Modification as a linguistic ‘relationship’: A just so problem in Functional Discourse Grammar

2.1 The functions of so

In the arguably earliest uses of so, it functions either as a deictic proform (König 2015) or a degree intensifier (OED), although König (2015) and König and Vezzosi (2022, 318) argue that the former is the source of the latter. The proform so, like its German cognate (see Wiese (2011, 995) for a description), is indexical in that it may exophorically point to extralinguistic elements in the communicative situation or the common ground, as well as endophorically to other linguistic elements. According to König (2015, 41, 48; see also Huddleston and Pullum 2002, 1535), the component of exophoric deixis could originally be coupled with the semantic dimensions of manner (e.g. Fold the paper in two, so), quality (e.g. so so for average ‘just o.k.’), and degree (e.g. The fish was so big); however, the first two of these are falling out of use, while the third is preserved, as is the non-deictic degree-intensifying meaning, e.g. It was so big that it didn’t fit into the bowl (Quirk et al. 1985, 591). Subsequently, so developed an endophoric function, allowing for multifold anaphoric reference to, for example, sentences (e.g. A: Are you happy? B: Yes, I believe so.), clauses (e.g. She was totally opposed to the idea and told the premier so), adjectives (e.g. Freedom of the press is sacrosanct and should remain so), and, with do support, individual verbs (e.g. John works in the library and Fred does so at home) or verbs together with their complements (e.g. A: Did he clean his room? B: He will do so tomorrow.) (Cornish 1986, 94, 97, Huddleston and Pullum 2002, 1529–30, König 2015, 43, Biber et al. 2021, 74).

It has been argued by König (2015, 45–6) that the later subordinator use stems from the earlier manner function, whereby so meaning ‘in this way’ or ‘in this manner’ as an anaphor in clause-initial position adopts a range of different connective marker meanings such as condition, cause, inference, concession, result, and purpose. Moreover, König (2015, 46) explains this transition in the following way: “Given the semantic versatility of the manner deictic so in English, … we could assume that what we have here is a vague meaning, still expressing the notion ‘manner,’ i.e. anaphorically relating to the kind of situation spelt out by the context.” The anaphoric manner proform referring back to the preceding clause is then reanalysed as a subordinating conjunction integrated into a now complex sentence (König 2015, 47–8), inter alia through “the semantic phenomenon of metonymic change, i.e. the incorporation of aspects of contextual meaning into the meaning of the connective” (König 2015, 47). Since the Middle English period (ME; ca. 1150–1500), so is generally considered a subordinator of result and purpose; it is less known that already in OE it also had a now obsolete condition use (Kortmann 1997, 316–7). Although the focus of this article is on the subordinating purpose and condition uses, the result function will be briefly discussed here in order to highlight the difference between the simple so and complex just so subordinator (Section 5.2).

In English, clauses of result are first and foremost introduced by the subordinator so or the variant so that, whereby the latter tends to be considered more formal (Quirk et al. 1985, 1108–10). Kortmann (1997, 86) defines the result clause as “q, so that p,” whereby “p expresses a (mostly factual and typically non-intended) result or consequence of q” (e.g. She is a good lecturer, so (that) all her courses are full). The two defining semantic criteria of result clauses are absence of intentionality or agentivity and/or event completion or factivity; see example (2). Therefore, the propositional content of the subordinate clause is presupposed and thus cannot be cancelled, as demonstrated in (3b) with respect to (3a) (Schmidtke-Bode 2009, 45–7; see also Huddleston and Pullum 2002, 733); cancelling the proposition in the subordinate clause makes the whole sentence questionable, if not non-sensical. Furthermore, Verstraete (2007, 202) argues that a strong sequentiality of the events denoted by the main clause and the result clause explains the subordinate clause’s syntactic immobility, as demonstrated in (3c)–(3d) (see also Palmer 1987, 99). In (3c), fronting would seem to construe a purpose meaning, as does clefting in (3d). Furthermore, the subordinate clause cannot be the focus of interrogation in the main clause without bringing in some aspect of intention, as demonstrated in (3e).

• (2) The dust clogged their throats, so that the women were always making ice water. (Huddleston and Pullum 2002, 1321)

• (3) a) He had to work late so that he couldn’t have any time with his son. (Huddleston and Pullum 2002, 733)

•   b) ? He had to work late so that he couldn’t have any time with his son, but in fact he did have time with his son.

•   c) #So that he couldn’t have any time with his son, he had to work late.

•   d) #It is so that he couldn’t have any time with his son that he had to work late.

•   e) #Did he have to work late so that he couldn’t have any time for his son?

So (that) as a purpose subordinator encodes a situation in which the main clause action is performed intentionally to bring about the situation in the subordinate clause (Schmidtke-Bode 2009, 20). Kortmann (1997, 86) provides the formula “in order to p, q,” whereby “p is an intended result or consequence of q that is yet to be achieved,” as in example (4). Some of the central semantic elements that constitute purpose clauses are intentionality, target-directedness, future orientation, and a hypothetical result state (Schmidtke-Bode 2009, 18–9).

• (4) He worked hard so that he might become rich. (Palmer 1987, 99)

Of all the semantic classification criteria proposed in the recent work on purpose clauses, intention is one of the defining features (and distinguishes them from the very similar result clauses; Kortmann 1997, 86, Huddleston and Pullum 2002, 733, Verstraete 2008, Schmidtke-Bode 2009, 47). This means that the purpose subordinate clause “describes what the agent of the main clause intends as the outcome of the activity described in that main clause” (Verstraete 2007, 200). Intention also plays a role in the second defining feature of purpose clauses, hypothetical result state, i.e. the activity in the subordinate clause is intended but not necessarily completed (Palmer 1987, 99, Schmidtke-Bode 2009, 178). Palmer (1987, 99) talks about potential result, and it has often been noted that this is marked by modality, such as might in (4), and the proposition of the subordinate clause can still be cancelled, as demonstrated in (5a). Syntactically, purpose clauses are adjuncts (Quirk et al. 1985, 1107–8, 1070). They can be questioned through an interrogative in the main clause as in (5b), taken as the focus of a cleft sentence as in (5c), and be preposed as in (6) (Quirk et al. 1985, 922, Palmer 1987, 99, Verstraete 2007, 201–2).

• (5) a) He worked hard so that he might become rich, but he died in poverty.

•   b) Did he work hard so that he might become rich?

•   c) It is so that he might become rich that he worked hard.

• (6) So that we might have more room for looms, the school allowed us to use the lower porch of old Ridgeway, its academic building, which was 101 feet long and roomy enough. (COHA 1958)

Historically, so as a subordinator by itself was also used to signal condition, either meaning ‘if’ or ‘as long as,’ as in example (7), but this function is said to no longer exist in PDE (Kortmann 1997, 315). As with many stages in the study of so, it is unclear when exactly the condition use disappears. According to Kortmann (1997, 315–6), this happens in Early Modern English (sixteenth and seventeenth centuries), although occasional examples can also be found as late as the nineteenth and even twentieth centuries, see (8) and (9), respectively.^[3]

• (7) It is no matter how dirty a bag it is conveyed to him in, …so the money is good. (OED 1750)

• (8) I’ll swiftly go…Nor care what land thou bear’st me to, So not again to mine. (OED 1812 LD. BYRON Childe Harold: Cantos I & II I. xiii. 14)

• (9) Yet the Black God still reigned, thought Kane dizzily, brooding back in the shadows of this dark country, bestial, blood-lusting, caring naught who lived or died, so that he drank. (Robert E. Howard “Red Shadows” 1928)

Finally, so has developed a number of discourse functions, such as introducing an explanation (explanatory so) as in (10), making a summary (11a), indicating a shift to a previously discussed topic (11b) or introducing a new sequence (11c), to name just a few (Schiffrin 1987, Bolden 2006, 2008, 2009, Buysse 2012, Denison 2020).

• (10) A: And you’re getting these things up to what sort of height?

•   B: So, we’re interested at the height of about 10 km, but they continue up to about 25–30 km.

•   (Denison 2020, 218)

• (11) a) so they are not free […] (Denison 2020, 212)

•   b) so what time is the flight tomorrow? [discussed earlier] (Denison 2020, 212)

•   c) …she obviously wants to look a bit more glamorous… er so he repaints the picture and she hangs it up very impressed… (Buysse 2012, 1773)

In sum, both historically and in PDE, so can fulfil a number of different functions, e.g. as an adverb, conjunction, or pragmatic marker.

2.2 The functions of just

Primarily, just functions as a so-called focus marker. Focus markers are employed by a speaker to draw a hearer’s attention to a certain element of a message (Quirk et al. 1985, 604). They are typically viewed as particles or adverb(ial)s (e.g. Nevalainen 1991, 4–8) and can sometimes be scalar, like degree modifiers. However, unlike degree modifiers, focus markers do not intensify the element they modify (the focus), but activate some sort of scale or set that is related to that element (Traugott 2006, 341), as demonstrated by also in (12). Unlike intensifiers, which are usually part of an adjective or adverb phrase, focus markers are more flexible (Traugott 2006, 337–8). Furthermore, focus markers do not necessarily need to be placed right next to the element they modify (König 1991, 10, Nevalainen 1991, 39–43). This is shown in the examples below, where also in one and the same position can put focus on the verb plus direct object (12a), the verb alone (12b), or solely the direct object (12c):

• (12) a) They also <feed the cats>. (‘They do not only water the plants.’)

•   b) They also <feed> the cats. (‘They do not only stroke them.’)

•   c) They also feed <the cats>. (‘They do not feed only the dogs.’)

•    (Nevalainen 1991, 42)

Moreover, scalar focus markers “not only invoke alternatives on a scale but also rank the focus on that scale, usually with other elements of the same type (qualities, events, persons, etc.)” (Traugott 2006, 342). They may signal some sort of discourse evaluation like, for example, expectation (Traugott 2006, 342) or select a referent or proposition in relation to alternative values (Ghesquière 2017, 34). This is demonstrated in (13) by just, where the implication would appear to be that being friends is a lower value on the relationship scale than being someone’s girlfriend.

• (13) So if Walter thinks you’re his girlfriend, tell him you guys are just friends . (COHA 1989)

A number of terminologies, categories, and subcategories have been suggested for focus markers, such as non-conjunctional vs conjunctional, restrictive vs additive, exclusive vs inclusive, and particularization (see Quirk et al. 1985, 604, König 1991, Nevalainen 1991, 55, Ghesquière 2017, 35 for different classifications). Nevalainen (1991, 151–4) classifies just both as a particularizer and as an exclusive focus marker (these for her being two subtypes of the so-called restrictive focus markers). Since these are the functions of just found to be used with so, only these two focus categories will be discussed in more detail.

As already briefly mentioned in Section 1, just stems from the French adverb juste, it enters English at around 1400, and in the ensuing Late Middle English period (ca. 1400–1500) it began to be used as a so-called particularizer with the meaning ‘exactly,’ ‘precisely,’ and ‘particularly’ (Nevalainen 1991, 151). Particularizers are in essence non-contrastive and non-scalar, although alternative values may be implied and the element in focus is emphasized as the paradigm example or value in the implied set (Ghesquière 2017, 35). This non-contrastive behaviour is a feature distinguishing them from exclusive focus particles (see below). In general, particularizers are considered to be non-truth-conditional (Nevalainen 1991, 58). Example (14) illustrates the particularizer just, used to identify precisely the right person to organize the games and implying that there are potentially other candidates for this position as well.

• (14) Leo will be just the one to get games organized. (Ghesquière 2017, 35)

Just developed its exclusive function after the particularizer function, in the middle or at the end of the seventeenth century (Nevalainen 1991, 58, 151–2). Only is the paragon of the exclusive focus marker category, but other exponents, next to just, include simply, barely, and merely (Nevalainen 1991, 123–62). Similar to particularizers, exclusive focus markers restrict the focused element, but in a somewhat different way in that they contrastively evoke a focus value, exclude any values higher than this value or any alternative values (Ghesquière 2017, 35; see also Nevalainen 1991, 59) and are truth-conditional (König 2017, 26). In (15), just excludes anyone else from taking the stage that evening, in this case the expected opening act.

• (15) There was no opening act, just Bruce listed at 7:30 pm. (COCA 2012)

There are two exclusive focus marker subtypes: scalar and non-scalar (or categorical) (König 1991, 95–6, Nevalainen 1991, 59–64). The scalar exclusive locates the focused referent on a scale, as in (16), where being a plumber is located toward the bottom of a social scale. Similarly, in (13) above, the status of being just friends marks the bottom of an implied relationship scale which potentially includes friends with privileges, a relationship, or marriage. In the non-scalar or categorical use, a focus value is singled out and all other alternatives are excluded (Ghesquière 2017, 35). This has already been demonstrated in (15), where expectations are negated.

• (16) He is just a plumber. (König 1991, 96)

König (1991, 116, 118) makes another distinction within the exclusive class between the ‘only’ use in (17a) and the ‘simply/emphasis’ use of just in (17b) (see also Nevalainen 1991, 153). The ‘only’ reading is only possible when “the value denoted by the context can be identified as a medium or low value on some scale,” i.e. the scalar exclusive; in contrast, the ‘simply’ use has “an emphatic effect” and is only possible when “the focus of just denotes an extreme value on some scale” (König 1991, 118). Both of these exclusive uses as well as the particularizer function of just play a role in the analysis below.^[4]

• (17) a) I just want two apples.

•   b) That’s just marvellous.

•    (König 1991, 116)

In sum, previous literature has classified the adverb just as a focus particle with a number of functions: as a particularizer meaning ‘exactly,’ and later as an exclusive focus particle meaning, first, ‘only’ and then ‘simply.’

5.1 Just so of manner

The manner use of just so is the most frequent form found at the beginning of the nineteenth century and, therefore, is considered to be the earliest function of this combination (together with the intensifier use not treated here). Just so of manner is often (but not necessarily) positioned clause-initially and seems to mainly serve the function of anaphorically referring back to portions of discourse of varying magnitude as exemplified in (19a) and (19b).

• (19) a) Mr. B. perceived this, he, with his characteristic calmness and resolution, gave up his hold of his son, and, motioning to him to save himself, resigned himself to his fate. His son reached the shore, but was too much overwhelmed by his loss to leave it. […] “The body was found, and on the countenance was the sweetest smile; and Bradford said, ‘Just so he smiled, upon me when he let go and pushed me away from him.’” (COHA 1856)

•   b) The original fountains may be remote and unknown; but the river laves our fields, and passes by to diffuse its treasures among other regions; and even if its waters are lost to our sight by evaporation, they descend again in showers to embellish and fructify the earth in a thousand forms. Just so it is with intellectual treasures. Very few persons now read the works of Aristotle, and not many those of Bacon: but the wisdom which they first taught, or perhaps collected, is now spread far and wide by numerous modes of diffusion, and is incorporated into the minds of thousands who know nothing of its origin… (COHA 1824)

In (19a), so links the sweetest smile seen on Mr. B.’s dead face to that of his final smile, meaning he smiled in the same manner. In (19b), so metaphorically links the picture of the distribution of water over a region to the distribution of knowledge among people, meaning either in ‘the same way’ or ‘in the same manner.’

In connection with this manner use of so, just functions as a particularizer meaning ‘exactly’ or ‘precisely’ (Section 2.2). Particularizers of this kind express what Dik et al. (1981, 66) have called restricting focus within Functional Grammar (FG), FDG’s predecessor theory, and work by either identifying or specifying the element in focus, particularly in terms of its validity (Nevalainen 1991, 6, 58; see also König 1991, 117).^[11] So in its meaning ‘like this/that, (in) this/that way’ (Giomi 2020, 140, 178) has been formalized in FDG as a unit of the semantic category Manner with an absent head (m_i) on the RL, corresponding to a Referential Subact (R_I) with an identifiability (+id) and specificity (+s) operator (i.e. a referent identifiable for both the speaker and the hearer; Hengeveld and Mackenzie 2008, 122) on the IL and inserted as a proform on the ML:

• (20) IL: (+id +s R_I)

•   RL: (m_i)

•   ML: so    (adapted from Hengeveld and Mackenzie 2008, 265)

Furthermore, just has been classified as a grammatical particle by Hengeveld and Mackenzie (2008, 401, Table 8) or as a grammatical(ized) interpersonal adverb by Keizer (2018, 75), and it is analysed as a (lexical) interpersonal grading operator by Giomi (2020, 302, 328, 2021, 217). Combining different aspects of these earlier proposals, we represent the relevant levels and layers of the underlined segment in example (21) as in (22):

• (21) (Context: how to stow an umbrella) Then, with finality, roll the strap the right way, and snap the snapjust so. (COHA 2002)

• (22) IL: (C_I: [(T_I) (R_I) (just +id +s R_J)] (C_I))

•   RL: (e_i: (f^c _i: [(f_j: snap (f_j): (m _i ) (f_j)) (x_i)_A (x_j: (f_k: snap (f_k)) (x_j))_U] (f^c _i)) (e_i))

•   ML: (Cl_i: [(Vw_i: snap (Vw_i)) (Np_i: [(Gw_i: the (Gw_i)) (Nw_i: snap (Nw_i))] (Np_i)) (Advp_i: [(Gw_j: just (Gw_j)) (Gw_k: so (Gw_k))] (Advp_i))] (Cl_i))

Essentially, this is a split analysis of just so between IL and RL, with just as a lexical operator at IL applying to an identifiable and specific Referential Subact (R_J) – arguably as a sort of exactness marker in the sense of Hengeveld and Keizer (2011) – which in turn corresponds to a Manner modifier with an absent head (m_i) at RL.^[12] Applying Hengeveld’s (2017, 31) distinctions and criteria (based on Keizer 2007), just has been classified as a lexical operator because it apparently can take focalization stress (e.g. snap the snap JUST so) but does not seem to be modifiable. Although examples with adverb modifiers can be found in the corpus, it is more likely that these modify just so together, i.e. the whole State-of-Affairs, instead of just by itself, as exemplified in (23).

• (23) A: “…Dear me, how confused everything is to-day!”

•   B: “It is always just so, Sundays,” said John…

•   (COHA 1843)

Furthermore, the absent head in (22) modifies the predicate (f_j) of the Configurational Property (f^c _i) (see also Hengeveld and Mackenzie 2008, 266) and is expressed by insertion of a proform so at ML. However, this analysis suggested by the existing literature calls for some refinement and, therefore, what follows is our preferred analysis.

At IL, it is not entirely clear that what is involved is indeed a Referential Subact or whether a second Ascriptive Subact (T_J) could possibly be involved instead. In other words, it seems difficult to conceive of the way in which something is (supposed to be) done as evoking an entity instead of a property. In fact, Hengeveld and Mackenzie (2008, 121) themselves state that “not all anaphora involves Referential Subacts,” a point already made more forcefully in Mackenzie’s (1997) study of English such within the earlier FG framework: “The antecedent of anaphoric such would appear never to be an entity, but rather a property of one or more of the entities involved in the textual world” (Mackenzie 1997, 92). For such, this pertains to antecedent nouns in predicate position (e.g. He was a genius but not recognized as such) as well as adjectives. Moreover, the author interestingly notes that adjectival properties can also be anaphorically referred to by so and that the latter expression is the only option for establishing an anaphoric relation to verbal properties, as in Do not accept anything as true unless you know it to be such/so in comparison with Bill was sleeping and so/*such was Fred (Mackenzie 1997, 92). From an FDG perspective, Ascriptive Subacts are the IL counterparts of Properties and thus the proper locus for the lexical operator introduced in (22) above to apply in the following way:

• (24) IL: (C_I: [(T_I) (R_I) (just T_J)] (C_I))

Representing Manner as an Ascriptive Subact rather than a Referential Subact in (24) means that the +id +s specifications do not apply, since these are not operators of the ascription layer. As Hengeveld and Mackenzie (2008, 36) equate identifiability with definiteness, this seems to match up with “the ‘indefiniteness’ of so (in its anaphoric use)” (Cornish 1986, 96, referring to Cushing 1972, 186–7). But the question now is what exactly is evoked by this Ascriptive Subact? In this regard, it is helpful to differentiate between explicit and implicit anaphoricity. Explicit anaphors work exclusively within the Grammatical Component by introducing a dedicated expression which is co-referential with an antecedent predicate expression that has been mentioned earlier (see also Mackenzie 1997, 92). This can be formalized in FDG by co-indexation of the relevant variables at the RL. For implicit anaphoricity, which seems especially relevant for just so, “there is no predicate or predicate complex which could function as an antecedent” (Mackenzie 1997, 94). Here “the grammar clearly interacts with elements of the textual world that is gradually being constructed,” for the formalization of which “a representation of that textual world, and feedback to the grammar, is an essential requirement” (Mackenzie 1997, 103). This gradual building-up of a textual (or discoursal) world and its ongoing feedback interaction with the grammar is provided for in FDG by the Contextual Component (a separate component interacting with the Grammatical Component), which appears to be crucial in establishing implicit anaphoric links via inferences over longer stretches of discourse, as well as exophoric links directly anchored in the discourse situation (see also Mackenzie 1997, 90, 93–6). This is underscored in the following conclusion drawn by Cornish (1986, 31): “The basic function of anaphora is a discourse one: namely, that of referring to a discourse referent already present within the speaker’s discourse model, or of actually creating one, via an inference on the addressee’s part, on the basis of the discourse context.” Thus, it could be argued that what the Ascriptive Subact (T_J) in (24) evokes is directly fed by such inferences from the Contextual Component.^[13]

At RL, there are also the alternatives for the absent Manner head to modify the whole State-of-Affairs or Configurational Property, as in (25a) and (25b), respectively. Although these different attachment points would be expected to show different semantic behaviours of the Manner modifier (Hengeveld and Mackenzie 2009, 208–9), this does not seem to seriously impinge on the rest of the analysis that we propose.

• (25) a) RL: (e_i: (f^c _i: [(f_j: snap (f_j)) (x_i)_A (x_j: (f_k: snap (f_k)) (x_j))_U] (f^c _i)) (e_i): (m _i ) (e_i))

•   b) RL: (e_i: (f^c _i: [(f_j: snap (f_j)) (x_i)_A (x_j: (f_k: snap (f_k)) (x_j))_U] (f^c _i): (m _i ) (f^c _i)) (e_i))

At ML, so being a proform triggered by an absent head at RL is analysed as a Grammatical Word, similarly to just being triggered by a lexical operator at IL. Both are then part of an adverb phrase as in (22).

5.2 Just so as subordinator

Just so as a subordinator has not been discussed in detail in the literature, being only briefly mentioned in Quirk et al. (1985, 1089) and Kortmann (1997, 315–6) (but see now Kaltenböck and ten Wolde 2020a,b, 2022). Although not as frequent as the older subordinator so (that), the different subordinating functions of just so raise a number of issues about the development of these subordinators and the way they interact with their modifiers. Just so develops two separate subordinating functions: purpose, which will be discussed first, and condition, which will be discussed second.

5.3 Pragmatic functions of just so

In the mid-twentieth century, some just so purpose subordinate clauses developed a purely interpersonal use, as demonstrated in (44)–(46). In these examples, the subordinate clause no longer denotes the purpose of the action in the main clause, but now provides the motivation for a request (44), extralinguistic action (45), query (46a), or statement (46b). Essentially, (44) is a subordinate clause linked to an ellipted main clause, saying [I am requesting that you tell me the rules] just so I’m sure I’ve got them straight, and something similar applies to (46a): Hey, darling. [I am asking you this] Just so I can finish my paperwork early, what are your people gonna steal today? This function also includes semi-formulaic phrases like just so we understand each other, just so we’re/you’re/I’m clear, just so we are all on the same page, and just so I/you know. Their extra-clausal status is clearest in those cases where the illocutionary force of the main clause is different from the one in the subordinate clause, as in (46a), where the main clause is an interrogative and the subordinate purpose clause a declarative. Furthermore, whereas the purpose subordinate clause usually occurs in final position, the interpersonal uses often occur in initial position.

• (44) And he shrugged. “Now the rules, Mrs. Blodgett – just so I’m sure I’ve got them straight.” (COHA 1967)

• (45) Dakin drew the dark heavy door-blind and snapped off the lights at the front of the store. “Just so we won’t be interrupted,” he explained. “Now come over here to the table, Alvin, and sit down so we can have a nice, friendly little talk.” (COHA 1945)

• (46) a) Hey, darling. Just so I can finish my paperwork early, what are your people gonna steal today? (COCA 2003)

•   b) His book Panther Dream follows the hunting-related adventures of a boy named Lokuli, who tries to save his village from starvation. (Just so you Don’t worry: He does.) (COHA 1991)

Such a use of subordinate clauses is what Sweetser (1990, 76–82) calls the speech act conjunction. Sweetser (1990, 77) distinguishes between conjunctions functioning in the content domain, whereby the subordinate clause gives the purpose of the action denoted in the main clause, as was the case for our purpose subordinators analysed in Section 5.2.1, and the speech act function, in which the subordinate clause modifies the speech act evoked by the main clause, in the case of just so this being the purpose/motivation for uttering the main clause.^[26] The listener/reader must assume the Gricean principle of relation to understand how the two clauses relate. For FDG, the content domain use of just so is a semantic function (on the RL) and the speech act function a pragmatic one (on the IL).

Here, as was also the case with condition in Section 5.2.2, purpose just so becomes an inseparable unit, and together with its clause fulfils an interpersonal function (unlike condition, which remains representational). Just is no longer a focus marker; none of the uses attributed to it in the literature would seem to fit in these instances, particularly not the ‘only’ or ‘simply’ just found together with the representational purpose clauses discussed in Section 5.2.1. Furthermore, deleting just would make the respective clauses awkward at best, as demonstrated in (47b) and (48b) ((47a) and (48a) being repeated from (44) and (45), respectively). Although there is a strong link to the original purpose semantics, these clauses perform pragmatic functions such as explaining or motivating particular actions or discourse contributions. Furthermore, such clauses are not completely fixed; there can be variation, particularly, in pronouns (e.g. just so I’m sure I’ve got them/it straight or just so we/you/I understand (each other)).

• (47) a) And he shrugged. “Now the rules, Mrs. Blodgett – just so I’m sure I’ve got them straight.” (COHA 1967)

•   b) “Now the rules, Mrs. Blodgett – so I’m sure I’ve got them straight.”

• (48) a) Dakin drew the dark heavy door-blind and snapped off the lights at the front of the store. “Just so we won’t be interrupted,” he explained. “Now come over here to the table, Alvin, and sit down so we can have a nice, friendly little talk.” (COHA 1945)

•   b) ? “So we won’t be interrupted,” he explained. “Now come over here to the table, Alvin, and sit down so we can have a nice, friendly little talk.”

We thus propose to analyse examples like (49) as in (50):

• (49) But, just so we understand each other, slick, we’re talking about a bag job, right? (COCA 2005)

• (50) IL: (M_I: [(A_I: (F_I: DECL (F_I)) (A_I))_Motiv (A_J: (F_J: INTER (F_J)) (A_J))] (M_I))_Contr

According to this analysis, just so is introduced at the IL in one go by the rhetorical function of Motivation for a Discourse Act (A_I) which is dependent on a nuclear Discourse Act (A_J) – note especially the different Illocutions (Declarative and Interrogative, respectively) involved, which are a clear sign for the respective Discourse Act status in FDG. From our perspective, this suggests the following diachronic change: (just ep₁)_Purp > (A₁)_Motiv, in which scope increases through pragmaticalization in that the combination of a lexical operator and a semantic function at RL becomes an interpersonal function at the Discourse Act layer of IL. However, Giomi (2020, 119, 360) also describes English so as pragmaticalizing into a Motivation discourse marker in and of itself via (secondary) grammaticalization, presumably from an earlier Consequence marker use (see also Section 5.2.1) pertaining to Episodes: (ep₁)_Cons > (A₁)_Motiv (Giomi 2020, 149, 164), see examples in (51).

• (51) a) The bridge has fallen down, so we will have to cross the river in a boat.

•   b) There will be trick questions in the exam, so watch out. (Hengeveld and Mackenzie 2008, 54)

This raises the question in how far the two analyses are compatible with or contradict each other. It seems obvious, though, that motivational so and just so do not appear in all the same environments and do not have exactly the same interpersonal import, the latter (i.e. with just) sounding somewhat more restrictive than the former. Moreover, as an only tangentially relevant aside, initial but in (49) is analysed as introducing a new, contrasting Move in the interlocution in (50), following Giomi (2020, 329–30).^[27]

Finally, the most frequent of these forms have developed into pragmatic markers, such as just so you know (Kaltenböck and ten Wolde 2022). There are a number of definitions and terminological variations surrounding discourse markers and pragmatic markers, which we cannot go into here, but instead we adopt Heine’s (2013, 1206–7) classification of discourse markers (in which he includes pragmatic markers as a subcategory). For him, discourse markers are defined as reflecting the following criteria: they (i) ascribe procedural meaning and are (ii) non-restrictive, (iii) non-compositional, (iv) syntactically independent as well as (v) prosodically set off from the rest of the utterance (Heine 2013, 1209). Naturally, not all pragmatic markers have to conform to all of these characteristics; however, in the cases under scrutiny here, all criteria would appear to apply.

The phrases in question, for example just so you know, are no longer compositional, a fact that can be most plainly seen in examples like (52), where the speaker wants to divulge some information that he/she expects the hearer does actually not want to know. Instead, they have a number of different discourse functions such as signalling a topic shift (53), or they are employed for expressing emphasis (54) (criteria i and ii). The phrases are relatively fixed (criterion iii) and positionally mobile (criterion iv). For example, in a sample of 1,050 tokens for just so you know from COCA, 81.4% prefer the left periphery, 1.2% medial position, and 14.7% the right periphery (2.7% are unclear). Just so you know is clearly not part of the main clause and could be read as a prosodically separate unit (criterion v), which is supported by the orthographic evidence that it is set off by commas in (52) and (53) (for more discussion, see Kaltenböck and ten Wolde 2022).

• (52) Look, I can’t believe I have to say this, but just so you know, there’s no way the two of you work.

•   Why would you say that? You don’t know that.

•   (COCA 2019)

• (53) KOTB: You missed the Cutest Baby Contest and we’re so bummed because – oh! Come on.

•   STOCKMAN: That’s Juice-Juice. […]

•   STOCKMAN: So that’s why I call her Juice-Juice. That’s my baby.

•   KOTB: Just so you know, Shawn’s on the road all the time. I didn’t realize that you guys were…

•   (COCA 2011)

• (54) HODA-KOTB# All right. So if you want one of those, oh, my god, what a cute video pieces we have that for you. There is a baby elephant –

•   KATHIE-LEE-GIFFORD# That’s what I feel like today.

•   HODA-KOTB# – in a kiddie pool. This is in Fort Worth, Texas. Her name is Belle. She was born on July 7th. She weighs three hundred pounds.

•   KATHIE-LEE-GIFFORD# Oh, my gosh. Look at that.

•   HODA-KOTB# Oh, my gosh. Okay. Just so you know that is the cutest three-hundred-pound baby.

•   KATHIE-LEE-GIFFORD# Oh, my gosh. (COCA 2013)

The pragmatic markers do not allow for pronoun variation, which can still be found in the earlier forms of the pre-pragmatic marker stage (see (55)), and they exhibit greater syntactic flexibility. In fact, the earlier constructions are often integrated into the clause structure, the speaker usually elaborating on what he or she actually wishes to know, as in example (55b).

• (55) a) Jake, I think we need a business plan. Just so we know what we have and where we’re going. (COCA 1993)

•   b) Just so I know, would you be open to taking his class? (COCA 2008)

The pragmatic marker just so you know as in (56), on the other hand, lacks the above features of compositionality, and can accordingly be formalized as in (57):

• (56) Just so you know, what you’re looking at here is the pith. (COCA 1994)

• (57) IL: (M_I: [(A_I: just so you know (A_I))_Orient (A_J: –what you’re looking at here is the pith– (A_J))] (M_I))

This IL representation analyses the fixed expression just so you know as heading a Discourse Act (A_I) of its own, with the rhetorical function Orientation in relation to a following nuclear Discourse Act (A_J).