (2015) argumentative, reasoning and post-modifying functions, considers elaborations of sentences uttered by the same or a different speaker. These elaborations cannot stand on their own, since they require the presence of the preceding utterance(s) to interpret and elaborate on them. For this reason, elaborative if-clauses of this type cannot be considered as purely insubordinate, but as adverbial clauses

6 For the purposes of this paper, however, prosody has not been taken into account, since many of the audio files are of insufficient quality, whereas others are simply not available.

C. Lastres-López / Language Sciences 66 (2018) 42–59

47

functioning as supplements,7 which have clear links with former pieces of discourse (Huddleston and Pullum et al., 2002: 25). Hence, elaborative insubordinate if-clauses have not been taken into account in the present study, on the grounds that insubordinate if-clauses are only truly insubordinate when they are fully independent and can stand on their own, as other main clauses do. Considering that D’Hertefelt (2015) is a cross-linguistic study and does not address if-insubordination specifically, it seems, to my knowledge, that only Stirling (1999), Mato-Míguez (2014a, 2014b, 2016) and Kaltenböck (2016) constitute the very few studies which have examined constructions such as the ones considered in this paper in English exclusively. As discussed, Kaltenböck (2016) considers a much wider approach to insubordination than that addressed in the present paper. Thus, he examines 97 clauses in the spoken component of ICE-GB, as compared to the 47 I will consider in the same corpus (c.f. Section 5). Stirling’s (1999) dataset comprises 44 clauses from Australian English, extracted from the General Practice Corpus – a relatively small spoken corpus of medical consultations – and from the Macquarie Dictionary Corpus – a 20 million-word corpus which mostly contains written language. Given the extremely low frequencies found considering the size of the sample, Stirling further analyses a corpus of Scottish English dialogues and reports considerably higher frequencies, but focuses on directives only. Mato-Míguez (2014a, 2014b) similarly concentrates on the directive subtype. Mato-Míguez (2016) extends prior analyses to compare insubordination, for which she only retrieves 52 clauses, with other constructions which may also express directive meaning which are more frequent. Thus, the studies of Stirling (1999) and Mato-Míguez (2014a, 2014b, 2016), probably due to the very few cases of insubordination found, consider only insubordinate directives and optatives – the former – and insubordinate directives and other directive constructions – the latter. This leaves aside other possible functions that insubordinate if-clauses may express, which I will address in Section 5.2.1, based on the analysis of data from ICE-GB and the BNC. 5. Empirical analysis 5.1. Methodology, description of the corpora and data extraction This paper aims to elucidate the uses and functions of insubordinate if-clauses in English. For this purpose, a corpus-based methodology has been adopted. The data analysed has been extracted from two corpora: ICE-GB and the BNC and, in line with prior studies, focuses on the spoken components therein. The analysis is restricted to spoken discourse since language innovations generally occur earlier in speech than in writing (Pintzuk, 2003: 525). Thus, a wider variety of functions is assumed to appear in spoken language prior to the incorporation of these into the written mode. ICE-GB is a one-million word corpus, fully tagged and parsed, which contains spoken and written language. For our purposes, only the spoken component has been analysed. This dataset contains 300 texts, totalling 637,682 words, and is divided into dialogic and monologic genres. Texts in the dialogic subcomponent are classified into private and public; the former category contains 90 direct conversations and 10 telephone calls, and the latter consists of 20 classroom lessons, 20 broadcast discussions, 10 broadcast interviews, 10 parliamentary debates, 10 legal cross-examinations, and 10 business transactions. The monologic dataset includes scripted and unscripted categories; the scripted component comprises 20 broadcast and 10 non-broadcast talks, and 20 broadcast news, whereas the unscripted category contains 20 spontaneous commentaries, 30 unscripted speeches, 10 demonstrations, and 10 legal presentations.8 The data analysed has been automatically retrieved from ICE-GB using the International Corpus of English Corpus Utility Programme (henceforth ICECUP) and a Fuzzy Tree Fragment (henceforth FTF). FTFs allow complex grammatical queries to be carried out by constructing a parse tree with as few or as many nodes as necessary. Each node contains three slots that can be completed or not depending on the specificity required in the search – hence the term fuzzy. The top left slot indicates the syntactic function, the top right specifies the category or form of the structure, and the bottom slot shows the features. Given that insubordinate clauses are not canonical structures recognised as such in the corpus, prior to the construction of the FTF, the Wizard function of ICECUP was used. This allowed for the detection of how instances of insubordination were marked in the corpus, as illustrated in Fig. 1 below, thus facilitating the construction of the FTF in the subsequent step. Manual analyses were required after the automatic extraction since other structures – mainly unfinished sentences, including false starts and changes of topic – were marked with the same parameters.9 Fig. 2 below illustrates the FTF used to extract cases of if-insubordination. This FTF indicates that the structure searched is a clause functioning as a parsing unit. This clause contains a subordinate phrase functioning as a subordinator and, within it, a subordinating conjunction if that functions as head of the subordinate phrase in clause-initial position.

7 Huddleston and Pullum et al. (2002: 25) define a supplement as a construction which “instead of being integrated into the constituent structure of the sentence as a dependent or coordinate, it is loosely attached, set off from the rest in speech by separate intonational phrasing and in writing by punctuation”. Supplements are similar to what Ono and Couper-Kuhlen define as an increment, “a grammatical extension of the already completed unit” (2007: 505); but while the latter only occur sentence-finally, the former are more flexible in their position. 8 For specific details on the structure of ICE-GB, see Nelson et al. (2002). 9 The automatic search with the FTF retrieved 173 tokens. From these, only 47 represent cases of insubordination (around 30% of the tokens extracted). This is due to the fact that many utterances are unfinished due to interruptions, false starts, or a change of topic, or are completed by a different speaker. Such cases have been excluded from the analysis.

48

C. Lastres-López / Language Sciences 66 (2018) 42–59

Fig. 1. Parsed tree of an insubordinate if-clause in ICE-GB.

Fig. 2. Automatic extraction of insubordinate if-clauses with the help of an FTF.

After the manual discard, a total of 47 clauses were established from ICE-GB. Their distribution in the corpus proved to be unequal across genres. Table 2 below sets out the absolute frequency of insubordinate if-clauses attested in each of the spoken subcomponents of the corpus, taking into consideration the number of sentences in each component and normalising the frequencies per million sentences,10 since the different text categories in the corpus are unequal in terms of size. Table 2 Frequency and distribution of insubordinate if-clauses across text categories in ICE-GB. Text category

Absolute frequency

Percentage

Number of sentences

Frequency per million sentences

Direct conversations Telephone calls Classroom lessons Broadcast discussions Broadcast interviews Parliamentary debates Legal cross-examinations Business transactions Spontaneous commentaries Unscripted speeches Demonstrations11 Legal presentations Broadcast news Non-broadcast speeches Broadcast talks Total

30 1 2 1 0 0 1 0 2 3 5 1 1 0 0 47

63.83% 2.13% 4.25% 2.13% 0% 0% 2.13% 0% 4.25% 6.38% 10.64% 2.13% 2.13% 0% 0% 100%

26,656 3337 4338 2949 1641 1075 1937 2777 4227 3500 1458 1065 2540 966 2435 60,901

1125.45 299.67 461.04 339.10 0 0 516.26 0 473.15 857.14 3429.35 938.97 393.70 0 0 8833.83

10 Frequencies have been normalised taking the number of sentences as a baseline rather than the number of words. Since we are dealing with a grammatical rather than a lexical phenomenon, it seems more reliable and accurate to use the number of sentences as a baseline, since not all words can be substituted by the construction studied here (Aarts et al., 2013). In spoken discourse, however, the notion of sentence is not always neat. In ICE-GB, sentences (‘text units’) correspond to either grammatical sentences or, in some cases, to coherent utterances. For further details about the compilation and annotation of ICE-GB and the BNC, see http://www.ucl.ac.uk/english-usage/projects/ice-gb/ and http://www.natcorp.ox.ac.uk/corpus/creating.xml respectively. 11 The results for demonstrations show the presence of an outlier. The 5 instances found in this component are misleading, since they all correspond to the same speaker. For this reason and despite their high frequency per million sentences, the results for this category have not been regarded as representative. Thus, considering the number of different speakers involved, it has been noted that the highest frequency of appearance is found in direct conversations. Likewise, although legal presentations and unscripted speeches have a normalised frequency similar to direct conversations, the results should be taken with caution, given the reduced number of tokens in these categories.

C. Lastres-López / Language Sciences 66 (2018) 42–59

49

Given that if-insubordination appears predominantly in direct conversations, in terms of both absolute and normalised frequencies,12 the results obtained from ICE-GB were implemented with data extracted from the conversations subcomponent of the BNC. The BNC is one hundred times larger than ICE-GB, and contains written and spoken discourse, totalling around 100 million words. In particular, the conversations subcomponent contains 4,012,457 words, making it more than 20 times larger than the conversations component of ICE-GB. This allows us to consider a greater number of insubordinate ifclauses in the analysis and, consequently, to test our research questions on more solid grounds. The data from the BNC have been extracted using the Lancaster University corpus interface.13 As opposed to ICE-GB, the BNC is tagged but not parsed, what makes the automatic extraction of the data more complex. Thus, the strategy used to retrieve instances of if-insubordination has been to extract if-clauses in initial position by doing a case-sensitive search, looking for capitalisation of the initial letter of the conjunction. This yielded a total of 2583 tokens which were then manually analysed to detect instances of insubordination. After this fine-grained analysis, a total of 137 instances of insubordinate ifclauses were considered from the conversation subcomponent of the BNC, which represent around 5% of the total number of if-clauses in sentence-initial position. The analysis of spoken discourse entails a number of difficulties compared with writing. One of these has been the considerable number of instances that have been excluded from the analysis of both corpora on the grounds that they include interruptions, false starts, changes of topic by the same or a different speaker, overlaps between speakers, or unclear words; examples are given in (13) and (14) below. These constitute highly frequent phenomena in speech, particularly in informal conversation (Biber et al., 1999: 1052–1066), but due to their difficulty of interpretation, such uncompleted clauses have been discarded from the analysis. (13) (14)

If it was me [unclear] [pause] [unclear] (BNC, KCF 1235) No. If you go <,> I tell you what go and ask Tony where it is (ICE-GB:S1A-017 #038-042:1:A)

In other cases, a close analysis has revealed many instances of discontinuous conditionals rather than insubordinate ifclauses. Two types of discontinuous conditional clause were recurrent across the two corpora: (i) a conditional protasis whose corresponding apodosis was uttered by a different speaker, as in (15), a subtype also discussed by Stirling (1999: 277–278); and (ii) a sequence of protasis and apodosis uttered by the same speaker but interrupted in the middle by a comment made by the addressee, as in (16). (15)

(16)

A: Danny could have had his operation three years ago! B: If you’d been on Bupa. (BNC, KCN 6852-6853) A: If you’re on the video channel and you want to change the channel B: When Nana’s got her Mitsubishi video <-j-> then you can use it all on one. A: <-j-> Then you use your video <-j-> remote control, oh (BNC, KC3 1566-1568)

In example (15), speaker B completes A’s utterance by stating the condition under which Danny could have had his operation. On the other hand, in (16) the condition expressed by A is interrupted by B; thus A utters the condition in two turns, the protasis in the first and the apodosis in the second. Other cases that may look like insubordinate clauses at first glance are in fact instances of multiple embedded if-clauses sharing the same apodosis, as in (17), in which the first if-clause is not insubordinated, but is part of a structure formed by two juxtaposed protases followed by a single apodosis. Thus, the two if-clauses in (17), highlighted in bold, are two conditions for the same consequent, italicised in the example. (17)

12 13

A: And he said er these Toshiba tellies have a, a timer on thirty sixty ninety one twenty B: Oh yeah A: you set the timer and hit the buttons whenever you want <-j-> it’ll turn itself off off the <-j-> B: <-j-> Switch off yeah <-j-> A: If ever you pick it up and er you get that bleeping noise of a fax somebody’s trying to send B: Mm A: if you just hit the start stop button and just hang on to the phone it’ll come through B: Oh right. A: after a little while. (BNC, KC1 1763-1771)

See footnote 11 above. http://bncweb.lancs.ac.uk/.

50

C. Lastres-López / Language Sciences 66 (2018) 42–59

5.2. Results and discussion As previously mentioned, after data extraction and discarding of examples, a total of 47 instances of if-insubordination was found in ICE-GB and 137 in the BNC, amounting to 184 in total. A database was created to analyse each clause manually, annotating each token according to a number of variables. In particular, the analysis concentrated on the different functions that insubordinate if-clauses perform in interaction and on other alternative syntactic patterns which may convey the same function. In addition, the issue of possible correlations between certain functions and specific verb types was addressed. 5.2.1. Discourse functions The corpus analysis has shown that insubordinate if-clauses can be employed for a wide range of functions in discourse, highlighting their polyfunctionality. Eight different functions have been attested in the corpora, belonging to two major groups: directives, including requests, suggestions, offers, threats, and permission; and non-directives, including, wishes, assertions/exclamations, and complaints. Directives are primarily used “to instruct somebody to do something” (Quirk et al., 1985: 804), whereas functions grouped under the heading ‘non-directives’ express different types of opinion, attitudes or evaluation on the part of the speaker. These functions, together with their frequencies, are summarised in Table 3 below and will be fully discussed in subsequent sections. In quantitative terms, the data analysed shows that insubordinate clauses function as directives in most cases (66.85%), but with different frequencies for each function, as will be discussed in Sections 5.2.1.1 to 5.2.1.5. On the other hand, non-directives, discussed in Sections 5.2.1.6 to 5.2.1.8, represent one third of the total number of insubordinate clauses, with wishes occupying the first position in the rank. Since in terms of illocutionary force directives are usually difficult to classify (Mato-Míguez, 2016: 302), this study distinguishes directives in terms of who is the agent and the beneficiary of the future action, in line with CouperKuhlen (2014).14 Table 3 Discourse functions of insubordinate if-clauses (absolute frequency and percentage), with special reference to the agent and beneficiary of the future action in the case of directives. Discourse function Directives 123 (66.85%)

Non-directives 61 (33.15%)

Total

Request 75 (40.76%) Suggestion 25 (13.59%) Offer 12 (6.52%) Threat 7 (3.81%) Permission15 4 (2.17%) Wish 39 (21.20%) Assertion/exclamation 12 (6.52%) Complaint 10 (5.43%) 184 (100%)

Agent of future action

Beneficiary of future action

Addressee

Speaker

(Speaker &) Addressee

(Speaker &) Addressee

Speaker

Addressee

Addressee

Speaker

Speaker

Speaker

5.2.1.1. Requests. As already noted, and in line with prior corpus-based studies, the results yielded from both corpora indicate that insubordinate if-clauses are most frequently used as requests (>40%, 75 tokens), as illustrated in example (18a) below. In requests, the speaker asks the addressee to carry out a certain action of which the former will eventually benefit. Requests can also be encoded by other syntactic constructions, such as interrogative clauses, as in (18b), or even imperative clauses, as in (18c).16 These structures may be considered as alternatives to the insubordinate pattern illustrated in (18a), in line with the

14 It should be noted that Couper-Kuhlen (2014: 634) classifies suggestions as those in which the agent and the beneficiary is the addressee (i.e. the “other” in Couper-Kuhlen’s terminology). Examples of insubordinate if-clauses in the corpora include many instances which can easily be paraphrased by a clause introduced by let’s, in which speaker and addressee are jointly agents and beneficiaries. For this reason, I have considered that the agent and beneficiary of a suggestion can be either the addressee alone or both the speaker and addressee. 15 Permission refers here to utterances that involve a request for permission (see Section 5.2.1.5). 16 Sometimes these constructions do not serve to request anything, but rather express an order, as in (19) below, uttered by a lecturer in a classroom context. Here, we have included these cases together with requests, since they are frequently ambiguous and difficult to classify, especially without knowing the intonation with which they were uttered, which is missing in some cases in the BNC. Another useful criterion to distinguish orders and requests is in terms of power relations between speakers and addressees (Pérez Hernández and Ruiz de Mendoza, 2002), but this information is, again, lacking in the BNC. (19) If I can continue <„> (ICE-GB:S1B-008 #124:2:A)

C. Lastres-López / Language Sciences 66 (2018) 42–59

51

findings reported by Mato-Míguez (2016) in her quantitative analysis of the variation between constructions with directive meaning, including insubordinate if-clauses and imperatives. (18a) (18b) (18c)

If you give me the key please (BNC, KB8 4079) Could you give me the key, please? Give me the key, please!

As expressions of politeness, insubordinate requests can be used together with downtoners, such as the adverb just; or with courtesy markers such as please, as in (18a) above. In addition, requests are often accompanied by modal verbs, such as can or could. Indeed, 42% (32 tokens) of the requests in the corpora contain modal verbs. Fig. 3 shows the percentage use of modal verbs in insubordinate requests and their absolute frequency.

50% 45%

14

40% 35% 30%

8

25%

7

20% 15% 10%

2

1

5% 0% Can

Could

Would

May

Will

Fig. 3. Absolute frequency and percentage of modal verbs in insubordinate requests.

Can is the most frequent modal verb in these constructions, followed by could and would, and to a lesser extent may and will. While can, could and may are frequently used in conditional protases, the use of would and will is less expected in ifclauses, since they are usually restricted to apodoses. The use of modal verbs in these constructions contributes to reinforce the request and can be paraphrased as if you wouldn’t mind, as can be seen in (20). (20)

If you’d like to put it on a little louder (BNC, KE0 3730)

Requests usually expect an action on the part of the addressee, as in (20), in which the expectation is that he or she will turn up the volume. For this reason, some authors have employed the term manipulative for this function (Declerck and Reed, 2001; Kaltenböck, 2016). 5.2.1.2. Suggestions. A total of 25 insubordinate if-suggestions were found in the corpora, more than 13% of the total number of insubordinate if-clauses. As with requests, suggestions also expect an action from the addressee, as in example (21a) below, in which the speaker makes a suggestion with the expectation that the addressee carries out the proposed action. Contrary to what happens in requests, the beneficiary of the future action suggested is the addressee (either alone or with the speaker). Very frequently, a sentence in isolation does not provide enough context to establish with certainty if the utterance represents a suggestion or another type of directive (Stirling, 1999: 278). Full contexts are usually required, as in (21a), in which we perceive the willingness of the addressee to accept the speaker’s suggestion. (21a)

A: If you book Serve (BNC, KC2 1146) B: Serve, yeah Serve takes them, yeah I know

Corpus-based results report that the most frequent subject in these constructions is the pronoun you, followed by we. Including the speaker in the suggestion may function as a softening device. This mitigation effect is also present in other syntactic constructions which, like insubordinate if-clauses, also encode suggestions, such as [let’s þ clause], as in (21b), or interrogatives with the form of [why don’t (we, you).?] or [how about.?], as shown in (21c) and (21d) below. Ford (1997: 401–405) argues that directives made through conditionals are a way of dealing with the negative face of the addressee since the conditional pattern allows more freedom of choice. This can be extended to insubordinate if-clauses, in which the construction seems to be more polite and detached compared to other directive constructions, and therefore is less imposing on the addressee.

52

C. Lastres-López / Language Sciences 66 (2018) 42–59

(21b) (21c) (21d)

Let’s book Serve Why don’t you book Serve? How about booking Serve?

5.2.1.3. Offers. Another function for insubordinate if-clauses attested in the corpora is that of an offer, representing around 6% of the total. Offers, as in (22a), constitute a different type of directive from requests and suggestions. Unlike these functions, an offer does not expect an action from the addressee, but rather from the speaker, who thus intends to benefit the addressee through some future action. This construction expresses meanings similar to (22b), but (22a) seems to convey the idea of asking the addressee whether he agrees with the offer or not, containing an implicature such as if that’s okay with you or other similar constructions. (22a) (22b)

If I bring you in a bottle of concentrate? (BNC, KP0 519) Can I bring you a bottle of concentrate?

5.2.1.4. Threats. Threats are infrequent functions for insubordinate if-clauses (<4% of the total, 7 tokens), and refer to a state of affairs “controlled by the addressee and evaluated as undesirable by the speaker” (D’Hertefelt, 2015: 99). In threats, the speaker dislikes the current states of affairs, but typically a threat would not only express this dislike but also indicate a consequence that will ensue if the situation does not change. In the case of insubordinate if-clauses this consequence is not explicitly mentioned, but the threat remains inherent in the construction, as in (23) below. In addition, it has been observed that these constructions are often accompanied by insulting and derogatory terms, as in (24) below. Extralinguistic factors, such as embodied conduct, would have been very useful for the classification of threats, but unfortunately this information is not included in the corpora analysed here. (23) (24)

If you don’t stand properly Matt! (BNC, KD7 2070) If you swear at me you cunt (BNC, KDA 8111)

5.2.1.5. Permission. Insubordinate if-clauses are also used to ask for permission, as in (25). D’Hertefelt (2015) establishes a parallelism between this function and requests, since both ask for something from the addressee. While in requests the speaker indirectly asks the addressee to carry out an action, in insubordinate if-clauses asking for permission it is the speaker who will perform the action rather than the addressee, the latter having a far more passive role, limited to granting the requested permission from which the speaker will benefit. This function has been attested as unusual for insubordinate ifclauses (2% of the total number of cases; 4 tokens). (25)

If I could add just a personal note (ICE-GB:S2B-020 #033:1:C)

5.2.1.6. Wishes. The data extracted from both corpora reports wishes as the second most frequent function for insubordinate if-clauses (>20%, 39 tokens). Among these, 15% follow the pattern [if only þ clause]. Unlike directives, wishes do not instruct anyone to do anything, but rather express a wish or regret on the part of the speaker. Two subtypes of wishes can be distinguished: irrealis and counterfactual wishes. While in the former the state of affairs expressed in the clause is unlikely to happen, in the latter “the potential realization [of the situation] lies in the past” (D’Hertefelt, 2015: 27) and is therefore impossible, since the speaker conveys a wish about a past event. Both types of wishes are intrinsically related to the verbal tenses used in the clause; irrealis wishes usually contain a verb in the simple past, whereas counterfactual wishes often use the past perfect. Examples (26) and (27) below show an irrealis and a counterfactual wish respectively. While in (26) the speaker regards the situation as unlikely, (27) conveys a regret about a situation that happened in the past. (26) (27)

If we had a regular bus service [pause] rather than a regular train service (BNC, KCS 332) If you’d made your mind up a bit earlier instead of farting about (BNC, KBE 4048)

The syntactic patterns with which they can alternate differ from those suggested for the various types of directives, since, for example, neither interrogatives nor imperatives can express wishes or desires. Instead, alternative patterns would include clauses introduced by I wish. 5.2.1.7. Assertions/Exclamations. Insubordinate if-clauses have been reported to express assertions/exclamations, as illustrated in (28) below, in approximately 6% of cases. All the examples retrieved belong to the BNC, indicating that they are characteristic of conversations. Previous studies recognise the existence of an exclamative function for insubordinate if-clauses, typically expressing a speaker’s indignation or surprise (Huddleston and Pullum et al., 2002; Kaltenböck, 2016). However, in structures such as (28), the speaker does not only show surprise, but also expresses his opinion by stating that something is the case (D’Hertefelt, 2015: 124). For instance, in (28) speaker B not only expresses surprise at the fact that A makes jam for someone, but also states that A is hardworking. Thus, since these constructions usually have the illocutionary force of an exclamation and express both functions simultaneously, the term “assertion/exclamation” seems more adequate for them.

C. Lastres-López / Language Sciences 66 (2018) 42–59

(28)

53

A: [.] and as he’s next door but one anyway, you know if, if anything goes wrong he can come along to me, but all he wanted was my homemade blackcurrant jam, and now he doesn’t come any more I’ve got to actually <-j-> give him <-j-> B: <-j-> Send some <-j-> round. A: I give him the blackcurrant jam, I give Myrene the blackcurrant jam, jam to take home for him. C: A: He loves it! C: Yeah, that’s right. I remember you saying. B: If you will be industrious. C: Mm mm mm. B: A: Oh I like, I qui– I quite like making jam. (BNC, KB8 4772)

5.2.1.8. Complaints. Corpus-based results reveal that insubordinate if-clauses may also be used by the speaker to convey a complaint or to reproach the addressee for something. Around 5% of the cases analysed are used with this function, one which does not seem to have been attested in prior studies, probably due to its low frequency (10 tokens). Example (29) illustrates an insubordinate if-complaint. In (29), speaker B complains about A’s previous question, and at the same time complains about the quality of the cleaning of the shoes. B’s utterance is ironic and at the same time reproaches A, since the implicature is that A is not able to clean the shoes better. (29)

A: Can’t you get them cleaner? (BNC, KDN 1902) B: If you can get them cleaner than that A: I haven’t got a decent pair of shoes you know!

5.2.2. Distribution per discourse function in the corpora As discussed in previous sections, insubordinate clauses introduced by if are polyfunctional constructions. The eight different functions identified in the corpora are presented in Fig. 4 in decreasing order of total frequency in ICE-GB and the BNC.

60% 50% 40% 30% 20% 10% 0%

ICE-GB

BNC

Fig. 4. Distribution of insubordinate if-clauses per function in ICE-GB and the BNC (in percentages).

We can observe that insubordinate if-clauses are most frequently used as requests in both corpora. The fact that their percentage is higher in ICE-GB than in the BNC may suggest that this function is not particularly characteristic of the genre of conversation, but it should also be noted that insubordinate if-clauses in other ICE-GB genres are only attested in very small numbers. The opposite is the case with wishes, the second most frequent function, which has been reported to be more likely in conversations than in spoken discourse in general. Suggestions are ranked third for frequency in these constructions. Of the remaining five functions (offers, assertions/exclamations, complaints, threats, and permission), these are all considerably less frequent and show much more variation between the two corpora; for instance, the assertion/exclamation and the threat function have only been reported in the BNC. This seems to reinforce the idea that more discourse-functional variety can be found in conversations than in other spoken genres, probably due to their spontaneity and informality, which makes them more suitable for linguistic innovation; but also, that the two functions attested by Stirling (1999) are the most frequent, despite the fact that they only represent around 60% of the total number of cases. Details of the absolute frequency and percentage by function in each corpus are provided in Table 4 below.

54

C. Lastres-López / Language Sciences 66 (2018) 42–59

Table 4 Absolute frequency and percentage of insubordinate if-clauses in ICE-GB and the BNC by discourse function. Discourse function

Request Wish Suggestion Offer Assertion/Exclamation Complaint Threat Permission Total

ICE-GB

BNC

Total

N

%

N

%

N

%

26 8 5 2 0 3 0 3 47

55.32% 17.02% 10.64% 4.26% 0% 6.38% 0% 6.38% 100%

49 31 20 10 12 7 7 1 137

35.77% 22.62% 14.60% 7.30% 8.76% 5.11% 5.11% 0.73% 100%

75 39 25 12 12 10 7 4 184

40.76% 21.20% 13.59% 6.52% 6.52% 5.43% 3.81% 2.17% 100%

5.2.3. Correlation between discourse functions and verbs The corpus-based analysis reveals a relatively high number of functions for insubordinate if-clauses. In this section, the relationship between verbs and the discourse functions of the constructions is examined. On the one hand, this correlation is explored considering the dual classification of modals and non-modals and, on the other, the classification of lexical verbs following the semantic typology for processes (i.e. verbs) proposed by Halliday and Matthiessen (2014: 213–220), looking at whether certain functions favour specific sets of lexical verbs. 5.2.3.1. Correlation between discourse functions and modal verbs. The results show a tendency for modal verbs to appear in insubordinate if-clauses; of the 184 tokens analysed, 58 contain a modal verb, representing around 31% of the total. Table 5 below shows the absolute frequency and percentage of insubordinate if-clauses per function containing modal verbs. Table 5 Absolute frequency and percentage of insubordinate if-clauses containing a modal verb (per function). Function

N

N and % of clauses containing a modal verb

Request Wish Suggestion Offer Assertion/Exclamation Complaint Threat Permission

75 39 25 12 12 10 7 4

31 (41.33%) 18 (46.15%) 1 (4%) 4 (33.33%) 1 (8.33%) 1 (10%) 0 (0%) 2 (50%)

Per function, the highest percentage of clauses containing a modal verb corresponds to permission (50%),17 followed by wishes (46%), requests (41%) and offers (33%). Modal verbs are much more infrequent in the other functions. These differences in the use of modal verbs with certain functions seem to be attributed to issues of politeness in the case of directives. Directives uttered by means of an insubordinate clause soften the strength which an imperative would have. As such, modal verbs are more frequent in the cases in which the speaker will be the beneficiary of the future action (see Table 3 above), since those involve a stronger imposition for the addressee that the speaker tries to mitigate by means of an insubordinate clause. In the light of the notion of entitlement, as discussed by Heinemann (2006), it seems that the speaker is less entitled to issue the directives in certain cases and uses a modal in the clause for such contexts. Thus, the insubordinate construction together with the modal verb mitigates the force of the directive and deals with the negative face issues of the addressee. Excluded from this are threats, which unexplicitly invoke a sanction or a negative consequence and are far away from being polite; and also other types of less imposing directives for the addressee, such as suggestions or offers. On the other hand, in the case of wishes, the considerable high frequency of modal verbs can be explained in connection with the notion of irrealis. As it happens with prototypical hypothetical conditionals, insubordinate wishes convey a regret about something that is unlikely or impossible to happen and the modal verb reinforces this idea of irreality or unlikelihood. This may also explain the low frequency of modal verbs in the other two types of non-directive functions (assertions/exclamations and complaints), since these do not refer to hypothetical states of affairs but to real ones. Newcombe-Wilson intervals were carried out at a 95% confidence level to determine whether having a modal verb in the clause increases the likelihood of having a certain discourse function, in particular in the case of functions that contain in more than a third of the cases if with modal verb. The results from the test are not statistically significant at p < 0.05, with the exception of suggestions, as shown in Fig. 5 below. It has been found that the probability of a suggestion occurring without a

17 Results for permission should be taken with caution, since there are only 4 instances of this function. In functions with low absolute frequencies, further studies should be carried out with larger datasets to confirm any correlation with modal verbs.

C. Lastres-López / Language Sciences 66 (2018) 42–59

55

modal verb, rather than with one, is statistically significant. However, in the remaining cases, modal verbs have not been found to be predictors of the discourse function, despite their considerably high occurrence in certain functions.

Fig. 5. Probability of insubordinate if-suggestions occurring with modal verbs, including 95% Newcombe-Wilson confidence intervals.

5.2.3.2. Correlation between discourse functions and lexical verbs. As noted in 3.2.2, lexical verbs have been classified semantically following the process typology proposed by Halliday and Matthiessen (2014: 213–220), also earlier discussed within the framework of Systemic Functional Grammar, for example, in Halliday (1994). This typology seems to be adequate for the study of insubordinate if-clauses, since the set of process types identified by Halliday as making up what he terms the system of clause transitivity offer a comprehensive and useful catalogue of the various kinds of verbal events that can be found in clauses. Thus, following this classification, verbs have been categorised depending on their meaning in material (verbs of doing), verbal (verbs of saying), mental (verbs of thinking), and relational (verbs of being or having)18. Material verbs are verbs of doing or happening (e.g. bring, fall or give), as in John fell down, which represent a change in the flow of events. An example of an insubordinate clause with a material verb is presented in (30). (30)

If you could just bring it here (BNC, KBW 2791)

Verbal verbs are verbs of saying and communicating (e.g. say, tell or announce). These are “symbolic relationships constructed in human consciousness and enacted in the form of language” (Halliday and Matthiessen, 2014: 215), as in John said that he was tired or as in (31), which shows an insubordinate clause with this type of verb. (31)

If you tell anybody (BNC, KD2 2995)

Mental verbs are verbs of thinking or sensing (e.g. think, see or feel). They are concerned with our experience of the world in our consciousness, as in John thinks Sarah is wise. An insubordinate if-clause with a mental verb is illustrated in (32) below. (32)

If she could only remember where she’d been [pause] and how much she’s enjoying it while she’s actually doing it (BNC, KB8 6708)

Relational verbs are verbs of being or having (e.g. is or have), as in Sarah is wise, in which a relation is established between two separate entities. Example (33) shows a relational verb in an insubordinate clause. (33)

If that’s not greasy! (BNC, KBL 862)

18 Halliday and Matthiessen (2014) also recognise two further categories: behavioural and existential, but these are usually regarded as less frequent. In fact, neither of these two types was found in the corpus-based analysis.

56

C. Lastres-López / Language Sciences 66 (2018) 42–59

Table 6 below sets out the absolute frequency and percentage of the correlation between the functions performed by ifinsubordination and the types of verbs used in the clauses. Table 6 Correlation between functional types of insubordinate if-clauses and types of verbs. Function

Verb

Request Wish Suggestion Offer Assertion/Exclamation Complaint Threat Permission

N % N % N % N % N % N % N % N %

Material

Verbal

Mental

Relational

Other19

Total

54 72% 22 56.41% 20 80% 4 33.33% 0 0% 3 30% 4 57.13% 2 50%

1 1.33% 1 2.56% 0 0% 0 0% 0 0% 2 20% 1 14.29% 0 0%

16 21.34% 4 10.26% 3 12% 6 50% 4 33.33% 3 30% 1 14.29% 2 50%

4 5.33% 11 28.21% 2 8% 2 16.67% 8 66.67% 2 20% 1 14.29% 0 0%

0 0% 1 2.56% 0 0% 0 0% 0 0 0 0% 0 0% 0 0%

75 100% 39 100% 25 100% 12 100% 12 100% 10 100% 7 100% 4 100%

It can be observed that certain functions clearly favour the use of a specific set of verbs. Requests most frequently occur with material verbs, as in (30) above. Material verbs express actions and thus seem appropriate for this type of if-clauses, usually uttered with the intention of provoking an action on the part of the addressee. This may also explain the high proportion of material verbs in other types of directives as well, as is the case of suggestions. Non-directive functions, such as wishes, also show a considerable proportion of material verbs, probably because the speaker wishes that something (an action) would happen or had happened, as in (34) below. (34)

If they would just cooperate a bit (BNC, KB8 11558)

Other frequent combinations of functions and verb types include offers with mental verbs, assertions/exclamations with relational verbs, and threats with material verbs. Newcombe-Wilson confidence intervals were conducted to examine the probability of all possible combinations, resulting in three function-verb type patterns which show statistical significance. Newcombe-Wilson intervals at 95% confidence level show that the probability of a material verb occurring increases when the clause expresses a request. The results reveal a probability of around 0.7, and this difference is statistically significant at an 0.05 error level, as shown in Fig. 6 below, which compares requests with other discourse functions.

Fig. 6. Probability of material verbs occurring with insubordinate requests, including 95% Newcombe-Wilson confidence intervals.

19

The label other includes an instance of an elided lexical verb.

C. Lastres-López / Language Sciences 66 (2018) 42–59

57

Likewise, the relationship between requests and relational verbs also displays statistical significance at a 95% confidence level. As observed in Fig. 7, the probability of a request being used with a relational verb is very low, around 0.1, and this difference is statistically significant at p < 0.05 when compared with other functions. This is not surprising since requests are not used to ascribe properties, as would happen with relational verbs, but rather they are used with the expectation of provoking certain actions on the part of the addressee.

Fig. 7. Probability of relational verbs occurring with insubordinate requests, including 95% Newcombe-Wilson confidence intervals.

Turning to assertions/exclamations, Newcombe-Wilson confidence intervals reveal that this function appears more frequently with relational verbs. It makes sense to me that insubordinate if-clauses conveying assertions/exclamations should show a preference for relational verbs, because the essence of an assertion and an exclamation is ascribing a property to something else, resulting in an abundant use of the verb be, as in example (33) above. This difference in the probability when compared to other functions is statistically significant at a 0.05 error level, as shown in Fig. 8.

Fig. 8. Probability of insubordinate if-assertions/exclamations occurring with relational verbs, including 95% Newcombe-Wilson confidence intervals.

As also mentioned, relationships between other functions and other verb types have not been found to be statistically significant. In the light of the above, there seems to exist a correlation between lexical verbs, taken as families of meanings, and the discourse function expressed by the clause, with some functions favouring or disfavouring the use of certain verb types.

58

C. Lastres-López / Language Sciences 66 (2018) 42–59

6. Concluding remarks This study has addressed insubordinate if-clauses in English as independent constructions, different from full conditionals and indirect interrogative complement clauses. The need for corpus-based studies on the topic has been noted by Evans and Watanabe (2016a) as a priority in the study of insubordination. Previous research on if-insubordination in English has attested very low frequencies for the constructions under analysis, resulting in typologies of functions that are not exhaustive enough so as to cover all the possibilities insubordinate if-clauses can have in English. This paper has aimed at filling this gap by drawing data from a larger sample which comprises two corpora that can be regarded as complementary, ICE-GB and the BNC. While ICE-GB provides an overview of insubordination in different genres, the data from the BNC concentrate on the type of text in which they appear more recurrently, that is to say, conversations. A fine-grained analysis of these constructions from a quantitative and qualitative perspective has revealed their polyfunctionality and has uncovered some functions that were absent in prior analyses on the topic, such as complaints. Functions usually grouped together as a homogeneous group in previous studies, as is the case of directives, have been redefined here in terms of agent and beneficiary of the future action, providing a more detailed typology. In line with previous studies, requests and wishes have been attested as the most frequent discourse functions of insubordinate if-clauses. The data analysed from British English in this paper does not differ in this respect from the results reported by Stirling (1999) for Australian English, but further research is needed to compare these results with other national varieties. The results, however, have also reported other functions for insubordinate if-clauses, such as suggestions, offers, assertions/exclamations, complaints, threats, and permission. It has also been pointed out that these constructions compete with other syntactic structures that may convey the same function in discourse, and that these syntactic alternatives are different depending on the function the clause expresses. Furthermore, we have looked at whether certain discourse functions favour the occurrence of certain verbs, considered both in their classification modals/non-modals and in the semantic taxonomy proposed by Halliday and Matthiessen (2014), given that prior research has not examined possible connections between discourse functions and other constituents in the clause. The results reveal that one third of the total number of insubordinate if-clauses occur with modal verbs, which are used as mitigating devices to soften the force of the directive or as expressions of irrealis meaning to emphasize the idea of hypotheticality in wishes. Regarding lexical verbs, 95% Newcombe-Wilson confidence intervals have reported statistically significant results for certain combinations of functions and verb types, namely: requests and material verbs, requests and relational verbs, and assertions/exclamations and relational verbs. This suggests a tendency for a correlation between these types of verbs and the function expressed by the insubordinate if-clause, leading to the conclusion that certain functions may predict the type of verb used in the clause, although further research is needed to confirm these results in low-frequency functions. Acknowledgements I would like to express my gratitude to Bas Aarts, Teresa Fanego, and Sean Wallis for invaluable suggestions and feedback on earlier versions of this paper, as well as to an anonymous reviewer for very useful comments and observations. My gratitude is also extended to Nick Evans and Gunther Kaltenböck, who provided me with some of their (then) unpublished materials on insubordination. I would also like to thank the financial support provided by the Spanish Ministry of Education, Culture and Sport (FPU2015/02519), the Spanish Ministry of Economy, Industry and Competitiveness (FFI2014-52188-P), and the European Regional Development Fund and the Regional Government of Galicia (Directorate General for Scientific and Technological Promotion, grants ED431D 2017/09 and ED431B 2017/12); these grants are hereby gratefully acknowledged. References Aarts, B., Close, J., Wallis, S., 2013. Choices over time: methodological issues in investigating current change. In: Aarts, B., Close, J., Leech, G., Wallis, S. (Eds.), The Verb Phrase in English. Investigating Recent Language Change with Corpora. Cambridge University Press, Cambridge, pp. 14–43. Athanasiadou, A., Dirven, R. (Eds.), 1997. On Conditionals Again. John Benjamins, Amsterdam. Biber, D., Johansson, S., Leech, G., Conrad, S., Finegan, E., 1999. Longman Grammar of Spoken and Written English. Longman, Harlow. Bybee, J.L., Noonan, M. (Eds.), 2002. Complex Sentences in Grammar and Discourse. Essays in Honor of Sandra A. Thompson. John Benjamins, Amsterdam. Couper-Kuhlen, E., 2014. What does grammar tell us about action? Pragmatics 24 (3), 623–647. Couper-Kuhlen, E., Kortmann, B. (Eds.), 2000. Cause, Condition, Concession, Contrast: Cognitive and Discourse Perspectives. Mouton de Gruyter, Berlin. Cristofaro, S., 2016. Routes to insubordination. A cross-linguistic perspective. In: Evans, N., Watanabe, H. (Eds.), Insubordination. John Benjamins, Amsterdam, pp. 395–424. Dancygier, B., Sweetser, E., 2005. Mental Spaces in Grammar: Conditional Constructions. Cambridge University Press, Cambridge. Declerck, R., Reed, S., 2001. Conditionals: A Comprehensive Empirical Analysis. Mouton de Gruyter, Berlin. D’Hertefelt, S., 2015. Insubordination in Germanic: A Typology of Complement and Conditional Constructions. KU Leuven, Leuven. Ehmer, O., Barth-Weingarten, D., 2016. Adverbial patterns in interaction. Lang. Sci. 58, 1–7. Elvira-García, W., 2015. Prosody as a phonological cue for differentiating between elliptical and insubordinated constructions in Spanish. In: Paper given at the 48th Annual Meeting of the Societas Linguistica Europaea. Leiden, 2–5 September 2015. Evans, N., 2007. Insubordination and its uses. In: Nicolavea, I. (Ed.), Finiteness: Theoretical and Empirical Foundations. Oxford University Press, Oxford, pp. 366–431. Evans, N., Watanabe, H., 2016a. The dynamics of insubordination: An overview. In: Evans, N., Watanabe, H. (Eds.), Insubordination. John Benjamins, Amsterdam, pp. 1–38. Evans, N., Watanabe, H. (Eds.), 2016b. Insubordination. John Benjamins, Amsterdam. Fillmore, C.J., Kay, P., O’Connor, M.C., 1988. Regularity and idiomaticity in grammatical constructions: The case of ’let alone’. Language 64 (4), 501–538. Ford, C., 1993. Grammar in Interaction: Adverbial Clauses in American English Conversations. Cambridge University Press, Cambridge.

C. Lastres-López / Language Sciences 66 (2018) 42–59

59

Ford, C., 1997. Speaking conditionally: Some contexts for if-clauses in conversation. In: Athanasiadou, A., Dirven, R. (Eds.), On Conditionals Again. John Benjamins, Amsterdam, pp. 387–413. Gras, P., 2011. Gramática de construcciones en interacción: Propuesta de un modelo y aplicación al análisis de estructuras independientes con marcas de subordinación en español. Universidad de Barcelona, Barcelona. Greenbaum, S., Nelson, G., 1995. Clause relationships in spoken and written English. Funct. Lang. 2 (1), 1–21. Greenbaum, S., Nelson, G., 1996. Positions of adverbial clauses in British English. World Englishes 15 (1), 69–81. Halliday, M., 1994. An Introduction to Functional Grammar. Arnold, London. Halliday, M., Matthiessen, C., 2014. Halliday’s Introduction to Functional Grammar. Routledge, London. Heine, B., Kaltenböck, G., Kuteva, T., 2016. On insubordination and cooptation. In: Evans, N., Watanabe, H. (Eds.), Insubordination. John Benjamins, Amsterdam, pp. 39–64. Heinemann, T., 2006. ’Will you or can’t you?’: Displaying entitlement in interrogative requests. J. Pragmat. 38, 1081–1104. Huddleston, R., Pullum, G.K., et al., 2002. The Cambridge Grammar of the English Language. Cambridge University Press, Cambridge. Kaltenböck, G., 2016. On the grammatical status of insubordinate if-clauses. In: Kaltenböck, G., Keizer, E., Lohmann, A. (Eds.), Outside the Clause: Form and Function of Extra-clausal Constituents. John Benjamins, Amsterdam, pp. 341–378. Kaltenböck, G., Heine, B., Kuteva, T., 2011. On thetical grammar. Stud. Lang. 35 (4), 848–893. Kortmann, B., 2012. Adverbial Subordination. A Typology and History of Adverbial Subordinators Based on European Languages. De Gruyter Mouton, Berlin. Lindström, J., Lindholm, C., Laury, R., 2016. The interactional emergence of conditional clauses as directives: constructions, trajectories and sequences of actions. Lang. Sci. 58, 8–21. Lombardi Vallauri, E., 2010. Free conditionals in discourse: the forming of a construction. Linguisticae Investigationes 33 (1), 50–85. Mato-Míguez, B., 2014a. If you would like to lead: on the grammatical status of directive isolated if-clauses in spoken British English. In: Alcaraz-Sintes, A., Varela-Hernández, S. (Eds.), Diachrony and Synchrony in English Corpus Linguistics. Peter Lang, Bern, pp. 259–283. Mato-Míguez, B., 2014b. Are isolated if-clauses independent clauses? Evidence from spoken British and American English. In: Rodríguez-Puente, P., Fanego, T., Gandón-Chapela, E., Riveiro-Outeiral, S.M., Roca-Varela, M.L. (Eds.), Current Research in Applied Linguistics: Issues on Language and Cognition. Cambridge Scholars Publishing, Newcastle upon Tyne, pp. 48–73. Mato-Míguez, B., 2016. The expression of directive meaning: A corpus-based study on the variation between imperatives, conditionals, and insubordinate if-clauses in spoken British English. In: López-Couso, M.J., Méndez-Naya, B., Núñez-Pertejo, P., Palacios-Martínez, I.M. (Eds.), Corpus Linguistics on the Move: Exploring and Understanding English through Corpora. Brill, Leiden, pp. 291–312. Matthiessen, C., Thompson, S.A., 1988. The structure of discourse and ’subordination’. In: Haiman, J., Thompson, S.A. (Eds.), Clause Combining and Grammar in Discourse. John Benjamins, Amsterdam, pp. 275–330. Mithun, M., 2008. The extension of dependency beyond the sentence. Language 84 (1), 69–119. Montolío Durán, E., 1999. ¡Si nunca he dicho que estuviera enamorada de él! Sobre construcciones independientes introducidas por si con valor replicativo. Oralia 2, 37–69. Nelson, G., Wallis, S., Aarts, B., 2002. Exploring Natural Language: The British Component of the International Corpus of English. John Benjamins, Amsterdam. Ohori, T., 1995. Remarks on suspended clauses: A contribution to Japanese phraseology. In: Shibatani, M., Thompson, S.A. (Eds.), Essays in Semantics and Pragmatics. John Benjamins, Amsterdam, pp. 201–218. Ono, T., Couper-Kuhlen, E., 2007. Increments in cross-linguistic perspective: Introductory remarks. Pragmatics 17 (4), 505–512. Pérez-Hernández, L., Ruiz de Mendoza, F.J., 2002. Grounding, semantic motivation and conceptual interaction in indirect directive speech acts. J. Pragmat. 34, 259–284. Pintzuk, S., 2003. Variationist approaches to syntactic change. In: Joseph, B.D., Janda, R.D. (Eds.), The Handbook of Historical Linguistics. Blackwell, Oxford, pp. 509–528. Quirk, R., Greenbaum, S., Leech, G., Svartvik, J., 1985. A Comprehensive Grammar of the English Language. Longman, Harlow. Sansiñena, M.S., De Smet, H., Cornillie, B., 2015. Between subordinate and insubordinate. Paths toward complementizer-initial main clauses. J. Pragmat. 77, 3–19. Schwenter, S., 1996. The pragmatics of independent si-clauses in Spanish. Hisp. Ling. 8, 316–351. Schwenter, S., 1999. Sobre la sintaxis de una construcción coloquial: oraciones independientes con si. Anu. Filolog. 21, 87–100. Schwenter, S., 2016a. Independent si-clauses in Spanish: Functions and consequences for insubordination. In: Evans, N., Watanabe, H. (Eds.), Insubordination. John Benjamins, Amsterdam, pp. 89–112. Schwenter, S., 2016b. Meaning and interaction in Spanish independent si-clauses. Lang. Sci. 58, 22–34. Stirling, L., 1999. Isolated if-clauses in Australian English. In: Collins, P., Lee, D. (Eds.), The Clause in English: In Honour of Rodney Huddleston. John Benjamins, Amsterdam, pp. 273–294. Traugott, E.C., Meulen, A., Reilly, J.S., Ferguson, C.A. (Eds.), 1986. On Conditionals. Cambridge University Press, Cambridge. Traugott, E.C., 2017. ’Insubordination’ in the light of the Uniformitarian Principle. Engl. Lang. Ling. 21 (2), 289–310. Verstraete, J.-C., 2011. Rethinking the Coordinate-Subordinate Dichotomy. Interpersonal Grammar and the Analysis of Adverbial Clauses in English. De Gruyter Mouton, Berlin.