1 Introduction: aim and scope of the special issue

Fragments constitute a pervasive characteristic of everyday language use, particularly, but not only, in spoken language. Fernández & Ginzburg (Reference Fernández and Ginzburg2002), for instance, identified 11.15 per cent of all utterances in a subcorpus of the British National Corpus (Burnard Reference Burnard2000) as fragments. Given their ubiquitous nature, fragments have, however, been surprisingly neglected in previous research. The aim of the special issue is therefore to move them further centre-stage and contribute to the linguistic characterisation of fragments in Present-Day English by combining empirical and theoretical perspectives.

One reason why fragments have garnered comparatively little attention is the unclear definition of the term itself, on which there has been no consensus in the literature. As will be reported in section 2, various terms have been used to refer to the same linguistic phenomenon: irregular sentences, non-sentences and non-sentential utterances, among others. We have chosen the term ‘fragment’ in this special issue for its widespread use and transparency.

Fragments are taken here as linguistic units which are functionally (semantically, discursively, communicatively) felicitous, although intonationally and syntactically independent and non-canonical (non-sentential) (see Haegeman Reference Haegeman, Chiba, Ogawa, Yamada, Fuiwara, Koma and Yagi1991; Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016; Hall Reference Hall, van Craenenbroeck and Temmerman2019). Typical examples of fragments to be discussed in this issue are thus the following (for further illustrations, see also section 2):

1. (1) Better not to pretend it is an option. (‘All we want’, The Economist, 26 November 2022)

2. (2) (A: Hurry up with the scissors Pauly.)

   B: Why the rush? (BNC1994 DS: KD0 499)

3. (3) (What would you go for?) Not alcohol. (COCA 2018 MOV)

4. (4) Last I recall, you were studying the law. (COCA 2017 TV: TURN:Washington'sS…)

5. (5) (B: They do my German tea, so…)

   C: What German tea? (LLC-2)

This special issue provides an up-to-date overview of current research on the topic. The unifying framework adopted for the individual contributions is that of a usage-based approach (e.g. Langacker Reference Langacker, Barlow and Kemmer2000; Bybee Reference Bybee, Díaz-Campos and Balasch2023). As alluded to in the title of the issue (viz. Fragments: construction and reconstruction), the focus is on questions such as the status of fragments as a potentially independent grammatical unit (i.e. a ‘construction’) and their ‘construction’ over time into conventional form–meaning pairs as well as their relationship to ‘complete’ constructions (i.e. ‘reconstruction’). In line with its usage-based orientation, the special issue also adopts a corpus-based methodology, providing qualitative as well as quantitative analyses of naturally occurring language data. Each contribution thus brings together, in its own unique way, two perspectives, viz. (grammatical) form and (situational, pragmatic) function, and investigates the dynamic relationship between them.

More specifically, the contributions address one or more of the following interrelated research questions:

1. (i) Why do speakers use fragments? Unlike generative theories of fragments, which are mainly preoccupied with their form, the usage-based outlook takes into account their discourse functions in equal measure. This includes the question of what determines the choice between a fragment and a full sentence (an underexplored question according to Lemke Reference Lemke2021: 3). See Fernández-Pena & Pérez-Guerra's and Haselow's contributions.

2. (ii) How are fragments used in discourse? This question relates to the discourse functions of fragments and their use in different registers and text types (e.g. Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016). Cappelle's and Mustafa & Kaltenböck's articles address this issue.

3. (iii) How can fragments be classified? More specifically, how can corpus data help us distinguish between different types and develop taxonomies of fragments (e.g. Fernández et al. Reference Fernández, Ginzburg and Lappin2007; Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016; Fernández-Pena Reference Fernández-Pena2021)? Various parameters of categorisation come into play here, such as their communicative functions, degrees of independence from the co(n)text, and degrees of conventionalisation, i.e. whether we are dealing with fully conventionalised fragments or cases of spontaneous ellipsis. See the contributions by Fernández-Pena & Pérez-Guerra and Kim.

4. (iv) How can fragments be modelled in a usage-based framework? Drawing on the findings from corpus data, many contributions consider how they can be generalised and incorporated into a grammatical model, particularly a constructionist account. Kim's article addresses this question, and also Mustafa & Kaltenböck's and Nykiel & Põldvere's.

In the remainder of this introduction, we look at these and related questions in more detail, attempting to give a brief overview of some of the pertinent topics in the research on fragments. Section 2 presents different definitions and classifications of fragments. Section 3 discusses their discourse functions, while section 4 surveys different approaches to the modelling of fragments in grammatical theory. Section 5, finally, provides an overview of the individual contributions to the volume.

2 What are fragments? Definitions and classifications

As already mentioned, prior research on fragmentary utterances lacks a comprehensive and exhaustive definition of these structures, as well as a fine-grained classification. The term ‘fragment’ itself denoting a ‘part broken off or otherwise detached from a whole’ or a ‘detached, isolated, or incomplete part’ (OED, s.v. fragment 1, 2.a.), fragmentary utterances can come in many shapes and sizes. In the broadest sense of the term, many different structures can be considered to be fragmentary, in the sense that they lack a fully fledged clausal structure: from interjections (6), discourse markers (7), reaction signals (8) and other types of formulae (9) to exclamatives (10), echo questions (11), short answers (12), freestanding subordinate clauses (13) and instances where the subject and/or operator is/are elided (14).

1. (6) Wow. (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 853)

2. (7) Right. OK. (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 852)

3. (8) Certainly not. Absolutely. Not at all. (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 852)

4. (9) Congratulations. Thanks. (Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 1103)

5. (10) How cool! (Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 1102)

6. (11) (A: Yeah, we ended up coming home Saturday?

   B: Are you serious?

   A: Yes.)

   B: Saturday? (Biber et al. Reference Biber, Johansson, Leech, Conrad and Finegan1999: 1101)

7. (12) (A: What did she give you?)

   B: A t-shirt. (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 1542)

8. (13) That I should live to see such ingratitude! (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 841)

9. (14) Hope you're right. Never seen anything like it! (Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 1540–1)

Prior research has mostly focused on fragments where clausal status is intended (e.g. (10)–(14) above). More specifically, the literature has explored structures (i) with varying degrees of compositionality, such as the more compositional example in (15) and the more conventionalised (though not fully idiomatic) instance in (16), (ii) with and without an explicit antecedent, such as (17) and (18) respectively, and (iii) structures which can or cannot be potentially reconstructed, such as (19) and (20) respectively.

1. (15) (Benigno got into a taxi and said:) To Segovia. To the jail. (Stainton Reference Stainton2006: 96)

2. (16) (We didn't have a choice, Jim.) Not a real one. Not one that was right. (Cappelle Reference Cappelle2021: 71)

3. (17) (A: Who did she see?)

   B: Bob Dylan. (Harnish Reference Harnish2009: 252)

4. (18) [Uttered by a waiter displaying a bottle of wine to customers] From Italy. (Hall Reference Hall, van Craenenbroeck and Temmerman2019: 605)

5. (19) (A: Which movie did you see?)

   B: Casablanca. (Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006: 2)

6. (20) Me first. (Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006: 2)

Some studies have used examples such as the last two to differentiate between fragments proper (19) and nonsententials (20), a contrast which reflects the two generative approaches adopted in the analysis of fragmentary utterances: sententialist accounts, which claim that all fragments have a silent sentential structure that can be reconstructed via ellipsis (i.e. I saw Casablanca in (19)), and nonsententialist approaches, which maintain that all fragments are subsentential expressions which are enriched pragmatically to full propositions within a given context (*Me am first in (20); see Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006: 1 and section 4 for further information).

Apart from the previous distinction, the literature lacks an overall homogeneous treatment of these structures. Starting with comprehensive grammars, they discuss examples like the ones above as well as related structures in sections dealing with elliptical phenomena and non-canonical syntax. Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985: 849), for instance, classify examples such as (6)–(8) as ‘nonsentences’, as they are ‘usually but not exclusively noun phrases [that] occur frequently in speech, mostly in informal conversation’. Instances of freestanding subordinate clauses and of sentences with elided subjects and/or operators ‘recoverable from the linguistic form of the sentence’, such as (13) and (14) above, are considered to be ‘fragmentary, lacking constituents that are normally obligatory’ and, hence, termed ‘irregular sentences’ (Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 838; emphasis in the original). Apart from considering clausal units with ellipsis, Biber et al. (Reference Biber, Johansson, Leech, Conrad and Finegan1999: 1099) also refer to fragmentary instances in their discussion of ‘syntactic non-clausal units’: for instance, Not really, Absolutely or Good for you, whose ‘“fragmentary” nature … reflects a dependence of the message on context, explicable in general terms either by anaphoric or situational ellipsis’. However, they also point out that further ‘non-clausal material’ which ‘extends far beyond ellipsis[,] cannot be analysed in terms of clause structure [and] is not analysable as part of any neighbouring clause’ involves, among other examples, ‘sentence fragments’. This is a category which for Biber et al. (Reference Biber, Johansson, Leech, Conrad and Finegan1999: 224–6) comprises any ‘non-clausal material in running written text’, such as headlines, captions, lists or tables. Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 944–5), finally, apart from accounting for cases of ellipsis of various types within the clause, devote a section to what they label ‘minor sentences’. This category includes the abovementioned freestanding subordinate clauses, as well as optatives (e.g. So be it.), verbless directives (e.g. Careful!) and parallel structures (e.g. The sooner, the better.). Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 855) also very briefly introduce the concept of ‘clause fragment’ in reference to ‘various kinds of verbless construction, such as open interrogative What about the others? or exclamative What a disaster!’.

The lack of a homogeneous treatment is also evident in the broad range of terms that the literature and previous studies have used to refer to utterances lacking an overt fully-fledged clausal structure:

1. (i) ‘fragmentary elliptical sentences or utterances’ (Merchant Reference Merchant2004);

2. (ii) ‘irregular/minor sentences’ (Kline & Memering Reference Kline and Memering1977; Sadock & Zwicky Reference Sadock, Zwicky and Shopen1985);

3. (iii) ‘nonsentential utterance types or units’ (Barton Reference Barton1990; Fernández & Ginzburg Reference Fernández and Ginzburg2002; Schlangen & Lascarides Reference Schlangen, Lascarides and Rudnicky2003; Barton et al. Reference Barton, Progovac, Elugardo and Stainton2005; Culicover & Jackendoff Reference Culicover and Jackendoff2005; Fernández Rovira Reference Fernández Rovira2006; Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006; Fernández et al. Reference Fernández, Ginzburg and Lappin2007);

4. (iv) ‘non-sentences’, ‘subsentences’ (Stainton Reference Stainton, Horn and Ward2004, Reference Stainton2006; Hall Reference Hall2007);

5. (v) ‘reduced constructions’ (Bauer & Hoffmann Reference Bauer and Hoffmann2020);

6. (vi) ‘clause fragments’ (Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016; Bowie & Popova Reference Bowie, Popova, Aarts, Bowie and Popova2019; Hall Reference Hall, van Craenenbroeck and Temmerman2019), and

7. (vii) ‘sentence fragments’ (Morgan Reference Morgan, Kachru, Lees, Malkiel, Pietrangeli and Saporta1973, Reference Morgan, Music, Graczyk and Wiltshire1989; Goldberg & Perek Reference Goldberg, Perek, van Craenenbroeck and Temmerman2019).

The most recent literature on fragmentary structures of a very diverse nature has opted for the term ‘fragment’: e.g. Cappelle Reference Cappelle2020, Reference Cappelle2021; Lemke et al. Reference Lemke, Schäfer, Drenhaus and Reich2020, Reference Lemke, Reich, Schäfer and Drenhaus2021; Nykiel & Hawkins Reference Nykiel and Hawkins2020; Fernández-Pena Reference Fernández-Pena2021; Goldberg & Herbst Reference Goldberg and Herbst2021; Lemke Reference Lemke2021; Abeillé & Kim Reference Abeillé and Kim2022; Nykiel & Kim Reference Nykiel and Kim2020; Fernández-Pena & Pérez-Guerra Reference Fernández-Pena, Pérez-Guerra, Conejero Magro, García, Márquez and Ruiz-Morgan2023, Reference Fernández-Pena, Pérez-Guerra, Maci and Garofalo2024 (for further information on taxonomies, see section 4).

Despite the research carried out, there is still no agreement on how to define ‘fragment’ either. In their study of clause fragments in spoken English, Bowie & Aarts (Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016: 262) define them as

Their definition thus excludes instances of (loosely attached) discourse markers and vocatives (e.g. Oh, did you want to see me, Jo?; Okay), incomplete utterances and free-standing constructions such as What about Sam? or The sooner the better.

Unlike previous accounts, Fernández-Pena (Reference Fernández-Pena2021) and Fernández-Pena & Pérez-Guerra (Reference Fernández-Pena, Pérez-Guerra, Leuschner, Barðdal, Delaby and Vajnovszkiforthcoming) propose a corpus-driven definition of ‘fragment’ based on data extracted from the British component of the International Corpus of English (ICE-GB; Nelson et al. Reference Nelson, Wallis and Aarts2002). Building on Bowie & Aarts’ (Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016) study, Fernández-Pena & Pérez-Guerra (Reference Fernández-Pena, Pérez-Guerra, Leuschner, Barðdal, Delaby and Vajnovszkiforthcoming) manually analysed a total of 9,563 fragments in samples of both spoken and written English. Their analysis resulted in a database of 957 examples of ‘valid’ fragments, a concept which Fernández-Pena & Pérez-Guerra (Reference Fernández-Pena, Pérez-Guerra, Leuschner, Barðdal, Delaby and Vajnovszkiforthcoming) define as follows:

Their definition accounts for examples of clausal fragments with nonfinite (21) or finite (22) verbs, Small Clauses (23) and other verbless clauses (24)–(25), wh-interrogatives (26) and exclamatives (27) and various types of phrasal fragments (28)–(30):

1. (21) Good interception coming in from Enrique at the back (ICE-GB:S2A-010 #018:1:A)

2. (22) If only she would admit it! (ICE-GB:W2F-008 #066:1)

3. (23) Better go now. Good old Hendon next stop. (ICE-GB:W1B-003 #105:1)

4. (24) Enough about me for the moment. (What about you?) (ICE-GB:W1B-001 #123)

5. (25) Good luck to Simon for his exams! (ICE-GB:W1B-004 #111:3)

6. (26) (I'll be able to have my first shorts and burgers Bar-B-Q on my balcony in no time at all.) And why not. (ICE-GB:W1B-002 #116:2)

7. (27) (A: We're going into Kingston.)

   B: <laugh> What a nice thought (ICE-GB:S1A-036 #059:1:B)

8. (28) (A really lovely day[.]) Brilliant views of the bridge & bay. (ICE-GB:W1B-011 #080:2)

9. (29) (Debbie's got the clock!) About 1 AM? (ICE-GB:W1B-003 #046:1)

10. (30) (a new era to my life-) Quite frightening but exciting too. (ICE-GB:W1B-014 #131:6)

Fernández-Pena & Pérez-Guerra's (Reference Fernández-Pena, Pérez-Guerra, Leuschner, Barðdal, Delaby and Vajnovszkiforthcoming) definition thus encompasses structures which had already been the focus of attention of previous studies, as is the case of the (semi-)insubordinate constructions in (22) above and (31) below (see Kaltenböck Reference Kaltenböck, Kaltenböck, Keizer and Lohmann2016, Reference Kaltenböck2021; Beijering et al. Reference Beijering, Kaltenböck and Sansiñena2019) and the Small Clauses in (23) and (32) (see Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006). Apart from these, previous research has also explored further fragmentary structures, such as comment clauses (33) (Brinton Reference Brinton2008; Kaltenböck Reference Bezuidenhout and Goldstein2013), parentheticals (34) (Dehé Reference Dehé2014; Griffiths Reference Griffiths2015) and appositives (35) (Vries Reference Vries, Aelbrecht, Haegeman and Nye2012).

1. (31) Well, funny you should ask, Florence (Kaltenböck Reference Kaltenböck2021: 127)

2. (32) Class in session. (Progovac Reference Progovac and Progovac2006: 33)

3. (33) It was that sort of time of the year I suppose (Kaltenböck Reference Kaltenböck, Aarts, Close, Leech and Wallis2013: 286)

4. (34) John will, I fear, be late. (Griffiths Reference Griffiths2015: 199)

5. (35) Yesterday, Joop asked Anna – and I might add that she, who in fact loves Jaap (also a nice guy, as you will agree), didn't see this coming – to marry him, the poor fellow. (Vries Reference Vries, Aelbrecht, Haegeman and Nye2012: 186)

Given the scholarly attention that fragments have garnered, we can also find more recent publications on specific fragmentary constructions, such as, among others, the coffee-construction (36) (Heine Reference Heine2011), the nice-of-you construction (37) (Goldberg & Herbst Reference Goldberg and Herbst2021), elliptical turns out (38) (Bauer & Hoffmann Reference Bauer and Hoffmann2020), not-fragments (39) (Cappelle Reference Cappelle2020, Reference Cappelle2021), me too fragments (40) (Abeillé & Kim Reference Abeillé and Kim2022) and why-fragments (41) (Fernández-Pena & Pérez-Guerra Reference Fernández-Pena, Pérez-Guerra, Conejero Magro, García, Márquez and Ruiz-Morgan2023, Reference Fernández-Pena, Pérez-Guerra, Maci and Garofalo2024):

1. (36) (Gillian shook hands briskly.) ‘Coffee? Tea? Do have a seat’. (Heine Reference Heine2011: 57)

2. (37) Nice of you to be with us today. (Goldberg & Herbst Reference Goldberg and Herbst2021: 19)

3. (38) Turns out, I was right. (Bauer & Hoffmann Reference Bauer and Hoffmann2020: 241)

4. (39) (It will never happen.) Not today. Not tomorrow. Not ever. (Cappelle Reference Cappelle2021: 55)

5. (40) (A: I can't deal with you right now.)

   B: Me too. (Abeillé & Kim Reference Abeillé and Kim2022: 16)

6. (41) (I mean, it was a lot of trouble to return it.) WHY DO IT? (Fernández-Pena & Pérez-Guerra Reference Fernández-Pena, Pérez-Guerra, Conejero Magro, García, Márquez and Ruiz-Morgan2023: 104)

This special issue extends the scope of previous research by including structures and issues that have so far received less attention in the literature, among others, the relationship between fragments and imperatives (42), why-fragments (43), Mad Magazine sentences (44), negated fragment answers (45), evidential last I fragments (46) and the reactive what-X construction (47), and by focusing particularly on their ‘elliptical’ nature:

1. (42) What to do now? Remember that we have to discuss publication options.

2. (43) Why not submit a special issue on fragments to ELL?

3. (44) Us submit a special issue on fragments to ELL!

4. (45) (Have you sent a proposal for a special issue before?) Not to ELL.

5. (46) Last I checked (they required a four-page proposal).

6. (47) What? A four-page proposal?

3 Why use fragments? Discourse functions

The raison d’être of fragments has not been at the forefront of generative approaches to language, whose main preoccupation is with form. In fact, the question why speakers use fragments rather than unreduced forms is, as noted by Lemke (Reference Lemke2021: 243), ‘almost unexplored’ in this research paradigm. Especially from the perspective of more formalist, computational theories of language, the discourse function of fragments is often seen as a ‘linguistic puzzle’ (Schmeh et al. Reference Schmeh, Culicover, Hartmann, Winkler and Winkler2015), a problem which needs to be ‘resolved’ (Schlangen & Lascarides Reference David, Lascarides, Bos, Foster and Mathesoni2002). Incidentally, the term ‘fragment’ itself is liable to evoke negative connotations, as something that is ‘fragmented’ and thus ‘incomplete’ and inherently ‘lacking’. And yet, language users make frequent use of such structures: as noted in the introduction, a corpus study based on a subcorpus of the British National Corpus (Burnard Reference Burnard2000) found that fragments made up 11.15 per cent of all utterances (Fernández & Ginzburg Reference Fernández and Ginzburg2002). So, clearly, they must have some role to play.

Usage-based approaches to language, by comparison, are by their very nature much more at ease with linguistic phenomena such as fragments, with a clear interest also in their discourse functions. The focus of usage-based investigations is, however, not usually on the category of fragments in general but on specific constructions that may have fragmentary form. We thus find a plethora of studies dealing with the use and functions of structures such as insubordination (e.g. If you come this way), discourse markers or comment clauses (e.g. I think, you know), parentheticals (e.g. in brief, frankly), imperatives (e.g. hold on!), formulae of social exchange (e.g. good morning), vocatives (e.g. darling!), interjections (e.g. damn!), exclamations (e.g. good boy!) and the like. A detailed discussion of these is clearly beyond the scope of this overview and we will therefore content ourselves with just a few general observations.

On a very general level, the principle seen as motivating the use of fragments, as opposed to fully fledged forms, is typically that of economy and communicative efficiency (e.g. Grice Reference Grice, Cole and Morgan1975; Hankamer & Sag Reference Hankamer and Sag1976; Greenbaum & Nelson Reference Greenbaum, Nelson, Collins and Lee1999: 117; Goldberg & Perek Reference Goldberg, Perek, van Craenenbroeck and Temmerman2019: 189). Quirk et al. (Reference Quirk, Greenbaum, Leech and Svartvik1985: 860), in this context, argue that – all things being equal – language users follow the maxim ‘reduce as much as possible’. Such reduction may also enhance clarity ‘by reducing items which are shared as “given information”, so that attention will be focused on fresh material, or “new information”’ (ibid.), except in cases where this would lead to ambiguity. The text types which most frequently make use of such reduction are informal spoken texts, while written texts generally have a much lower incidence of fragments. In a study of the British component of the International Corpus of English (Nelson et al. Reference Nelson, Wallis and Aarts2002), for instance, Fernández-Pena (Reference Fernández-Pena2021, Reference Fernández Pena, Gallardo-del-Puerto, Camus Camus and González-López2022) found that fragments occur most frequently in the spoken text categories ‘private dialogue’ and ‘unscripted monologue’ (with 36.8 and 71.1 occurrences per 1,000 parsing units, respectively). At the same time, however, they are not uncommon in the written registers either, particularly in the category ‘correspondence’, where they have a normalised frequency of 37.3 occurrences per 1,000 parsing units (see also Greenbaum & Nelson Reference Greenbaum, Nelson, Collins and Lee1999; Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016).

While economy and clarity undoubtedly play a central role in the use of fragments, this is only one part of the story. As elliptical structures, fragments arguably also require more active involvement from the hearer, who has to ‘reconstruct’ the intended structure or meaning by filling in the ‘missing’ parts. As such, fragments may be considered more interpersonal or interactive, to the extent that they ‘draw in’ and activate the hearer. On the textual level, they also contribute to greater cohesion, to the extent that the elements or structures to be ‘recovered’ are furnished by the preceding co-text. This (typically) anaphoric orientation of the fragment establishes a cohesive link with the preceding text.

Many fragments are also extra-clausal constituents (or disjuncts) in terms of their syntactic status (e.g. Kaltenböck et al. Reference Kaltenböck, Kaltenböck, Keizer and Lohmann2016). In fact, fragmentary form may act as an important cue for signalling this extra-clausal status and thus their use as, what have been referred to as, theticals (e.g. Kaltenböck et al. Reference Kaltenböck, Heine and Kuteva2011; Heine et al. Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Jany2013). Theticals are elements of sentence grammar (of varying sizes) that are being redeployed (‘coopted’) for discourse-pragmatic purposes and, as such, are syntactically, semantically and prosodically independent, positionally mobile, and may be elliptical. The communicative function of theticals has been identified as relating to the immediate ‘situation of discourse’. More specifically, this involves the dimensions text-organisation, expression of speaker attitude and speaker–hearer interaction (among others).

Many fragments have developed their own, highly specific discourse functions, often resulting from a process of conventionalisation, which can be described in terms of grammaticalisation, pragmaticalisation or constructionalisation. These functions often distinguish them from their non-reduced counterparts and may be highly idiosyncratic to a particular fragment, in which case these structures qualify as constructions in their own right in a Construction Grammar sense (e.g. Goldberg Reference Goldberg2006). Examples of such fragments with their own quirky (unpredictable) meaning are, for instance, insubordinate if-clauses, such as If you'd like to take a seat (e.g. Kaltenböck Reference Kaltenböck, Kaltenböck, Keizer and Lohmann2016), semi-insubordinations, such as Funny you should say that (e.g. Kaltenböck Reference Kaltenböck2021), or not-fragments of the type Not in a million years! (e.g. Capelle Reference Cappelle2021).

4 What is their syntactic status? Modelling fragments in grammatical theory

Most scholarly focus on fragments has been directed towards unravelling the mechanisms that elucidate their structure and meaning. As already mentioned, investigations into the structural, syntactic aspects have predominantly operated within the Generativist framework, where two contrasting theories emerge: derivational, ellipsis-based or sententialist approaches on the one hand, and non-derivational, non-ellipsis-based or non-sententialist accounts on the other. Adhering to the notion that ‘speakers mentally represent the full grammatical structure, even if they utter only fragments’ (Newmeyer Reference Newmeyer2003: 690), sententialist approaches argue that fragments possess a sentential structure that can be reconstructed through ellipsis (Hankamer Reference Hankamer1979; Morgan Reference Morgan, Music, Graczyk and Wiltshire1989; Stanley Reference Stanley2000; Merchant Reference Merchant2004), as in (17) above, adapted here as (48) for convenience. From this perspective, even the absence of an explicit antecedent in the immediate context, as in (18) above, repeated here as (49), does not pose an issue as long as the extralinguistic context enables the invocation of a relevant linguistic expression that facilitates the interpretation of the fragment (Stanley Reference Stanley2000).

1. (48) [A: Who did she see?] B: She saw Bob Dylan. (Harnish Reference Harnish2009: 252)

2. (49) [Uttered by a waiter displaying a bottle of wine to customers] From Italy. (Hall Reference Hall, van Craenenbroeck and Temmerman2019: 605)

Conversely, there are scholars who reject the assumption of such silent syntax and advocate instead for a non-derivational explanation of fragments, where ‘what you see is basically all the linguistic structure you get’ (Hall Reference Hall, van Craenenbroeck and Temmerman2019: 605), as in (20) above, adapted here as (50). Fragments are thus considered to be generated as subsentential expressions, which are pragmatically enriched to full propositions within a given context (Barton Reference Barton1990; Ginzburg & Sag Reference Ginzburg and Sag2000; Carston Reference Carston2002; Barton & Progovac Reference Barton, Progovac, Elugardo and Stainton2005; Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006; Stainton Reference Stainton2006; Bezuidenhout Reference Bezuidenhout and Goldstein2013).

1. (50) Me first. [*Me am first] (Progovac et al. Reference Progovac, Paesani, Casielles and Barton2006: 2)

One approach that is particularly suited for the analysis of fragments is a usage-based one, where linguistic structure is assumed to arise out of usage events (Goldberg Reference Goldberg2006; Bybee Reference Bybee2010). Such a ‘bottom-up’ view of language, which does not postulate a strict separation between actual usage and speaker's grammatical knowledge, and crucially allows for all kinds of non-sentential structures to be stored in the grammar (‘grammat-i-con’) can account not only for their grammatical status (e.g. independent constructions in their own right) but also for their diachronic development (e.g. increasing conventionalisation). The usage-based perspective posits, among its principles, that speakers exhibit both impressive creativity and remarkable repetitiveness in their linguistic expressions (Goldberg Reference Goldberg, Hoffmann and Trousdale2013: 26). According to this view, these dynamics of creativity and repetition extend beyond the scope of individual lexical items to influence entire constructions (Van de Velde et al. Reference Van de Velde, De Smet, Ghesquière, Smet, Ghesquière and de Velde2015). Along these lines, a particularly promising usage-based framework for analysing multi-word patterns is Construction Grammar. As non-derivational approaches, usage-based construction grammars operate on the premise that our understanding of language is rooted in language use. According to (Cognitive) Construction Grammar, a linguistic expression is a construction if it is non-predictable (Hilpert Reference Hilpert2019: 13) and/or sufficiently prevalent in language. In consequence, phenomena that have been traditionally attributed to ellipsis, such as fragmentary expressions, can be better explained as unique pairings of form and meaning, that is, as constructions in their own right. In Bauer & Hoffmann's (Reference Bauer and Hoffmann2020: 241) words, fragments can ‘become entrenched in a speaker's mental grammar provided they occur with sufficient token frequency’.

To illustrate constructionalist studies on fragments, in one of the earliest publications on this topic, Heine (Reference Heine2011) explores the so-called coffee construction. She argues that instances of non-coordination-based ellipsis, such as Coffee?, differ from their non-reduced counterparts (Would you like some coffee?) in that the former (i) lack stylistic markings, being exclusively used with consumption items, and (ii) exhibit a more specific correspondence between the reduced form and its meaning. Heine (Reference Heine2011: 74) thus asserts that the schema [NP?] ‘is stored as a rather independent construction in its own right, which overlaps with the full form but also displays independent features’. In a similar vein, Goldberg & Herbst (Reference Goldberg and Herbst2021) delve into the nice-of-you construction (e.g. It's nice of you to be worried about my health), a pattern that instantiates a distinctive form–meaning pairing: it tends to associate with specific adjectives (e.g. nice, good, wrong, typical) and conveys the precise meaning of evaluation of how an action reflects on its agent. Finally, Cappelle (Reference Cappelle2021) and Abeillé & Kim (Reference Abeillé and Kim2022) are examples of constructionalist, usage-based studies of emphatic constructions. Cappelle (Reference Cappelle2021) investigates the use of emphatic negative fragments, such as Not today! or Not him again!, which he includes within the constructional schema Not X!. Abeillé & Kim (Reference Abeillé and Kim2022) propose a construction-based Head-driven Phrase Structure Grammar account of me too fragments in English and French. These fragments consist of nominal remnants and additive adverbs that can be used emphatically (e.g. [A: John left.] B: HIM too!) or contrastively (e.g. [A: John left.] B: Me too). Their corpus-based study demonstrates that these fragmentary expressions exhibit various types of (person, number, gender) mismatches with respect to their antecedent clauses, may lack overt antecedents and can occur in island configurations. These findings challenge derivational approaches and thus advocate for a constructional account.

Finally, scholarly attention has also been directed towards the empirical investigation of fragments. The literature can be categorised into more qualitative analyses that depict or classify fragments extracted from various types of English texts (e.g. plays, journals, essays) (Kline & Memering Reference Kline and Memering1977; Malá Reference Malá2000; Schuster Reference Schuster2006), as well as quantitative analyses based on corpora. In the latter category, there are studies that propose taxonomies driven by corpora (Fernández-Pena Reference Fernández-Pena2021), most of them utilising English oral corpora to facilitate the automatic identification and classification of fragmentary structures through Machine-Learning techniques, and/or interpret those fragments within the framework of Head-driven Phrase Structure Grammar (Purver et al. Reference Purver, Ginzburg, Healy, van Kuppevelt and Smith2001; Fernández & Ginzburg Reference Fernández and Ginzburg2002; Schlangen & Lascarides Reference Schlangen, Lascarides and Rudnicky2003; Fernández Rovira Reference Fernández Rovira2006; Fernández et al. Reference Fernández, Ginzburg and Lappin2007). Only a limited number of corpus studies have explored the usage and/or communicative function of specific types of fragments in spoken and written English (Greenbaum & Nelson Reference Greenbaum, Nelson, Collins and Lee1999; Bowie & Aarts Reference Bowie, Aarts, López-Couso, Méndez-Naya, Núñez-Pertejo and Palacios-Martínez2016; Cappelle Reference Cappelle2020). Finally, to the best of our knowledge, only Fernández-Pena & Pérez-Guerra (Reference Fernández-Pena, Pérez-Guerra, Leuschner, Barðdal, Delaby and Vajnovszkiforthcoming) have investigated from a constructional perspective the cognitive mechanisms operating in fragments which allow them to work as alternatives in the competition between fragments and complete clauses in language.

5 The contributions to this special issue: an overview

As highlighted in the preceding sections, while there are still a number of open questions in the research on fragments, a usage-based approach can provide a useful framework for answering them. This is precisely what the six contributions to this special issue set out to do: they offer new insights into the use, discourse functions and syntactic status of English fragments based on empirical evidence provided by natural language data.

The contribution by Bert Capelle entitled ‘The special position of fragments and imperatives in polished prose: Data from The Economist editorials’ identifies fragmentary clauses in a genre where one might least expect them: in carefully edited journalistic writing, more specifically, editorials in The Economist. The article first highlights the formal and functional similarities between the two categories in question, i.e. fragments such as What to do? and imperatives such as Take spending cuts as an example, both of which are shorter than canonical clauses and typically have non-truth-conditional semantics. It is further shown that fragmentary clauses and imperatives also have a similar use. Rather than occurring evenly distributed in a text, both constructions tend to appear in specific, conspicuous positions: at the beginning or end of a paragraph, particularly in the second paragraph, where the writer presents a contrasting view from the opening paragraph, and in the concluding paragraph. The study thus makes a case for considering stylistic properties in the characterisation of fragments and, in doing so, for expanding our scope to broad discourse phenomena, going well beyond the boundaries of the sentence.

The study ‘A constructionalist account of why-fragments and Mad Magazine sentences: The “Sceptical Small” construction’ by Yolanda Fernández-Pena and Javier Pérez-Guerra focuses on why-fragments or WFs (Why deal with fragments here?) and Mad Magazine sentences or MMs (Us deal with fragments here?). WFs and MMs, having a meaning and discourse function equivalent to complete (interrogative) sentences, occasionally convey an ‘enriched’ specific nuance of scepticism regarding a proposition. By analysing data retrieved from the 1994 and 2014 editions of the British National Corpus and the Corpus of Contemporary American English, Fernández-Pena and Pérez-Guerra show that, whereas WFs remain formally and semantically ‘under construction’, MMs prove to be fully conventionalised in contemporary English. Having dismissed a deletion-based explanation, the authors give support to the analysis of WFs and MMs as examples of an umbrella Sceptical Small (i.e. untensed) construction. Treating these patterns as constructions within the framework of (Cognitive) Construction Grammar allows for an account of both their non-compositional enriched interpretation and their formal (intonational, structural, syntactic) similarities.

In his contribution ‘Syntactic fragments in social interaction: A socio-cognitive approach to the syntax of conversation’ Alexander Haselow focuses on the reciprocal interplay of speakers in successful language-based interaction and argues that fragments play a crucial role in this. Drawing on the principles of Conversation Analysis, the article argues for a reorientation in syntactic description from a single-mind to a dual-mind view of syntax, one in which interacting speakers are not studied as isolated agents but as forming a social unit. To this end, an analytic framework labelled ‘dual-mind syntax’ is proposed, which includes a social signature in descriptions of spoken syntax, more generally, and fragments, more specifically. Accordingly, fragments are not seen as deficient or incomplete syntactic units, but, on the contrary, as resulting from a successful communicative practice which designs structures in a responsive-contingent fashion in social interaction. As such, fragments are part of a network of mutually dependent, coherent structures. Based on empirical data from the Santa Barbara Corpus it is further shown how speakers use fragments for coordinating actions, for collaborative structure-building, and, more generally, for contributing to the emergence of a structurally integrated, coherent whole.

Jong-Bok Kim in ‘(Negated) fragment answers in English: A discourse-oriented and construction-based perspective’ conducts a corpus-based analysis of fragment answers to wh-questions, such as [A: What was his motive?] B: Not money, where not syntactically combines with a remnant phrase. Kim examines the category of the remnant in the fragment answer (AdvP, PP, NP, AP, VP, subordinate clause), its syntactic function concerning a reconstructed propositional source (modifier, subject, object, predicative complement, oblique complement), c/overtness of the antecedent question and constituent versus sentential negation in the fragmentary construction. The data, retrieved from the Corpus of Contemporary American English, reveal that these negated fragment answers often refer to syntactic/semantic structures of the antecedent questions and that, at times, they are associated with discourse structures that diverge from the questions. Consequently, Kim rejects a deletion-based approach and advocates a surface-oriented direct-interpretation approach couched within the framework of construction-based Head-driven Phrase Structure Grammar, according to which the status of the fragment is not clausal but a maximal phrasal category that is projected into a non-sentential utterance.

The article by Ozan Mustafa and Gunther Kaltenböck is entitled ‘Last I heard: On the use of evidential last I fragments’ and investigates a fragment that has emerged only recently. It takes the elliptical form of ‘last I + Verb’, as in last I checked, last I heard, and is evidential in meaning. Data from COCA show a steep increase in frequency for this construction in recent decades. Syntactically, these fragments are identified as disjuncts, which are positionally mobile with respect to their host clause and whose ‘elliptical’ form can be linked to different ‘full’ forms, viz. specificational sentences and temporal adjuncts. What makes these fragments interesting, apart from their relative novelty, is that their underlying evidential meaning gives rise to different discourse functions. Depending on contextual use, these range from downtoner to booster and ironic use. Fragmentary form, moreover, is shown to be more closely associated with evidential meaning than the unreduced counterparts, with reduced form thus acting as an important marker of evidentiality. Finally, it is argued that the grammatical status of these fragments is best captured by a constructional account which identifies them as constructionalising units.

In ‘Ellipsis meets the reactive what-x construction in English’, Joanna Nykiel and Nele Põldvere offer an analysis of expressions such as [A: but this museum is double the size of Tate Modern even with the the uh the extension] B: what the extension that's not open as instances of ellipsis, whose interpretation and usage in discourse can be justified by categorising them as either ‘reprise’ (i.e. clarification) or ‘direct’ fragments. Drawing on data retrieved from the spoken London–Lund corpora, and adopting a nonsententialist approach to the study of fragments couched within Head-driven Phrase Structure Grammar, Nykiel and Põldvere describe the discursive (never discourse-initial), structural (optional what plus S/NP/PP/VP/CP), syntactic (matrix-clause single-tone-unit phenomenon) and semantic (non-interrogative) characteristics of the reactive what-x construction in British English. The construction's formal and semantic properties are accounted for by the assimilation of reactive reprise and direct wh-x types into, respectively, merger and sprouting strategies.