Insubordinate if-clauses in FDG: Degrees of independence

2.1 How to account for an “unruly” category?

As mentioned before, insubordinate clauses present a challenge for grammatical representation. Their formal features (e.g. presence of a subordinator, subordinate-clause word order) signal subordination, while their syntactic independence suggests main clause status. It is this Janus-like nature that makes them live up to the double-meaning of the term “insubordination” as “unruly” constructions which are difficult to account for in grammar.

Not surprisingly, therefore, insubordinate clauses were either ignored in earlier linguistic work or marginalized as anomalies. Quirk et al. (1985, 841–3), for instance, classify insubordinate clauses as so-called “irregular sentences” on account of their “not conform[ing] to the regular patterns of clause structures” (Quirk et al. 1985, 838). Similarly, Huddleston (2002, 944) includes them in the category of “minor clause types,” which comprises “a number of main clause constructions that do not belong to any of the major clause types.”^[1]

Another way of dealing with insubordinate clauses in the grammatical literature is to treat them simply as performance features which involve ellipsis. The problem with ellipsis-based accounts is, however, that the missing matrix clause is not always fully and unambiguously recoverable from the context (as shown, e.g. for if-clauses by Stirling 1998, Lombardi Vallauri 2004, Mato-Míguez 2014a,b). Moreover, an ellipsis approach fails to account for cases where the subordinate clause has a complete (i.e. terminal) prosodic contour (e.g. Schwenter 2016, Elvira-García et al. 2017), has its own illocutionary force, and shows structural signs typical of the main clause, such as the ability to coordinate with another main clause or the ability to take a subordinate clause as its dependent (e.g. Mato-Míguez 2014a,b). Features such as these suggest that we are not dealing with incomplete structures which are the result of performance “accidents,” but rather with deliberately produced and complete constructions.

2.2 Distinguishing types of insubordination

As a typologically widespread phenomenon which potentially involves the whole range of subordinate clause types, insubordination encompasses a number of different forms and functions. In terms of form, it is possible to distinguish different types of subordinate features, such as infinitive, participial, or subjunctive inflections of the verb, subordinate word order, as well as different types of subordinators. In terms of function, Evans (2007), for instance, has identified a wide range of options which include interpersonal control (e.g. warnings and requests), modal meaning (e.g. epistemic, evidential, deontic, exclamation, and evaluation), and signalling presupposed material (e.g. negation, contrast, and reiteration). Despite the range of pragmatic possibilities offered by this form, the functions of insubordination are still relatively constrained in that they typically involve the expression of interpersonal relations such as conveying speaker attitudes and the management of speaker–hearer interactions (e.g. Van linden and Van de Velde 2014, 228, Sansiñena et al. 2015, 16, Heine et al. 2016, Martínez Caro and Alba-Juez 2021).

Another parameter for distinguishing different types of insubordination is an expression’s degree of autonomy or independence from the co-text. The possibilities here range from completely autonomous to semi-autonomous insubordinate clauses (cf., e.g. Kaltenböck 2019). D’Hertefelt and Verstraete (2014), for instance, distinguish two types of Swedish and Danish at(t) constructions: (i) expressives, which are fairly autonomous, and (ii) elaboratives, which elaborate on something that was said before by the same speaker or a different one and as such are pragmatically dependent. They argue that only the former are cases of insubordination proper, while the latter are identified as cases of dependency shift. Similarly, Verstraete et al. (2012) identify the type of “discursive insubordination,” which is used to expand on an utterance from the preceding discourse. An even wider scope of dependency has been observed by Mithun (2008) in her study of Navajo and Yup’ik markers, which operate over larger stretches of discourse. Yet another type of pragmatic dependency can be found in dyadically dependent insubordinations (Gras 2016, Sansiñena 2015, Sansiñena et al. 2015, Gras and Sansiñena 2015), which can be analysed as subordinate clauses which relate to the main clause in a previous turn (cf. Kaltenböck 2019). These cases have also been referred to as “collaborative insubordinations” (Hilpert 2015).

Finally, insubordinate structures can also be distinguished in terms of their degree of conventionalization and fixation (e.g. Heine et al. 2016; Section 3). While most insubordinations can be considered as constructionalized at least to some extent (i.e. being compositional but having schematic structure and function with some degree of fixation of morphosyntax and prosody; e.g. If you come over here, If only he wouldn’t say that), some have developed into even more formulaic structures (i.e. non-compositional and fairly invariable units; e.g. if you will, if I may, if you don’t mind). At the other end of the spectrum we find what has been called “instantaneous insubordination” (Heine et al. 2016), that is, instances that are felt to be elliptic with their meaning deriving only from their link to the context.

Insubordination thus does not represent a uniform, monolithic category but is marked by considerable variation in its structure, discourse function as well as degree of autonomy and conventionalization. Section 3 gives an overview of the different taxonomies proposed for insubordinate if-clauses in the literature.

4.1 Some general features

FDG is a functional theory in that it views language first and foremost as a means of communication and regards linguistic form as emerging from communicative function. More specifically, FDG takes a “form-oriented ‘function-to-form’ approach” to language analysis (Hengeveld and Mackenzie 2008, 39). It is “function-to-form” in that it takes as its input a speaker’s communicative intentions, which the grammar subsequently translates into the appropriate linguistic form. It is “form-oriented” in the sense that it only considers those aspects of meaning that are systematically reflected in the morphosyntactic and/or phonological form of an utterance.

One of the distinctive features of FDG is that it does not take the clause or the sentence (i.e. a formal category) as its basic unit of analysis, but the Discourse Act, a functional category (Hengeveld and Mackenzie 2008, 3–4). Discourse Acts, defined as the “smallest identifiable units of communicative behaviour” (Kroon 1995, 65), consist of (maximally) four other interpersonal units: an Illocution (F), the Speech Participants (P₁ and P₂, representing the Speaker and the Addressee), and a Communicated Content (C), which “contains the totality of what the Speaker wishes to evoke in his/her communication with the Addressee” (Hengeveld and Mackenzie 2008, 87). This Communicated Content is made up of one or more Subacts evoking either referents (Subacts of Reference, R) or properties (Subacts of Ascription, T). Each of these units (or layers) is provided with a slot for operators and modifiers, providing additional grammatical and lexical information, respectively, about the unit in question. Some examples are given in the following, where examples (38b) and (39b) provide (simplified) representations of the sentences in (38a) and (39a), respectively:

• (38) a. Holiday accommodation understandably experiences considerable wear and tear (BYU-BNC, advert)

•   b. (A_I: [(F_I: DECL (F_I)) (P_I)_S (P_J)_A (C_I: [(T_I) (R_I) (R_J)] (C_I): understandably (C_I))] (A_I))

• (39) a. I frankly don’t care what these scientists are out to prove (BYU-BNC, religion)

•   b. (A_I: [(F_I: DECL (F_I): frankly (F_I)) (P_I)_S (P_J)_A (C_I: [(T_I) (R_I) (R_J)] (C_I))] (A_I))

In (38b), we find a declarative Discourse Act (A_I) containing a Communicated Content (C_I) which, in turn, consists of two Subacts of Reference (evoking the entities described as holiday accommodation (R_I) and considerable wear and tear (R_J)), and a Subact of Ascription (T_I) (evoking the property “experience”). The Communicated Content as a whole is modified by the attitudinal adverb understandably. In (39b), we likewise have a Communicated Content consisting of two Subacts of Reference and a Subact of Ascription (expressed as I, what these scientists are out to prove and care, respectively). In addition, the adverb frankly is analysed as a modifier of the Illocution (indicating that the Speaker is frank in presenting the Communicated Content as a declarative).

Although FDG does not require there to be a one-to-one relationship between units at different levels of analysis (e.g. Hengeveld 2011, Leufkens 2015, García Velasco 2017, Contreras-García and Velasco 2021), certain default relations can be identified. Thus, Communicated Contents at the Interpersonal Level (IL) typically correspond to the highest layer at the Representational Level (RL), the Propositional Content. At the Morphosyntactic Level (ML), they are typically realized as Clauses; they may, however, also take the form of “sentence fragments” (NPs, PPs, etc.) or single words (yes, ok, why); in FDG, such expressions are not seen as communicatively or syntactically incomplete. In addition, there is a default relation between Discourse Acts at the IL and Intonational Phrases at the Phonological Level (PL; Hengeveld and Mackenzie 2008, 432). Intonational Phrases are characterized by a combination of external and/or internal features, the most important ones being the presence of a complete intonational contour and boundary tones preceding and following the unit in question (Hengeveld and Mackenzie 2008, 432). The overall intonational contour of the Intonational Phrase is determined by the Illocution, possibly in combination with certain operators (e.g. emphasis, reinforcement, or mitigation).

In FDG, a distinction is made between abstract and lexical Illocutions. The former are conventionalized realizations of the speaker’s communicative intention, coded syntactically and/or prosodically (e.g. declarative, interrogative, imperative).^[3] In those cases where the Discourse Act is expressed as a clause, abstract Illocutions are often expressed through a combination of morphosyntactic and prosodic means (e.g. no subject-finite verb inversion and falling intonation for Declaratives, inversion and rising intonation in Yes–No questions). In some cases, however, speakers rely on one form of formal coding; thus, in (40a), the interrogative nature of the last two Discourse Acts is coded using intonation only, while the imperative nature of sit down in (40b) is coded only by morphosyntactic means (absence of subject, bare form of the verb):

• (40) a. Peace is in the interests of both. You believe me? You trust me? (BYU-BNC, fiction)

•   b. Come in John. Sit down. (BYU-BNC, interview)

When a Discourse Act takes the form of a unit smaller than the clause, speakers typically resort to prosodic means, as shown in (41), with the falling tone on in the bin in (41a) expressing a declarative Illocution, and the rising tone on tomorrow in (41a) being triggered by an interrogative Illocution:

• (41) a. Absolutely. In the bin. That’s the best place for it (BYU-BNC, conversation)

•   b. “When do you want to move in?” “I don’t know. Tomorrow?” (BYU-BNC, fiction)

Illocutions can, however, also be expressed lexically, as in the case of interjections (ouch, yuck) or fixed expressions (good morning, congratulations, thanks). These elements, which express very specific Illocutions, are analysed as lexical heads of the Illocution, as illustrated in (42) for the expression Thanks:

• (42) (A_I: [(F_I: thanks (F_I)) (P_I)_S (P_J)_A] (A_I))

Two or more Discourse Acts together may form a Move, defined as “an autonomous contribution to an ongoing interaction,” which “either is, or opens up the possibility of, a reaction” (Hengeveld and Mackenzie 2008, 50). Discourse Acts within a Move can be related in two ways. If the two Discourse Acts have equal communicative status, the relationship between them is one of equipollence; both function as independent Discourse Acts. If, on the other hand, the Speaker wants to indicate that one Discourse Act (the Nucleus) is communicatively more important than some other Discourse Act (a Subsidiary Act), the relationship between them is one of dependence. In that case, a rhetorical function representing the relationship between the two Acts is assigned to the Subsidiary Act (Hengeveld and Mackenzie 2008, 52–8). In (43), for instance, the Subsidiary Discourse Act as for London is assigned the rhetorical function of Orientation, while in (44), the clause the Press will be down too has the function of Motivation (as it provides the reason for the speaker to express the following Nuclear Discourse Act).

• (43) a. As for London, I’ve no great yearning to go back there. (BYU-BNC, misc)

•   b. (M_I: [(A_I)_Orient (A_J)] (M_I))

• (44) a. The Press will be down too, so watch out. (BYU-BNC, fiction)

•   b. (M_I: [(A_I)_Motiv (A_J)] (M_I))

Other rhetorical functions include Aside (typically targeting a particular Subact), Correction, Clarification, as well as Elaboration and Specification (Keizer 2020a, Giomi and Keizer 2020). At the ML, Subsidiary Discourse Acts are positionally mobile (Dik 1997, 383, Heine et al. 2013, 159, 176): they may precede, interrupt, or follow the Nuclear Discourse Act. This means that within the Linguistic Expression, they appear in an extra-clausal position (pre-clausal, interpolated, and post-clausal), as shown in (45) for the Orientation in example (43):

• (45) P^Pre     | P^Centre

•   As for London, I’ve no great yearning to go back there.

Finally, as separate Discourse Acts, Subsidiary Discourse Acts have their own Illocution and are realized as separate Intonational Phrases at the PL.

4.2 Conditional clauses

In their most prototypical use, if-clauses specify a condition (P, the protasis) under which the proposition expressed in the main clause (Q, the apodosis) is true; i.e. utterances with such if-clauses take the semantic form of “if P, then Q.” An example is given in (46a). In FDG, the relation between if-clause and main clause can be captured at the RL by analysing the if-clause as a modifier of the Propositional Contents expressed in the main clause (Hengeveld and Mackenzie 2021, 36–7). Given that the if-clause does not seem to allow for propositional modifiers (e.g. probably, possibly; cf. Hengeveld 1988, 236, see also Olbertz et al. 2018, 142, Keizer 2020b, 10), we analyse this modifier as an Episode.^[4] This modifier is assigned the semantic function Condition, triggering the use of if (46c). At the IL, the whole expression is represented as a single Communicated Content and, as such, a single (Independent) Discourse Act (46a). At the ML, this Discourse Act is expressed as a main Clause with an embedded subordinate Clause, taking the central position in the Linguistic Expression (46d). At the PL, the default realization will be that of a single Intonational Phrase (46e).

• (46) a. If you pass the test, you are a member. (COCA, movie)

•   b. IL: (A_I: (F_I: DECL (F_I)) (P_I)_S (P_J)_A (C_I: [ ….] (C_I)) (A_I))

•   c. RL: (p_j: –you are a member– (p_j): (ep_i: –you pass the test– (ep_i))_Cond (p_i))

•   d. ML: (Cl_i: [… (Cl_j) …] (Cl_i)) in P^Centre

•   e. PL: (ip_I)

In addition, representational (conditional) if-clauses can also be used as separate (Subsidiary) Discourse Acts, as in (47), and even as complete Moves, as illustrated in (3) (repeated here for convenience). The analysis of such instances as representational, that is, as cases of subordination, is motivated by the existence of a clear matrix clause (e.g. it’s obligatory to have something in the company report), even if it is provided by another speaker as a form of co-construction. This view differs from the one proposed by Sansiñena et al. (2015) and Gras and Sansiñena (2015), who identify such cases as dyadically dependent insubordinations.

• (47) And it worried me that these are the sort of people that would have been advised into the committee. If our officer had not sent them away with lots of facts and information. (BYU-BNC, meeting)

• (3) A: Uh uh it’s obligatory is it to have something in a company report

•   B: If you’ve got more than a hundred in the uh workforce (ICE-GB:s1b-062-138)

After this brief section on some of the interpersonal and representational features of FDG, we can now turn our attention to the focus of this article, namely the various types of non-conditional, insubordinate if-clauses and their analysis in FDG.

5.1 Independent Discourse Acts

We will start with the most independent type of insubordinate if-clauses, those that function as fully autonomous utterances, not part of, nor dependent on, another Discourse Act (cf. Stirling’s (1998) and Mato-Míguez’s (2014a,b) “isolated if-clauses”). The clearest representatives of this category are if-clauses that are completely lexicalized (formulaic), and which, in FDG, are analysed as lexical Illocutions (Section 4.1). Examples would be (if you) please, as well as its Dutch and French counterparts (Du. alstublieft/alsjeblieft; Fr. s’il vous plait, lit. “if you please”) when used, independently, as direct expressions of a request or appeal (the Dutch example in (48)). Note that in Dutch, the same phrase can also be used when giving or offering something to someone else (Eng. “here you are”; see example (49)), or when accepting an offer (Eng. “yes please”; example (50)). In addition, it can be used emphatically as an expression of disapproval/annoyance (Eng. “for heaven’s sake!”). The FDG analysis (given in (51)) would be the same for all these uses, the exact interpretation being a matter of context. Thus, all these expressions would be represented at the IL as Discourse Acts consisting of an Illocution and the Speech Participants, but no Communicated Content: as direct expressions of the Illocution, they do not involve any Subacts of Reference or Ascription. This also means that they are not represented at the RL. Since they also lack morphosyntactic structure, they are not captured at the ML either. At the PL, they are represented as separate IPs.

• (48) ja, ga mee joh. alsjeblieft. (CGN, N01055)

•   yes, go with PART please

•   ‘Yes, do come along. Please.’

• (49) A: (offering biscuits)

•    alsjeblieft (CGN, N01097)

•    ‘Here you are/Have some.’

•   B: dankjewel

•    ‘Thank you.’

• (50) A: cola?

•   B: alsjeblieft. (CGN, V80023)

•    ‘Yes, please.’

• (51) a. IL: (A_I: (F_I: alsjeblieft (F_I)) (P_I)_S (P_J)_A (A_I))

•   b. PL: (ip_I)

The examples in (48)–(50) clearly show that, ^[5]even as formulaic expressions, with conventionalized meanings, the use of these expressions, and their exact interpretation, depends on the (linguistic or situational) context. This seems to hold even more for a number of other (semi-)fixed expressions, such as If you say so (expressing doubt; Declerck and Reed 2001, 77–8, 386–7), If you must, If you insist (both expressing deprecation/disapproval; Declerck and Reed 2001, 384), or If only (functioning as an optative; Stirling 1998, 286, Declerck and Reed 2001, 383–9). It has been argued that in these cases, there is an implicit apodosis (Declerck and Reed 2001, 383–4), in which case they would not really be insubordinate. Note, however, that in the examples in (52)–(54), it would be difficult to envisage what the implicit consequence would be; even if one could add one, it is clear that this is not what the speaker really wishes to communicate. Thus in (50), the speaker merely wishes to express doubt, which would be lost on a subordinate reading (If you say so, ^# then that must be the case). Similarly, if you must in (53) serves to indicate the speaker’s disapproval, not merely the sense of inevitability expressed in the subordinate form (If you must, ^# then you have to go), while in (54), If only expresses a sense of longing and hopelessness. Therefore, although clearly less fixed than the examples in (48)–(51), we would still propose an analysis along the same lines.

• (52) “Now, all we can do is wait for his report and get on with more work.” “If you say so,” Paula said dubiously. (BYU-BNC, fiction)

• (53) “I’m sorry, Frau Fegel,” Erika said. “I have to go to the gym.” “Ah, athletics.” Frau Fegel was less than enthusiastic. “If you must.” (BYU-BNC, fiction)

• (54) Reaching the top, Ruth looked eastward; she could see beyond the forest to purple hills and faint mountains. If only, she thought; but it was no good wishing, especially when she didn’t even know what she was wishing for. (BYU-BNC, fiction)

Note that the same is true for the Dutch expression Als je het maar laat (expressing threat or warning; lit. “If you PART leave it,” Eng. “Don"t you dare!/Don’t even think about it!”), which can only be used autonomously and without apodosis:

• (55) Gem wijst naar de ijsbox die nog op straat staat. “Daar zit meer dan genoeg in om zat van te worden.” “Als je het maar laat,” waarschuwt Jolanda. (CHN)

•   ‘Gem points at the icebox that is still standing in the street. “There’s more than enough in there to get drunk”. “Don’t even think about it,” Jolanda warns him.’

Finally, as shown in the Dutch example in (56a), some formulaic expressions can also be used in combination with an abstract Illocution. In this particular case, it seems more appropriate to analyse the formulaic expression as a Communicated Content, combined with an interrogative Illocution. Thus, in (56), the lexical item alsjeblieft adds the meaning aspect of imploration to the Interrogative Illocution:

• (56) a. “Doe je het?” zei ze. “Alsjeblieft?” (CHN)

•    ‘“Will you do it?” she said. “Please?”’

•   b. (A_I: (F_I: INTER (F_I)) (P_I)_S (P_J)_A (C_I: –alsjeblieft– (C_I)) (A_I))

It is, however, not only formulaic expressions that can be used as Independent Discourse Acts. Some examples of such non-formulaic, largely compositional phrases are given in (1) (repeated here), where the if-clause is used either as an opening Move (its appropriateness depending on the discourse situation) or as (part of) a reactive Move, as in (57). Since we are dealing here either with a polite request (example (1)) or a directive (example (57)) (cf. e.g. Ford and Thompson 1986, 365, Ford 1993, 49–50, Ford 1997, 401–5), the particular form of the expression (the presence of the element if, as well as the particular prosodic patterns) may be assumed to be triggered by a mitigation operator on an imperative Discourse Act:

• (1) If you’ll just come next door. (ICE-GB:s1a-089-159)

• (57) A: How could I have had nineteen eighty-six’s accounts in nineteen eight-six I’m unclear on that <,>

•   B: If you look at page seventy-seven <,>

•   A: Yes.

•    (ICE-GB:s1b-065-158)

If-clauses functioning as Moves can also function as optatives, as in (58); see also examples (16) and (22) above; in this case, the formal features of the clause (use of the sequence if only in combination with a subjunctive) are triggered by the presence of an Optative Illocution.

• (58) “My heart has joined the Thousand, for my friend stopped running today,” he said to Blackberry, quoting a rabbit proverb. “If only it were not Bigwig,” said Blackberry. (BYU-BNC, fiction)

In other cases, such independent Discourse Acts are used as exclamations of surprise, or expressions of indignation or concern (cf. Quirk et al. 1985, 842, Declerck and Reed 2001, 386, Panther and Thornburg 2003, 139), as in examples (59), (60), and (61), respectively. Sometimes, the insubordinate if-clause is simply used in response to what has been said before, as in (62), where the utterance If it meant anything is a self-effacing comment on B’s own guitar playing, prompted by A’s first utterance and, in turn, prompting a response by A. The same can be said of the idiomatic expression If the worst comes to the worst in (63), except that here A reflects on something she/he said her/himself in the first Move.^[6]

• (59) Well, if it isn’t my little companion, my theatre-going pal. (BYU-BNC, fiction)

• (60) Karen stood up, shrugging. “Suit yourself. If you think I’m prying. Right, I’m off to bed. See you.” (BYU-BNC, fiction)

• (61) A: That’s a job in itself then …. Yeah. Mm.

•   B: If you think of the trauma on the people who have to deal with that.

•   A: (unclear) aye.#

•   B: Now you never forget those sort of things.

•    (BYU-BNC, unscripted speech)

• (62) A: I’ll just say you know that your s guitar playing is so pianistic

•   B: If it meant anything

•   A: No It is It’s so exciting

•    (ICE-GB:s1a-045-233)

• (63) A: Or you can uhm order like a meat and three veg without the meat They should let you do that

•   B: Yeah

•   A: If the worst comes to the worst

•    (ICE-GB:s1a-071-366)

As shown in (64) and (65), if-clauses functioning as Independent Discourse Acts can also constitute Moves within a Move (a digression; Hengeveld and Mackenzie 2008, 51–2). In (64), the speaker interrupts him/herself briefly to ask the judge for permission to continue, and in (65) to parenthetically comment on what he/she is doing at that moment. Note that after the digression the original Move continues.

• (64) Now <,> of course there is some com there is some uhm if I’m talking too much My Lord But say under hypnosis it is possible to get a person to have a paralysed arm under hypnosis (ICE-GB:s1b-070-079).

• (65) Well the reason is that if I just go back one slide the attenuation that we measure is proportional to this the extension <unclear-word> (ICE-GB:s2a-053-091).

Finally, if-clauses can occur as Independent Discourse Acts within a larger Move. An example can be found in (66), where the if-clause starts a new Move which also includes another Discourse Act (But he couldn’t). Although the contents of this second Discourse Act are clearly (and explicitly) related to the first, the relation is not one between a Nuclear and Subsidiary Discourse Act, but between the representational contents of the two Discourse Acts (i.e. between the two Episodes of Denis Betts being able to pick that ball up and not being able to do so). The two Discourse Acts can, therefore, be regarded as equipollent.

• (66) If only Denis Betts could have picked that ball up and got it out to Offiah But he couldn’t <,> (ICE-GB:s2a-4-377)

In (66), if only clearly marks the Discourse Act as an optative (in this case indicating a wish that has not been fulfilled). In other cases, the if-clause functions as an exclamation (example (67)) or a directive (examples (68)–(69)):

• (67) Another great conflict in the whole all of this was to you know during those decades the sixties and the seventies to discover uh just how I I should lead my life If you think it was a whole generation doing that a whole generation of y of young gay men and women And that was much more important than any work that I could do that sort of that coming to terms with my sexuality and <,> (ICE-GB:s2b-045-057)

• (68) We will start again If you’d like to engage neutral <,> That’s the way Kickstart the bike (ICE-GB:s2a-054-076)

• (69) Right now if you’d come round here/we have the Ottomans <,> (ICE-GB:s2a-059-001)

We’d like to propose that all the non-formulaic if-clauses in this section be analysed as Communicated Contents at the IL, corresponding to an Episode at the RL. The actual form of the expression (i.e. its realization as an if-clause, as well as any other formal features), however, may be triggered in different ways, e.g. by the presence of a particular Illocution in combination with an operator). Thus, if-clauses like those in (1), (57), (68), and (69) could be regarded as imperatives with a mitigating operator (at the layer of the Discourse Act), while the if-clause in (62) could be seen as being triggered by a declarative Illocution in combination with a mitigating operator. In the case of example (57), this would lead to the following representation (note that, since we are dealing with a non-formulaic expression, this if-clause, unlike the ones in (48)–(50), is represented at the RL and ML):

• (70) a.  If you look at page seventy-seven (see (55))

•   b. IL: (A_I: (mit F_I: IMP (F_I)) (P_I)_S (P_J)_A (C_I: [(T_I) (R_I) (R_J)] (C_I)) (A_I))

•   c. RL: (ep_i: –you look at page seventy-seven– (ep_i))

•   d. ML: (Cl_i) in P^Centre

•   e. PL: (ip_i)

In a similar way, the sequence if only in (58) and (66) can be assumed to be triggered by an optative Illocution, while the sequence “if it isn’t + NP” in (59) could be regarded as the result of an exclamative Illocution. Finally, examples (60) and (61), which are perhaps more emphatic than exclamative, could be analysed as declarative Discourse Acts with an emphatic operator.

5.2 Subsidiary Discourse Acts

In this section, we will look at insubordinate if-clauses that are semantically, syntactically, and prosodically non-integrated, but which are pragmatically dependent on the presence of another (previous or following) Discourse Act (cf. Eilfort’s (1987, 56–8) “peripheral conditionals”; Haegeman’s (1984) “pragmatic conditionals”). Such if-clauses will be analysed as Subsidiary Discourse Acts, relating in different ways to the Nuclear Discourse Act (functioning as comments, orientations, elaborations, specifications, motivations, etc.). These (broadly defined) discursive functions will be reflected in the rhetorical functions assigned to the Subsidiary Discourse Act. These functions, in turn, partly determine the position the if-clause takes vis-à-vis the host, i.e. which extra-clausal position it occurs in at the ML (pre-clausal, interpolated, and post-clausal; Keizer 2020a, Giomi and Keizer 2020).

Starting with if-clauses appearing in the pre-clausal position, we find that they can fulfil various functions. In example (71), for instance, the if-clause “serves to orient the Addressee to the Speaker’s communicative intentions” (Hengeveld and Mackenzie 2008, 55), indicating in which respect the entities discussed (doors) are flimsy. As such, the Subsidiary Discourse Act will be assigned the rhetorical function of Orientation, which automatically leads to placement in the pre-clausal position. Similarly, the if-clause in (72) serves to draw the hearer’s attention to “the apology at page d seven” before proceeding to comment on that apology. The if-clause in (72) is also prosodically non-integrated, as indicated by the long pause (<,>) following it (in the previous example prosodic non-integration is suggested by the use of a comma separating it from the following clause).

• (71) And if you think of a the main fabric of them, they’re quite flimsy. (BYU-BNC, spoken-unclassified)

• (72) Now My Lord if we look at the apology at page d seven <,> the BBC accept in the last sentence there is no truth in a number of allegations and apologises for them <,> (ICE-GB:s2a-064-083)

In (73) and (74), on the other hand, the pre-clausal if-clauses function as a Motivation, indicating why the speaker utters the Nuclear Discourse Act (voluntarily so in (73), reluctantly in (74)).^[7] As pointed out by Hengeveld and Mackenzie (2008, 54), however, motivating Subsidiary Discourse Acts can also follow the Nuclear Discourse Act, as shown in (75) and (76).^[8] Note that their prosodic non-integration is signalled by pauses (<,>) separating example (73) from the rest of the utterance and by the use of commas in the other examples.

• (73) If you are interested, there is still time to do something about it. (BYU-BNC, misc)

• (74) “If you must know, our Emily wasn’t an only child. She had a brother, Tom.” (BYU-BNC, fiction)

• (75) Uhm and this is actually quite a good quality tent <,> if you hadn’t noticed <,> (ICE-GB:s2a-047-024) (= (3))

• (76) “In fact, I sent Geoffrey away, if you must know!” she shrilled. (BYU-BNC, fiction)

In (77)–(78), the clause if you remember seems to function as an Appeal to the hearer, irrespective of which position it takes. In (77), this phrase additionally seems to serve as a mitigating device (“I know that you probably know this”), while in (78) it has a reinforcing function; in both cases the phrase is forward-looking. In (79), on the other hand, the if-clause is added, as an Afterthought, to remind the hearers that they were supposed to be familiar with the information provided in the preceding Discourse Act.

• (77) if you remember we looked at the index <,> (ICE-GB:s1b-057-103)

• (78) a Labour Government would mean HIGHER rates. Higher, if you remember, by TWO PER CENT. (BYU-BNC, newspaper)

• (79) That’s right. Also, Helen Burns isn’t an orphan (pause) if you remember. (BYU-BNC, classroom)

The specific interpolated position an if-clause appears in is often triggered by the fact that the clause targets a particular element within the Nuclear Discourse Act. In (78), for instance, the clause if you remember draws attention to the information that follows (by two per cent); in addition, it also neatly separates topical from focal information. In other cases, targeting if-clauses simply function as Asides, providing additional information about the preceding or following unit. In example (80), for instance, the clause if I may respectfully say so serves as a comment on (or to mitigate) the potentially offensive phrase on the plump side, while in (81), the clause if that’s the right term comments (meta-discursively) on the suitability of the lexical item advertise (cf. Dancygier’s (1998) “conversational conditionals”).

• (80) She was then a stable <,> outgoing person <,> on the plump side if I may respectfully say so <,> fun-loving <,> sociable <,> independent <,> confident <,> happy-go-lucky <,> with a <,> good marriage <,> including <,> the sexual side <,> and in general she had very good health <,> (ICE-GB:s2a-062-011)

• (81) And it was well known at that time because we advertised it if that’s the right term to the to the uh people at large that we were looking to acquire businesses (ICE-GB:s2b-065-078)

Finally, as in the case of if-clauses functioning as Independent Discourse Acts, the if-clauses discussed in this section can also be formulaic in nature. Some examples were already provided in (74) and (76) (if you must know) and (77)–(79) (if you remember). Some more examples can be found in (82)–(85):

• (82) The other difference, Chair, if I may, was the cooperatives that currently put in Shropshire, (BYU-BNC, meeting)

• (83) “Continue, Craig, if you please,” she said sternly. (BYU-BNC, fiction)

• (84) He I’d say he’s uh he finds it difficult to look at himself <,> if you see what I mean and see his own responsibility for the marriages not working out or is it anything to do with him (ICE-GB:s1a-076-044)

• (85) Very short skirt on if you don’t mind me saying (ICE-GB:s1a-040-089)

As mentioned at the beginning of this section, all the if-clauses discussed here will be analysed as Subsidiary Discourse Acts at the IL, provided with a rhetorical function specifying the relation between the Subsidiary and the Nuclear Discourse Act. This is illustrated in example (86) for the if-clause in (73). Note that the descriptive (designating) information contained in the if-clause is represented at the RL.

• (86) a.   If you are interested, there is still time to do something about it. (BYU-BNC, misc)

•   b. IL: (A_I: (F_I: DECL (F_I)) (P_I)_S (P_J)_A (C_I: [ ….] (C_I)) (A_I))_Motiv

•   c. RL: (ep_i: (e_i: –you are interested– (e_i)) (ep_i))

•   d. ML: (Cl_i) in P^Pre

•   e. PL: (ip_i)

In those cases, where the if-phrase is entirely formulaic, it will be analysed as a lexical Illocution within the Subsidiary Act, as illustrated in (87) for example (82) (compare example (51)). In examples like these, the Subsidiary Discourse Act does not contain a Communicated Content and is not represented at the RL and ML.

• (87) a IL: (A_I: (F_I: if I may (F_I)) (P_I)_S (P_J)_A (A_I))_Aside

•   b. PL: (ip_i)

5.3 Interpersonal modifiers

The if-clauses to be discussed in this section are also semantically non-integrated (non-truth-conditional); unlike the if-clauses discussed in the previous sections, however, they are syntactically and prosodically integrated; as such they are positioned between subordination and complete insubordination and can be regarded as semi-insubordinate. In example (88), for instance, the if-clause is prosodically integrated, i.e. realized as part of a larger Intonational Phrase (as can be seen from the representation given in Figure 1, produced in PRAAT (Boersma and Weenink 2019)). The same is true for the phrase als je wil (‘if you like’) in the Dutch example in (89):

• (88) Uh if I may I would like to add one final word to the families of those men who are engaged in this conflict in the Gulf <,> (ICE-GB:s2b:014-043)

Figure 1

Prosodic representation of the first part of example (88).

Evidence for the syntactic integration of such clauses comes from V2-languages, where they trigger subject–verb inversion when appearing in the clause-initial position. Compare in this respect examples (90) and (91) from Dutch. In (90), the if-clause als je het mij vraagt (‘if you ask me’) in the first position does not trigger subject–verb inversion; as such, it is regarded as syntactically non-integrated (i.e. a Subsidiary Discourse Act); in (91), on the other hand, the same phrase does trigger inversion, indicating that it is syntactically integrated, i.e. that it forms one syntactic unit with the following main clause. A similar pair of examples is provided in (92) and (93), now with the if-clause als je zin hebt (‘if you like,’ ‘if you feel like it’). In FDG, the former will be analysed as a Subsidiary Discourse Act (Section 5.2), the latter as an interpersonal modifier (Keizer 2018, 2020a; Giomi and Keizer 2020, see also Kojadinović 2022).

Due to the non-conditional nature of these if-clauses (note that in all the examples in (90)–(93) they do not specify a condition), they will be analysed as having an interpersonal function. As syntactically and prosodically integrated elements, however, they will not be analysed as separate Discourse Acts but rather as modifiers within a larger Discourse Act (Keizer 2018, 2020a, Giomi and Keizer 2020, see also Kojadinović 2022).

Analysing such clauses as interpersonal modifiers of course raises the question of which interpersonal layer they modify, i.e. which part of the Discourse Act do they comment on? Clearly, this depends on the modifier itself. The highly formulaic politeness marker if I may in (88), for instance, seems to scope over the Illocution, as it serves to mitigate the force of the declarative. The clause if you don’t mind in (94) seems to perform a similar function. The expression als je het mij vraagt in (91), on the other hand, is used to indicate the source of the information: what is expressed in the main clause is not a generally accepted truth, but merely the speaker’s opinion; as such, it functions as an evidential, modifying the Communicated Content; an English example, with the equivalent clause if you ask me, can be found in (95). The Dutch expression als je zin hebt in example (93) serves to mitigate the force of the Illocution: the declarative main clause could be perceived as face-threatening in that it may be interpreted as making a suggestion; by adding the if-clause the speaker explicitly indicates that the hearer is under no obligation to perform the action in question. The if-clause als je wil can be used in the same way. In example (89), however, it seems to add an element of disapproval, as though the speaker cannot imagine that anyone should want to perform the action in question.

• (94) I’m just going to top up my tea again if you don’t mind (ICE-GB:s1a-067-169)

• (95) Looks a bit lethargic if you ask me. (BYU-BNC, conversation)

In addition, if-clauses can have a meta-discursive function, as in examples (96)–(98). In that case, they are used as comments on the selection (appropriateness) of the property expressed in the preceding unit (examples (96) and (97)) or the following unit (example (98)).

• (96) And you know it wasn’t kind of wasn’t really stimulating me enough really after after kind of like years of of of being kind of like <,> sort of battery educated if you like You know what I mean uhm (ICE-GB:s1a-034-056)

• (97) Refined or classic French cookery – call it haute cuisine if you must – and French regional, provincial, farmhouse and peasant styles of cookery cannot arbitrarily be isolated and set apart one from the other. (BYU-BNC, misc)

• (98) Well it isn’t accountable on the basis that it merely is a group of Chief Constables who sit down together to consider the new issues and and devise if you like systems of policing what is best in terms of a common policy. (ICE-GB:s1b-033-079)

The examples provided so far clearly suggest that if-clauses functioning as modifiers tend to be quite conventionalized. It will therefore not come as a surprise that the completely lexicalized if-clauses mentioned in Section 5.2 can also have a modifying function, as in example (99) from Dutch:

As for the exact FDG analyses of these modifying if-clauses, let us consider the clause if you ask me in (95). Here, the main clause and the if-clause are both parts of the same Independent Discourse Act (A_I in (100b)). Within this Discourse Act, the if-clause functions as a modifier of the Communicated Content (C_I). As an interpersonal modifier, the if-clause is not present at the RL (reflecting its non-truth-conditional nature); the proposition expressed in the main clause, however, is represented here (in the form of a Propositional Content, p_i). At the ML, the if-clause takes a clausal position (as an embedded clause, Cl_j), reflecting its syntactic integration. Similarly, at the PL, the if-clause is part of the larger Intonational Phrase (ip _i) corresponding to the Discourse Act as a whole, indicating its prosodically integrated status.

• (100) a. Looks a bit lethargic if you ask me. (BYU-BNC, conversation)

•   b. IL: (A_I: (F_I: DECL (F_I)) (P_I)_S (P_J)_A (C_I: [ ….] (C_I): if-you-ask-me (C_I)) (A_I))

•   c. RL: (p_i: –looks a bit lethargic– (p_i))

•   d. ML: (Cl_i: [(Vp_i: –looks– (Vp_i)) (Ap_i: –a bit lethargic– (Ap_i)) (Cl_j: –if you ask me– (Cl_j))] (Cl_i))

•   e. PL: (ip_i: –looks. if you ask me– (ip_i))

The existence of an intermediate category between subordination and insubordination, with a close functional relation to the category of Subsidiary Discourse Acts (Section 5.2) and at the same time a close structural relation to subordinate if-clauses (Section 4.2), raises the question of how these interpersonal modifiers may have developed historically. While this question is clearly outside the scope of the present article, two obvious pathways suggest themselves: representational modifier if-clauses developing an interpersonal function (possibly leading to broader discourse uses; cf. Mithun’s (2008) functional extension), or Subsidiary Discourse Acts becoming integrated into the Nuclear Discourse Act (Discourse Act integration; e.g. Giomi 2020, 121, 213, 361–2).^[9] Since the data presented in this article are purely synchronic, we clearly cannot provide an answer to this question. However, the fact that if-clauses functioning as modifiers tend to be formulaic may be taken as an indication that Discourse Act integration may have led to the current situation. Given that insubordinate clauses are generally assumed to have developed out of ordinary conditionals (e.g. Evans 2007, Mithun 2008, Heine et al. 2016), i.e. representational modifiers, which subsequently gain greater pragmatic independence (i.e. Independent Discourse Act or Subsidiary Discourse Act), interpersonal modifier if-clauses would thus represent a case of insubordinate clauses being re-integrated at the IL.