Narrative before in English

1.1. Statement of Aim and Method

This paper takes up the descriptive challenge posed by narrative before-clauses in English: how should this relatively rare but fully productive construction be described? Can it be accounted for adequately and at the same time in a manner consistent with the description of the other uses of before? I approach this topic on the basis of several thousand examples (for details, see section 1.4), a collection that serves as an empirical platform for my observations of patterns and tendencies, as a checklist to ensure proper representation of usage and as a source of illustrative examples for my exposition. I aim at an identification and characterization of the properties of narrative before and its delimitation from other uses rather than an actual corpus analysis with collocation patterns, statistics, and frequencies, which would presuppose the conclusions arrived at in this paper. My method is largely explorative and interpretative in a traditional, eclectic manner in the spirit of Otto Jespersen. I freely employ relevant concepts from formal, functional, and cognitive linguistics (e.g., “main clause phenomena” or “root transformations” from generative linguistics and “assertive” in Langacker’s (1991) and Cristofaro’s (2003) cognitive-typological sense) in my attempt to arrive at a better understanding of the complex syntax, meaning, and stylistics of narrative before-clauses. My aim is to make up for the lack of attention to this construction in comprehensive grammars of English (such as Quirk et al. 1985; my own Bache & Davidsen-Nielsen 1997, and Bache 2000; Huddleston & Pullum 2002), and therefore I employ several of their terms and diagnostic measures (e.g., “comparative governor” and “clefting”) in order for this paper to qualify as a supplement to their work.

1.2. What Is a Narrative before-Clause?

To give the reader a real sense of what a narrative clause is, I will go back a few years to a time when I myself was completely unaware of the existence of such clauses in connection with the temporal linkers before, when, and until. I was engaged in writing a text and exercise book on sentence analysis for first-year university students of English (Bache 2014). I had reached the chapter on subordinate clauses and had offered the example found in (1).

In this example, I argued, when is a subordinating conjunction, not an adverbial (as students often suggest). It initiates an adverbial subordinate clause (when she looked up) and links it to the main clause for temporal specification and contextualization of the situation expressed by the main clause (‘her immediately seeing him’). Its role within the subordinate clause is not to convey temporal specification of ‘her looking up,’ as an adverbial in the subordinate clause might, but to relate this content to the main clause as a temporal adverbial (Bache 2014:99). To clinch this fairly traditional (but theoretically of course not entirely uncontentious) point, I decided to offer another example (2) containing a when-clause—after all, this was meant to be a pedagogical exposition.

I started analyzing this sentence in exactly the same way as I did (1), defining when as a subordinating conjunction and the clause it initiated as a temporal adverbial clause qualifying the main clause, when it dawned on me that something completely different was going on here (as, indeed, in the current sentence!) and that what I had written was embarrassingly wrong. Unlike (1), example (2) was not so much about the main clause situation (‘Jenny reading the newspaper’) but about the situation expressed by the when-clause (‘the telephone suddenly ringing’), and this situation happened while Jenny was reading the newspaper. Somehow the balance between the main clause and the when-clause in terms of their relative communicative impact was very different from what I had expected and from what we saw in (1): the main clause in (2) offered contextual information in support of the situation expressed by the subordinate clause, or rather by what looked like a subordinate clause. Judging from its communicative impact, the when-clause in fact behaved more like a superordinate clause.

To cut a long story short: I dropped the second example in my book—it illustrated too complex a phenomenon for the level I was aiming at—but decided to follow up on examples like (2) in a new separate project on what I came to appreciate as the “narrative” use of when-clauses (Bache 2015, 2016)—“narrative” in the sense originally defined by Labov and Waletzky (1967) and Labov (1972:359-362) and later developed in connection with when by Couper-Kuhlen (1988, 1989) and Declerck (1997): a clause is narrative if it adds a new event to the progressive sequence of main events making up a storyline. A sequence of narrative clauses iconically matches the sequence of events depicted. The term narrative hence applies unproblematically to many main clauses while many subordinate clauses are non-narrative because the events they express simply support, contextualize, or elaborate on the event expressed by the main clause. In (2), however, the when-clause is used narratively, despite the fact that it is not a main clause, in that it “push[es] forward the action” (Declerck 1997:213, 2006:731-732), i.e., it brings a new event to the foreground sequence of events making up the storyline. By comparison, the when-clause in (1) is used as a temporally specifying adjunct qualifying the event expressed by the main clause. There is in other words a temporal-narrative contrast in the use of when-clauses.

When is not the only linker in English with a recognized temporal-narrative contrast, as also acknowledged by Declerck (2006). This is illustrated in (3)–(6), which are similar to the ones offered in Bache (2015:11). ¹

Although when, before, and until differ somewhat both syntactically and in semantic detail, they are all three basically temporal and relational in meaning: X when Y; X before Y; and X until Y. In all three relations, the specific linker chosen plus Y may be expressed by a sentence-initial or sentence-final construction. In (1), (3), and (5), the combination of the linker and Y is expressed by a clause in sentence-final position, while in (7)–(9) it is expressed by the corresponding sentence-initial option.

But the three linkers share the restriction of narrative usage to clauses in sentence-final position: a narratively linked clause takes its onset from a preceding main clause, and the combination always displays iconic sequencing in that the linear order of clauses reflects the order in which the situations expressed by the two clauses take place or begin. Thus in (4) the main-clause situation (‘them hearing her explanation’) happens before the before-clause situation (‘Jack bursting out crying’), and in (6) ‘her so enjoying life’ takes place before ‘her suddenly realizing that all her efforts had been in vain.’ Temporal clauses are not necessarily iconic but may or may not be iconic depending on position and, in the case of when, tense choice (Bache 2016:276).

1.3. Why a Separate Study of Narrative before?

This paper takes a closer look at the narrative use of before on the basis of my findings in connection with narrative when-constructions (Bache 2015, 2016). In addition to the stylistic effect, which is very similar for narrative when-clauses and narrative before-clauses, there are certain formal and pragmatic parallels (to be specified below). But because of a comparative element inherent in before (= ‘at an earlier time than’), which is absent in when, and because of its different syntactic properties, a different descriptive approach is required for before. More specifically I treat narrative when as a special metaphorical extension of the use of when as a sentential relative, i.e., as a relative pronoun that takes all of the preceding main clause as its antecedent (for discussion and examples of sentential relatives, see Bache & Davidsen-Nielsen 1997:274-275). A narrative when-construction is metaphorical in the sense proposed by Halliday (1994:ix) (cf., also Halliday & Matthiessen 1999:7, 227-296, 2004:7, 592-593, 613-658) in that when takes a non-temporal event expression as its antecedent and treats it as if it were a temporal expression (Bache 2016:287-290). Unlike when, however, before is never used as a relative pronoun, and the concept of metaphor plays no role in narrative before-constructions. In sections 2.1 and 2.2, I will elaborate on the comparative meaning of before and its different syntax.

1.4. The Empirical Platform of this Study

The material on which this study is based consists in part of 4000 occurrences of before made up by four subsets of 1000 randomly selected examples from each of the following four electronic corpora available at www.corpus.byu.edu:

NOW (News On the Web) 1000 hits

COCA (Corpus of Contemporary American English) 1000 hits

TIME (Time Magazine Corpus, limited to 2000-2016) 1000 hits

BNC (British National Corpus) 1000 hits

These corpora include a variety of genres and both spoken and written language. One more electronic corpus was searched:

COLLINS (Collins Wordbanks Online) 2000 hits

The 2000 examples from this corpus derive from five samples: a general unspecified one (500 hits), US fiction (500 hits), British fiction (500 hits), US spoken (250 hits), and British spoken (250 hits). ² To get as precise a picture of before and its uses as possible, and since I did not know in advance exactly what would be relevant, I did not make any a priori restrictions on these hits: all sorts of construction with before were harvested. In each corpus I simply asked for a random set of x hits to be generated. The broad scope of the material helped me formulate the general presentation of the syntax and meaning of before in sections 2.1 and 2.2.

As a supplement to the electronic corpora I screened ten Kindle novels for all occurrences of before and got 1301 additional hits. ³ No principle was involved in the selection of the novels apart from an attempt to get a certain spread of recent types of narrative styles and a wish to see the use of before in more coherent contexts. The reason for selecting Kindle editions of these works rather than normal book form was that Kindle provides an extremely convenient search option for individual words and phrases.

In addition to the total of 7301 examples in my basic collection I have made many searches for specific combinations and illustrative examples in Google (e.g., before suddenly, before finally, before gradually, hardly + before, scarcely + before, barely + before, before in came, before out rushed, before down came, etc.). I also refer to and discuss examples provided by fellow grammarians and by myself. I offer source information after each example from my collection. When an example is used several times I only supply the source the first time round. As already mentioned, the purpose of working with this corpus of examples is simply to ensure comprehensiveness and authenticity, not to conduct a formal corpus investigation.

1.5. Structure of Exposition

I begin my presentation of before with a general characterization of its syntactic and semantic properties as a temporal linker in section 2. The purpose of this section is to provide proper contextualization of the narrative use of before, more specifically the background against which this use stands out as rather special. Section 3 offers a discussion of both formal and pragmatic characteristics of narrative before in some deliberately clear-cut cases in an attempt to capture the essence of narrativity in connection with this linker and to identify the mechanisms at work. In section 4, I argue that there is a gray area between clearly temporal and clearly narrative uses of before, and that we therefore need a finer-tuned categorization of before-clauses. Finally in section 5, I arrive at my conclusion.

2.1. Syntax

In the basic relation “X before Y” with “before Y” in constructional postposition, Y may be realized by a finite clause, a non-finite clause, or a nominal, as in (10)–(12).

While traditionally a distinction is maintained between before as a conjunction (as in 10 and 11) and before as a preposition (as in 12), ⁴ Huddleston and Pullum (2002:1011-1014) provide interesting arguments for always treating before as a preposition (and hence a constructional head), even in examples like (10) and (11). The implication of this approach is that prepositional heads may take not only nominals and non-finite clauses but sometimes also finite clauses as their complement. Subordinating when is also considered a preposition taking both finite and non-finite clauses but unlike before it cannot take nominal complements. I am not going to take a firm stand in this debate (as it does not affect the central argument of this paper) but will continue to speak in traditional terms of (temporal and narrative) before-“clauses” even if, strictly speaking, they may qualify as prepositional phrases with prepositional heads and clausal complements.

When temporal before takes a nominal complement, its referent is always construed as a temporally specifiable event (as in 12). ⁵ Thus before often congruently takes a regularly recurring event or time expression as its complement, as in (13).

We do encounter nominals that are not usually considered event expressions serving as the complement of before, exemplified in (14).

But penicillin must here be understood metonymically as representing some event, such as the discovery of penicillin (i.e., an object or “thing” expression used as an event expression). Particularly frequent collocations are BC (‘before Christ,’ i.e., ‘before the birth of Christ’) and before tax (i.e., ‘before tax is added’).

Significantly, before cannot be used as a relative pronoun like when (‘at which time’), as in Jack postponed the meeting to next Friday, when the test results will be ready (Bache 2016:285). This feature provides a syntactic frame for the narrative use of when (Bache 2016:291) but cannot play a part in the description of the narrative use of before.

Before may be used without an explicit Y, illustrated in (15) and (16).

In traditional grammar, before is here analyzed as an adverb.

Before may be premodified—largely irrespective of the kind of complement it has—by an expression further specifying or quantifying the anteriority expressed, as in (17) and (18).

In such constructions the premodifier + before (+ complement) sequence constitutes an adjunct of time.

2.2. Meaning

Before basically expresses anteriority and is used for the temporal ordering of “actualities” (situations, occasions, times, things, facts) such that one actuality (X) is anterior to another (Y). ⁶ Before thus indicates relative time. For example, in sentences containing a before-clause at the primary clausal level (as in He arrived before the ceremony began) the situation X expressed by the main clause (‘him arriving . . .’) is anterior to the situation Y expressed by the before-clause (‘. . . the ceremony beginning’).

As noted in section 1.3, there is something comparative about before, its meaning being close to the comparative expression earlier than, and it is in fact descriptively possible to approach before as a “comparative governor” (as suggested in passing by Huddleston & Pullum 2002:1104). I would like to explore this possibility and begin by identifying the standard components of comparative adjectival constructions like those in (19) and (20) (the following fairly traditional approach to comparison takes its point of departure in Bache & Davidsen-Nielsen 1997:473-475 and Bache 2000:244-245).

In such sentences there are four components relevant to the discussion of comparison: a gradable property, a comparative marker, a comparative basis, and a comparative focus. In (19) and (20) the adjectives (tall, beautiful) express the gradable properties involved, -er and more are comparative markers indicating a higher degree or rank (the suffix -er being a morphological marker and more a syntactic one), and than Arthur and than Greece express the comparative basis, i.e., the standard against which the comparison is made. In the two sentences, Nithe and Calvi are in the comparative focus, i.e., the ones being described in terms of the gradable property. In the case of morphological comparison (taller), the gradable property and the comparative marker are fused into one word.

Similarly, one way to describe (21) is to say that before represents a synthetic fusion of a gradable property and a comparative element (resulting in a meaning corresponding to ‘at an earlier time than’).

In (21), the ceremony began expresses the comparative basis, and He arrived is in the comparative focus (being sequentially positioned relative to ‘the ceremony beginning’). More specifically, the comparative basis is the time of ‘the ceremony beginning,’ and the comparative focus is on the time of ‘him arriving.’

The important point, both in connection with adjectival comparison and with before-constructions, is to note the asymmetry between the two actualities X and Y involved in the comparison: one serves as the comparative basis (Y) while the other is in the comparative focus (X), i.e., it is characterized in terms of the gradable property. Thus (19) and (20) are descriptions of Nithe and Calvi , not of Arthur and Greece, and (21) is a temporal specification of He arrived, not of the ceremony began. More technically, the main clause situation in (21) is asserted (in conjunction with the before-clause situation) while the before-clause situation is presupposed. In this respect, regular temporal before-clauses are just like regular temporal when-clauses (Bache 2015, 2016). The main clause carries the main message, and the subordinate clause has a supportive role. It is a truth condition of (21) that he arrived before the scheduled beginning of the ceremony, but not that the ceremony actually began: it could have been canceled at the last minute. As Huddleston and Pullum (2002:1007) note: “The presupposition triggered by before—that the event expressed in its complement subsequently took place—thus has only the status of an implicature even in the positive declarative, and as such it can be cancelled.” This is more obviously true of (22) from their discussion.

Quirk et al. (1985:1081) note that an example like (23) is ambiguous between a factual and non-factual reading of the before-construction.

In the factual reading, the subject “I” was asked after having already sent a donation; in the non-factual reading, “I” wasn’t asked but could have been asked. While the clause in the before-construction may or may not express something factual, it is not assertive like the main clause.

In addition to the central temporal meaning of before, there are a number of derived meanings and uses where the notion of anteriority is somewhat bleached or associated with other meanings. For the sake of completeness I will briefly comment on examples with such meanings. As noted by Quirk et al. (1985:1081), before-constructions can be used for the expression of implausibility, preference, result, and conditional threat, as in (24)–(27) (all from Quirk et al. 1985:1081).

What Quirk et al. (1985:1081) call preference, as in (25), sometimes comes across rather as strong determination not to do what is mentioned by the before-clause, as seen in (28).

In the data collected for this paper I have also encountered a fair number of examples in which before is used with an anticipatory meaning: X takes place to prevent, avoid, or prepare for Y, as in (29)–(31), respectively.

The important point to emphasize in this characterization of the semantics of before is that underlying the various pragmatic uses exemplified by these examples there is still in all of them a basic sense of anteriority and relative temporal ordering.

The introduction to before in this section and the preceding one has served to establish a brief overall picture of the syntax and semantics of this linker (particularly its basic non-narrative meaning of anteriority) as an appropriate contextualization of the more specific topic of this paper. As already mentioned, the temporal-narrative distinction exists only in a corner of this system, i.e., in clausal constructions in sentence-final position (with one possible exception, cf., below): narrativity concerns the sequencing of situations building up a storyline and thus crucially involves clausal expressions. What emerges from the general description of before above is that in the temporal-narrative distinction, temporal is unmarked and narrative is marked. The temporally specifying use of before is possible with all its complements and in all positions, while narrative before is restricted to a particular manifestation of its syntactic potential and is used for a special effect.

3.1. Clefting

Temporal before-clauses allow clefting while narrative before-clauses do not (or change their status to temporal when cleaved), illustrated in (21), which is repeated from above, and in (32).

(21) He arrived before the ceremony began. (temporal)

→ It was before the ceremony began that he arrived.

By allowing clefting, temporal before-clauses behave like adjuncts, i.e., like adverbials fully integrated in the structure of a main clause (cf., Quirk et al. 1985:1071), while narrative before-clauses, by not allowing clefting, behave more like disjuncts and conjuncts, which are less central to the structure of the main clause.

3.2. Main Clause Phenomena

Despite the fact that narrative before-clauses look like subordinate clauses they display a number of features characteristic of main clauses rather than subordinate clauses (including temporal clauses). One such main clause phenomenon (often simply referred to as MCP or “root transformation” in the generative tradition; cf., Hooper & Thompson 1973; Green 1976; Aelbrecht, Haegeman & Nye 2012) is the possibility of having a fronted adverbial particle followed by full inversion, as in (32): [. . .] before in came Governor McAuliffe and in (33).

Secondly, we encounter instances where instead of having a narrative before-clause we have a narrative non-clausal before-construction containing an onomatopoeic interjection, as in (34).

This phenomenon (which—depending on your definition of a clause—is an exception to the rule that narrative before must have a clausal complement) shows that narrative before-clauses, unlike temporal clauses, have a degree of illocutionary independence similar to that of main clauses: while the main clause is here declarative, the before-construction is exclamatory (for discussion of clause types and utterance functions, see Bache 2000:44-48). Temporal clauses are always part of the utterance function of the main clause.

Thirdly, in narrative before-clauses the speaker or writer may switch to the dramatic historic present independently of the tense choice in the main clause (see 35).

In temporal clauses (as in 36) this is not possible.

3.3. Actionality

In sentences containing temporal clauses there is no characteristic pattern in the main clause and the temporal clause of “actionality” (i.e., event/situation types, or “Aktionsarten,” such as “stative,” “dynamic durative,” “punctual,” “iterative,” etc.; see Bache 1982). But in sentences with a narrative before-clause, there is a tendency for the main clause to express something unbounded and communicatively incomplete (e.g., a state or a dynamic durative activity) and for the narrative clause to express something bounded and (potentially) communicatively complete (e.g., punctual or telic, often with an added element of suddenness), as in (37) and (38).

In these examples the main clauses employ a past catenative progressive (kept running) and a past perfect (had . . . placed), respectively. While the first is generally recognized as unbounded with respect to actionality, referring as it does to an activity in progress, the second perhaps requires explanation (especially because the past perfect is a frequent choice in a main clause followed by a narrative before-clause): although in (38) the placing of the foot itself is a bounded event, the past perfect adds the meaning of “resulting state” to this bounded event (‘the state of having her foot on the slope after placing it there’), and that resulting state is unbounded (Bache 1994), however brief, as in this example. Both expressions are communicatively incomplete in that the receiver will expect more to come when hearing the main clause. In fact, the main clause in (38) cannot stand alone without the subsequent before-clause: we could not simply say or write Ursula had barely placed her small, slippered foot on the slope beneath the window, and then leave it at that. In (37), by contrast, it is grammatically possible to have the main clause by itself: he just kept running getting faster and faster. But even here the textual impact of the past catenative progressive is such that the hearer will expect the speaker to offer more information to justify its relevance: it is mentioned for a reason, as a context for something else (for discussion of aspect and actionality in English, and the textual functions of the progressive form, see Bache 2008:102-113, 2013:89-93). Both before-clauses express something bounded and complete, using in each case a simple past tense to refer to a punctual situation (missed and slid).

3.4. Assertiveness and Focus

While the situation expressed by a temporal before-clause is presupposed, the situation expressed by a narrative before-clause is asserted (cf., section 2.2). The difference between asserted and presupposed is seen in a sentence like He arrived before the ceremony began. Here we intuitively understand the sentence as telling us something about the main clause situation (‘him arriving’), and more specifically that it happened before the situation expressed by the before-clause (‘the ceremony beginning’). The before-clause has a supporting role in simply specifying the main clause temporally. Both formally and functionally it is a subordinate clause. In Langacker’s (1991) and Cristofaro’s (2003) cognitive linguistic terms, “the profile of one of the two [situations] overrides that of the other” (Cristofaro 2003:33; for discussion of her approach, see Bache 2015:15-17, 2016:279-280). The overriding profile (‘him arriving’) is associated with superordination and assertiveness while the overridden profile (‘the ceremony beginning’) is associated with subordination and presupposition. In Matthiessen and Thompson’s (1988) rhetorical-structure approach to texts, the main clause codes the “rhetorical nucleus,” and the subordinate clause codes the “rhetorical satellite”—a different description of the same textual asymmetry (cf., especially Matthiessen & Thompson 1988:308-317). It is ‘him arriving’ as specified that is the crux of the message and it is that which invites any immediate further communicative attention: it can be negated, questioned, confirmed, debated, and elaborated on. Thus for example, if speaker A says He arrived before the ceremony began, speaker B may react by saying No, he didn’t, but she could not say No, it didn’t (with reference to the before-clause content). Or speaker B might say Really?, to which speaker A may confirm her original assertion by saying Oh yes, he did, but only just! Here speaker A couldn’t say: Oh yes, it did! (again with reference to the situation expressed by the before-clause).

If we place sentences with a narrative before-construction in the same kind of dialogue we find that it behaves differently from its temporal counterpart. If speaker A says They had hardly heard her explanation before Jack burst out crying, speaker B could react to the situation expressed by the before-clause by saying No, he didn’t. Jack never cries! Or speaker B might again say Really?, to which speaker A may respond by saying Oh yes, he did. I’ve never seen him like that before! Unlike a temporal before-clause, a narrative before-clause is assertive and may serve as the primary prompt for further dialogue.

Even tag-questions, which normally link anaphorically to the main clause subject and verb (as in He arrived before the ceremony began, didn’t he?), cannot be ruled out as an extension to a narrative before-construction. Compare (39) and (40).

While (39) is completely out of the question, (40) is more acceptable to the native speakers that I consulted, especially if spoken with falling, low intonation to weakly reinforce or signal expected acceptance of the statement expressed by the before-clause.

However, we should not jump to the conclusion that in narrative constructions the main clause and the before-clause simply switch their standard roles as identified in sentences with temporal before-clauses (where the before-clause offers contextualization of the main clause situation, thus supporting the assertion made by the main clause). While a sentence with a temporal before-clause consists of an assertion and a presupposition, a sentence with a narrative before-clause consists of two assertions, where the one expressed by the before-clause is in fact pragmatically the stronger. That the main clause also expresses an assertion is seen in, e.g., (41), where a tag-question is added immediately after the main clause without reducing the narrative impact of the subsequent before-clause.

With the shift in status from presupposition to assertion in narrative before-clauses, there is also a shift in temporal focus. In examples with a temporal before-clause like (21), repeated from earlier, the temporal focus is on X (‘him arriving’): X gets specified as happening before Y (‘the ceremony beginning’).

In narrative examples like (4), repeated from above, the focus is on Y as happening immediately after X.

In the terminology used in our comparative approach to before in section 2.2, the time of X in (21) is in the comparative focus while the time of Y is the comparative standard. In (4) it is the time of Y that is in the comparative focus and the time of X that serves as the comparative standard.

3.5. The Down-Ranking Potential of the Main Clause

As argued in 3.4, the main clause provides contextualization for the situation expressed by a narrative before-clause. This means that it is often possible to paraphrase such a sentence by reverting the formal status of the two clauses, as seen in (42)–(45).

Here (42) and (44) contain a main clause followed by a narrative before-clause while (43) and (45) are rough paraphrases in which the original main clause situation (X) of (42) and (44), respectively, is now expressed by a subordinate clause, and the original before-clause situation (Y) is expressed by a main clause. The former main clause expression in (42) and (44) has assumed adverbial function in relation to the new main clause expression of Y. The near-synonymy in these pairs of examples is indicative of the contextual nature of X and of a special relationship between X and Y, but it is important also to note what is different between (42) and (43) and between (44) and (45). The difference is a stylistic one relating to markedness: because the main message is conveyed by the subsequent before-clause rather than by the main clause (which we would normally expect), (42) and (44) offer a somewhat livelier account than (43) and (45), where we have the normal, expected, unmarked alignment of content with syntax (subordinate clause with subordinate content linked to superordinate clause with superordinate content), and which therefore come across as more factual. Our description of narrative before-clauses must take this into account and, as we will see, it will do so by invoking the notion of “functional superordination” in section 3.8 (cf., also Bache 2016:280, 290-292). This notion depends, among other things, on two kinds of cohesion.

3.6. Cohesion: Actionality

The special pattern of actionality noted in sentences with narrative before in section 3.3 (and which is present also in the other narrative constructions looked at so far) helps create a certain kind of progressive (or cataphoric, forward-directed) textual cohesion: by expressing something unbounded the main clause is communicatively incomplete, thus making the listener expect more to come, and the subsequent before-clause meets that expectation by expressing something bounded, illustrated further in (46) and (47).

The stative, unbounded nature of X expressed by the main clause (‘me standing there for hours, waiting’ and ‘the state of having scarcely recovered her composure’) creates an element of anticipation and the sudden, dynamic, bounded nature of Y in the before-clause (‘things suddenly returning to normal’ and ‘him inviting her to join him’) offers resolution. The rhetorical focus is thus directed from the main clause to the before-clause. The speaker’s textual strategy in choosing a narrative before-construction is to “set the scene” in the main clause for what is to come in the before-clause. While the textual cohesion is progressive in such constructions, textual cohesion in sentences with temporal before-clauses is regressive (or anaphoric, backward-directed), the information offered being simply appended as contextualization of the situation expressed by the preceding main clause.

3.7. Cohesion: Polarity

Actionality is not the only possible textual tie in sentences with a narrative before-clause. Polarity is sometimes a factor too, and in this respect sentences with narrative before differ from sentences with narrative when. The observant reader will have noticed that some of the sentences with narrative before looked at so far not only use a past perfect in the main clause but also one of the following three negative adverbials hardly, scarcely, or barely, as in the repeated examples (4), (38), and (47).

(4) They had hardly heard her explanation before Jack burst out crying.

(38) Ursula had barely placed her small, slippered foot on the slope beneath the window before it slid out from under her.

(47) She had scarcely recovered her composure from that frantic evening before he invited her to join him on the royal yacht Britannia during Cowes Week.

In these examples the communicative incompleteness of the main clause is a result not only of the choice of past perfect tense with its resultant-state meaning but also the presence of the italicized negative adverb. Although the negative adverb in each case technically makes the main clause negative, ⁷ the scope of the negation is limited. In fact the negative adverb does not as such negate the proposition made by the rest of the clause in isolation: in (4) they did hear her explanation, in (38) Ursula did place her foot on the slope, and in (47) she did recover her composure. Instead the negative adverb ties in with before and has the before-clause in its scope: hardly . . . before / barely . . . before / scarcely . . . before. The general meaning shared by these collocations is ‘almost not before . . .’: although X is positively factual, i.e., X happened, it almost did not happen before Y. In other words, in conjunction with before the negative adverb creates a sense of strong immediacy: Y happened immediately or promptly after X. This is a strong cohesive tie which supplements the tie established by actionality and which helps the switch of comparative focus from X to Y.

The textual cohesion created by the three adverbials can be further strengthened by fronting them—a syntactic operation which triggers partial inversion in the main clause and thus signals its communicative incompleteness even more strongly (see 48–50).

Even standard negation (operator negation by means of not rather than by other triggers of negative meaning) may add to the progressive cohesion, as in (42), repeated from above, and in (51).

(42) He wasn’t halfway up the stairs before I threw the phonebook again.

In these examples we get a sense that Y (the situation expressed by the before-clause) immediately affects (stops short or interrupts) X. The contribution of the negation becomes clear when we compare (42) to its positive counterpart in (52).

Although the before-clause in (52) is also narrative, the construction as a whole is stylistically less intense (signals less immediacy in the relationship between Y and X) than (42): through its lexical choices the main clause signals positive, if partial, accomplishment and is thus similar to unbounded expressions with, e.g., the progressive form. The cohesive tie between main clause and before-clause is somewhat weaker than in (42), relying exclusively on the unbounded-bounded actionality pattern in the sentence.

3.8. Functional Superordination

In my paper on narrative when (Bache 2016), I concluded that the special effect of the narrative clause was achieved by it “transposing” the narration to a higher level than the ordinary narrative level of the preceding main clause. The when-clause is in other words functionally superordinate, not to a subordinate clause but to a main clause (Bache 2015:15, 2016:280-281), and it is this being in a sense “super-superordinate” that makes a narrative when-construction stylistically marked. Not only does it attract the temporal focus of the sentence as a whole (having an overriding situational profile and assuming the status of rhetorical nucleus) but it does so against the background of a situation itself syntactically framed as superordinate by being expressed by a main clause: the narrative clause takes its onset to a higher textual level from a main clause. The same applies to the clearest cases of narrative before-clauses: they too are functionally superordinate to the main clauses they follow. Being functionally superordinate to a main clause does not make a narrative when- or before-clause less dependent on the main clause than a functionally (and syntactically) subordinate temporal when- or before-clause. As pointed out in Bache (2016:280), to reach a higher level of textual intensity, the narrative clause needs to have the main clause strike a neutral level of narration. The special effect achieved is a result of the combination of two of the characteristics mentioned above: the assertive weak-strong composition and progressive textual cohesion. It is the weakly assertive factual main clause that secures an ordinary narrative level, and it is progressive textual cohesion that transposes the narration to a higher assertive level from that onset. The notion of functional superordination is similar to Matthiessen and Thompson’s (1988) notion of rhetorical nucleus and Langacker’s (1991) and Cristofaro’s (2003) notion of overriding profile referred in section 3.4, but in addition it captures more precisely the complex relationship entered into by a functionally superordinate narrative before-clause to a formally superordinate main clause. Functional superordination ties in well with the presence of formal main clause phenomena in narrative before-clauses, as noted in section 3.2, and unlike the other approaches, my approach takes the role of syntactic level into the equation.

One consequence of this description of narrative before-clauses is that the comparative approach to before-clauses suggested in section 2.2, though an appropriate starting point, is insufficient to explain narrative constructions precisely enough. In sentences containing temporal before-clauses we found that the time of X was in the comparative focus and the time of Y served as the comparative standard. Significantly these roles are reversed in sentences containing narrative before-clauses so that the time of X serves as the comparative standard and the time of Y is in the comparative focus. But in itself this is not enough to explain the special effect of such constructions. We also need to refer to the special combination of a main clause assertion with a subsequent pragmatically stronger assertion, as well as the progressive cohesion created by the choices of actionality, and sometimes also by polarity.

The characteristics presented above capture the clearest cases of narrative before-clauses. But as indicated at the beginning of section 3, there is a certain gray area between temporal and narrative use of before where only some of the characteristics apply, and this calls for a finer categorization of before-clauses.

4.1. Context-Thematization and before-Rhematization

As noted by Declerck (2006:731-735), examples like (53) and (54) should be included in the discussion of narrative before-clauses.

Declerck treats these as clefts where the main clause simply highlights the time of Y. They can both be paraphrased by means of ‘It was not until . . . that’ (‘It was not until 7 p.m. that he arrived in London’ / ‘It was not until the following holidays that they fought again’). Since the only situation expressed by the sentence as a whole is Y (the main clause simply identifying a temporal aspect of Y), the before-clause “pushes forward the action” (Declerck 2006:732), and it is therefore tempting to classify this as a special type of narrative before-clause, as suggested by Declerck (2006).

He also discusses examples like (55) and (56).

But it is not entirely clear whether he regards them as narrative or simply factual. The implication, however, is that they are to be categorized as narrative like (53) and (54). Declerck (2006) considers (55) a kind of cleft like (53) and (54) but suggests an analysis of (56) in terms of extraposition. What they all share is that the main clause offers temporal contextualization of Y.

Examples like (53)–(56) must be accommodated in a description of the temporal-narrative distinction because they obviously share certain features with the genuinely narrative before-clauses discussed in section 3. One such feature is that, in addition to its contextualizing role, the main clause in examples like (53)–(56) is communicatively dependent on the subsequent before-clause, and we even encounter cases of intensification by means of polarity (cf., examples 55 and 56). But despite these features it is in my view not unproblematic to categorize such examples as truly narrative. Although the before-clauses in (53)–(56) “push [. . .] forward the action” (Declerck 2006:732) by expressing Y, the communicative prominence of Y is weaker than in sentences with genuinely narrative before-clauses. The reason for this is that the overall construction is a special syntactic device normally employed for emphatic identification of an element of a situation potentially expressed by a main clause (in the case of cleaving) or for reasons of weight and focus (in the case of extraposition). There is thus no real situational X in the main clause but rather an element of Y recruited for special mention. X serves to identify some contextual element and typically (though not inevitably) employs for its expression a form of be as predicator and the prop word it as subject. The element recruited is often temporal (as in 53–56) but it could be some other related contextual information, as in (57)–(60).

We can even have fairly elaborate contextual information, illustrated in (61).

The fact that the main clause is situationally “thin” (i.e., typically without, e.g., an agent, a substantial content verb, and other participants) means that it does not establish a solid narrative level from which the before-clause can transpose Y to an even higher level. Quite on the contrary, the before-clause in examples like (53)–(61) remains at a rather factual narrative level.

As mentioned, we do find intensifying polarity in Declerck’s (2006) examples, but there is little restriction on the choice of actionality in the before-clause. In (62), quoted by Declerck himself (2006:734), Y is clearly and unproblematically an unbounded activity.

In both cleft sentences and sentences with extraposition, the assertive power of the main clause remains unchallenged by the before-clause, illustrated in the repeated and expanded (53) and (60).

(53) A: It was 7 p.m. before he arrived in London.

B: No, it wasn’t! / *No, he didn’t!

(60) A: It took three more drafts before we finally hit on the right formulations.

 B: No, it didn’t! / *No, we didn’t!

B’s response here appropriately picks up the subject and predicate of the main clause but not of the before-clause. What this means is that the main clause makes a circumstantial element the main theme of the sentence as a whole, and this reduces the narrative effect of the before-clause. The textual strategy employed by the speaker or writer is to present a certain contextual meaning in relation to a situation, not to present a situation Y as dramatically occurring against the background of another situation X. I therefore regard sentences like (53)–(60) as context-thematizing constructions rather than as truly narrative constructions. Although the before-clause in such sentences expresses some new situation in the storyline (and thereby resembles a genuinely narrative before-clause), it does so in relation to some foregrounded contextual information. In other words, the before-clause completes the sentence after context-thematization and will thus be said to serve a rhematization purpose. By contrast, a genuinely narrative before-clause introduces a new theme.

4.2. Neutral-Narrative before-Clauses

Declerck (2006:732) offers other examples (such as 63 and 64) of narrative before which are less marked with respect to narrativity than the examples presented in section 3.

Unlike examples (53)–(62), these sentences refer to X and Y as rather independent situations (of ‘waiting’ and ‘pushing’ in 63 and ‘standing panting’ and ‘saying’ in 64) and they express these situations as sequential (basically X before Y) but unlike sentences containing a temporal before-clause, the focus on the two situations is more balanced and shared: X and then Y. The textual strategy is not to contextualize or specify X in terms of Y but to move the narration forward with a sequence of new situations, and in that sense the before-clause “pushes forward the action” (Declerck 2006:732) and qualifies as a narrative clause. But there is little dramatic effect and no sense of functional superordination. In other words, we have something in-between a temporal and a narrative clause. I refer to examples like (63) and (64) as neutral-narrative, as opposed to the examples in section 3, which I refer to as intensive-narrative. These terms are meant in a technical sense to indicate degrees of narrative effect in the relationship between main clause and before-clause, not as descriptions of the nature of the situations expressed. Thus we find intensely dramatic events described by a neutral-narrative construction, as in (65) and (66).

Here the sentences describe dramatic events, but there is no functional superordination in the before-clauses as a result of cohesion (triggered by actionality and/or polarity): the main clauses describe bounded situations and the before-clauses simply describe what happened next.

Often non-finite before-clauses are neutral-narrative, as seen in (67) and (68).

We typically get neutral-narrative constructions in both fiction and accounts of real events as an alternative to building up the storyline by means of sequences of paratactic main clauses. Stylistically the construction offers elegant variation in the presentation of a complex plot. A neutral-narrative before-clause is either assertive (if finite) or non-assertive (if non-finite), but always factual (presenting the situation as having taken place).

In fiction the neutral-narrative construction is regularly used to mediate the transition from description of situations to direct speech, as in (64) and in (69)–(71).

The neutral-narrative construction is also particularly frequent in sports commentaries (which of course often focus on sequences of events), exemplified in (72)–(74).

But the neutral-narrative construction is used in the presentation of any sequence of events, even the movements of stock-market figures (see 75).

As we see in all these examples, the effect of using the neutral-narrative construction is to signal the arrival at a crucial point in a sequence (of two or more situations) which then may or may not serve as a platform for new events, or give rise to comments or elaboration. In that respect, the before-clause expresses a juncture in the sequence of situations making up the narration.

4.3. A New Categorization of before-Clauses

The picture that emerges is that rather than having a simple binary, equipollent temporal-narrative contrast in connection with before-clauses, we have a number of constructions displaying different degrees of narrativity: at its purest, narrativity is characterized by a number of formal and functional features (as described in section 3), but when we consider actual usage more closely, we see a more complex scenario with two other main types of construction which fall somewhere in-between the two extremes of the temporal and intensive-narrative types. In other words, we have a cline of constructions categorized according to textual function and degree of narrativity. The four main function types are here listed and defined in textual and formal terms.

Y serves to specify/contextualize X. The comparative focus is on X. The before-clause is non-narrative and non-assertive.

He arrived before the ceremony began.

The main clause thematizes a contextual aspect of Y. The core of Y serves as the rheme. The comparative focus is on the contextual aspect expressed by the main clause as the theme. The before-clause is non-narrative and non-assertive.

It was dark before she reached the village.

Y is what happens next in a sequence. The comparative focus is shared between X and Y, or slightly balanced in favor of Y as marking a juncture in a narrative sequence of situations. The before-clause is weakly narrative and may be assertive (if finite) or non-assertive (if non-finite), but always factual.

A brick flew over the wall before hitting a small kid.

X serves as situational context for Y. Y receives comparative focus and is functionally superordinate to X. The before-clause is intensely narrative and assertive. There is a progressive cohesive tie between main clause and before-clause.

They had hardly heard her explanation before Jack burst out crying

As we have seen there is some variation even within the categories, especially the last two. In neutral-narrative before-clauses, the presence of assertiveness is dependent on the predicator being finite, and in intensive-narrative before-clauses, the exact degree of narrative intensity depends on whether functional superordination is created by actionality or polarity alone or by both.

In the second category, thematization may be created by clefting or by extraposition, as we have seen, but we also find existential constructions with there and idiomatic expressions with matter of time, as in (76) and (77), respectively.

Intensive-narrative examples are relatively rare, and one gets few hits in random general corpus searches. In my core corpus there are only fourteen sentences containing an intensive-narrative before-clause, while there are lots and lots of temporal, context-thematizing, and neutral-narrative examples. On the other hand, it is quite easy to search more specifically for examples with strong polarity cohesion, i.e., combinations like hardly + before, scarcely + before, barely + before, and expressions of suddenness emphasizing the bounded actionality of the clause (before suddenly), etc., and in this way elicit many more examples of the intensive-narrative construction type. Intensive-narrative constructions are a productive kind, and they are used for a specific communicative textual purpose. So even if they are relatively rare, it is important to accommodate them in our description as a marked option.