2.1. Collocation patterns and context as predictors of modal meaning

The present special issue features three articles that investigate the specialisation of modal meaning in specific contexts. By using either corpus-based or experimental methods, these three articles offer empirical insight on what drives the development of modal meaning in certain constructions. From a theoretical viewpoint, this contributes to testing one of the foundational hypotheses of Construction Grammar, namely that linguistic expressions reflect the interaction of constructions and the linguistic material they are made up of. As stated by Fried & Östman (Reference Fried, Östman, Fried and Õstmann2004: 22), ‘words … contribute specific semantic properties to any larger construction they occur in, but a construction may also modify some of those properties, as well as add features of its own’. Whether a specific sequence should be regarded as a construction or not depends on the degree of conventionalisation associated with the form–meaning pairing. Each of the three articles investigates how the contextual environment impacts this conventionalisation process, either synchronically or diachronically.

Nadine Dietrich’s contribution, entitled ‘Motivations for specialisation: Testing the feasibility of polysemous pre-emption in the competition between will and must’, considers when these modals are in competition in functionally equivalent expressions, and one may be preferred over the other and eventually specialise in conveying a certain meaning. Dietrich delves into the functional differences that might account for the specialisation of these modals for the specific contexts of command and inference. She sets out to test the value of three types of motivation for specialisation, namely construal pre-emption, statistical pre-emption and polysemous pre-emption. She argues that differences in construal and connotation (which she labels construal pre-emption) cannot account for specialisation. In the case of near synonymy, as illustrated in particle placement alternation, for instance, competing expressions do not imply a different construal, since alternations are functional equivalents. When a pattern is better entrenched in a specific meaning than a competing pattern with the same meaning, specialisation might result from statistical pre-emption, such as we see in the case of the ditransitive pattern with explain, where *explain me this is pre-empted by the better entrenchment of the to-dative pattern (explain this to me) (Goldberg Reference Goldberg2019). However, Dietrich points out that this explanation of specialisation is circular because it takes usage as evidence. She puts forward polysemous pre-emption as a more relevant motivation for construal differences. Polysemous pre-emption means that a pattern is prevented from being used with a certain meaning in a specific context because that pattern is already entrenched in that context with another meaning. She examines the competition between will and must with two specific meanings: the deontic ‘command’ meaning (e.g. you must listen to me!) and the epistemic ‘inference’ meaning (e.g. you must be tired). For each meaning, she identifies the collexemes that will and must collocate with and tests whether each of the three types of pre-emption may account for specialisation. Based on a quantitative analysis of the final period of the Corpus of Late Modern English (CLMET 3.1; De Smet et al. Reference De Smet, Flach, Diller and Tyrkkö2011), she provides quantitative evidence for the inferential specialisation of must and will in collocation patterns such as must be/know and will remember. The command meaning of must is more frequent in must remember/forgive/excuse, while the prediction meaning is better entrenched in will remember/forgive/excuse. This supports Dietrich’s claim that in addition to construal pre-emption, polysemous pre-emption is a plausible motivation for specialisation.

The other two articles in the set investigate the impact of the co-occurrence of adverbs with modal verbs. In their contribution entitled ‘A register approach to modal (non-)concord in English: An experimental study of linguistic and social meaning’, Stephanie Rotter and Mingya Liu revisit modal concord in American English in a register approach, addressing the linguistic and social meanings of may possibly and must certainly both with and without context. Modal concord is generally understood as a phenomenon whereby two modal elements with the same flavour and force give rise to a single modality, which implies that a modal adverb co-occurring with a modal verb is modally harmonic (Lyons Reference Lyons1977) and semantically vacuous (Zeijlstra Reference Zeijlstra2007). An alternative analysis has recently been proposed by Giannakidou & Mari (Reference Giannakidou and Mari2018), emphasising the ‘modal spread’ through adverbs that may strengthen the commitment associated by default with modal verbs. Building on these results, Rotter and Liu explain how interpretive differences between modal concord constructions – may possibly and must certainly – and their single modal counterpart can challenge the principle of semantic equivalence which is at the core of the modal concord analysis. Their results show that the linguistic meaning of modal concord constructions differs from that of single modal constructions in terms of speaker commitment. In addition to the strengthening effect predicted by the modal spread analysis, Rotter and Liu uncover a weakening effect in may possibly. The perceived social meaning of modal concord constructions is also reported to be different from that of single modal constructions. Modal concord constructions are perceived as less friendly, warm and cool than single modal constructions, especially in necessity conditions as compared to possibility conditions. Formality and confidence are reported to be stronger for necessity modal concord must certainly compared to single modal must. By contrast, confidence is reported to be weaker for possibility may possibly compared to single modal may. Modal concord constructions are rated as less grammatical than single modal constructions but do not appear to be sensitive to register. Overall, the results suggest that necessity modal concord constructions increase the speaker commitment ratings, in contrast to possibility modal concord constructions, and that modal concord constructions have distinct linguistic and social meanings. These results challenge the core assumption of semantic equivalence that underlies the concord analysis.

Benoît Leclercq and Graeme Trousdale’s contribution, entitled ‘Investigating diachronic shifts within a domain of English modality: A study of collocates with well’, points in the same direction by highlighting the role of certain collocation patterns in semantic change. Using data from the Corpus of Historical American English (COHA; Davies Reference Davies2010) that span each decade from 1830 to 1970, they present a corpus-based account of recent developments in a set of constructions involving adverbs and modal verbs, namely well may/might and may/might well. Their aim is to determine whether the four patterns have undergone a functional shift and possibly contributed to the development of may and might in concessive constructions, which are known to have emerged from the epistemic uses of might and may. In previous studies, well is viewed as serving to strengthen the epistemic value of may, as in He may well have left it downstairs (Huddleston & Pullum et al. Reference Huddleston and Pullum2002). Hoye (Reference Hoye1997) contrasts the idiomatic meaning of may well, which conveys epistemic probability, with that of well may, which reinforces epistemic possibility. Leclercq and Trousdale’s findings reveal a substantial shift from non-epistemic meanings to epistemic meanings. However, none of the four patterns display a significant association with concessive meaning. In terms of frequency, the use of the well may and well might in subject–auxiliary inversion (SAI) patterns have gradually declined, probably as a result of the weakening of modal force and the increasing use of modals as hedging devices over the period considered. The shift towards the epistemic meaning also appears to have triggered a change in collocational preferences by widening the semantic profiles of the lexical verbs used in the four patterns beyond verbs of locution and cognition. Leclercq and Trousdale’s study of collocation patterns that show a shift from the deontic to epistemic domain thus confirms a well-known trend in the evolution of modal meaning (Sweetser Reference Sweetser1990; Bybee, Perkins & Pagliuca Reference Bybee, Perkins and Pagliuca1994).

2.2. Diachrony and change

Two articles in the special issue provide in-depth diachronic analyses of modal constructions in spoken and written English, arguably offering the most historically oriented perspectives in the volume. Both contributions adopt a Construction Grammar framework to trace the development of specific modal expressions, highlighting recent changes in use across contemporary spoken varieties. Methodologically, the studies align in their use of large-scale corpora and mixed-methods approaches, combining quantitative techniques such as regression modelling with qualitative analyses of semantic and pragmatic shifts. This dual focus enables the authors to map not only the frequency and distribution of emerging patterns, but also the evolving communicative functions these modals perform. Both articles thus contribute to the large body of scholarship on the historical development of modality in English, including foundational studies (Palmer Reference Palmer1990) and those rooted in grammaticalisation (Bybee, Perkins & Pagliuca Reference Bybee, Perkins and Pagliuca1994; Krug Reference Krug2000; Traugott & Dasher Reference Traugott and Dasher2002; Narrog Reference Narrog2012, inter alia), as well as more recent work situated within Construction Grammar (Hilpert, Cappelle & Depraetere Reference Hilpert, Cappelle and Depraetere2021).

The article by Lucie Latouche, Samantha Laporte and Ilse Depraetere, ‘Hedged performatives in spoken American English: Recent change and variation in their use’, investigates the diachronic development of hedged performatives (HPs) in spoken American English. HPs, first analysed by Fraser (Reference Fraser, Cole and Morgan1975), consist of a (semi-)modal verb and a performative verb, as in I have to say, I must admit, etc. While the discursive functions of HPs have been studied, their development as constructions over time has not been addressed. The novelty of this article lies in its approach to examining HPs through a Diachronic Construction Grammar perspective, which differentiates between macro-level constructions (e.g. [I + MODAL + Vperf]), modal-specific meso-level constructions (e.g. [I must Vperf]), and micro-level constructions (e.g. [I must say]). More precisely, the authors aim to address four key research questions. The first question investigates whether the diachronic trend observed at the macro-level is mirrored in the development of individual meso-level constructions or whether divergent trends exist at this level. The second question examines whether modal-specific constructions at the meso-level follow the frequency changes of their respective modal verbs or exhibit distinct trajectories. The third question explores whether micro-level constructions align with their respective meso-level trends or show idiosyncratic variation related to specific performative verbs or their illocutionary categories. Finally, the fourth question assesses whether the same diachronic trends appear across scripted and unscripted speech. The analysis draws on three corpora based on data spanning from the 1950s to 2019: The TV Corpus (Davies Reference Davies2019b), The Movie Corpus (Reference Davies2019a; see also Reference Davies, Bednarek, Werner and Pinto2021) and the spoken subpart of the Corpus of Contemporary American English (COCA; Davies Reference Davies2008–). The results reveal that HPs exhibit divergent diachronic trends depending on the modal used: [I must Vperf] constructions show a decline, [I have to Vperf] constructions are on the rise, and [I can Vperf] constructions remain relatively stable. Interestingly, the diachronic patterns of HPs do not simply mirror the changes in the base modal verbs themselves, thereby reinforcing the argument for their status as constructions. Changes in HPs involving must and have to are shown to occur predominantly at the meso-level, reflecting a shift in discourse norms away from authoritative modality. In contrast, HPs involving can show more idiosyncratic changes at the micro-level. Another significant finding presented by the authors is the difference in register: HPs are more prevalent in scripted rather than unscripted speech, though the observed diachronic trends remain consistent across different registers. The article’s key contribution is highlighting that the evolution of HPs can be driven at different constructional levels depending on the base modal, providing new insights into the intersection of modality, discourse and constructional change.

The article by Carmelo Alessandro Basile, Christophe Lenoble and Debra Ziegeler, ‘The emergence of BHT: A cognitive-functional account’, investigates the rise of the be having to (BHT) construction within the English modal system, focusing on its semantic and functional traits, as well as its (recent) diachronic emergence. Their study – the first to introduce BHT in the literature on English modality – highlights the innovative combination of the progressive aspect with modal necessity – a development that sets it apart from traditional semi-modals like have to. The novelty of this study lies indeed in its analysis of BHT as an emerging construction, a claim that challenges established assumptions regarding the compatibility of progressive forms with modal verbs, given that modal verbs are typically considered stative and non-progressive. The study addresses three research questions. First, it explores to what extent BHT and have to differ semantically. Second, it investigates the factors prompting the diachronic emergence of BHT within the English modal system. Third, it examines why BHT is predominantly observed in British English rather than in postcolonial varieties. The article’s approach combines qualitative and quantitative corpus analyses to identify patterns of use and variation, particularly focusing on British English. The findings reveal that BHT is semantically distinct from the semi-modal have to by emphasising contingency, intensity, and a lack of control over the situation of the subject referent. It conveys a sense of external necessity imposed on the subject, contrasting with the future-oriented necessity projected by have to. This distinction underscores BHT’s non-compositional meaning, which does not simply result from combining the progressive aspect with the modal verb. The study also highlights that the emergence of BHT can be linked to the grammaticalisation cycle of have to, which originally expressed possession but evolved to encompass dynamic and deontic modal functions. The higher production of BHT in British English than in postcolonial varieties of English today is attributed to ‘negative retentionism’ – the tendency of contact-based varieties to lack linguistic features that emerged after colonisation (as shown in the investigation of diachronic data from the CLMET corpus). The article’s contribution lies in documenting the BHT construction’s rise and establishing its role within the English modal system, offering new insights into the interaction between modality and aspect. The authors suggest future research on the potential spread of BHT to other varieties of English and its productivity as a future marker, emphasizing the need for further diachronic and cross-varietal investigations.

2.3. Contractions and constructional status

The final axis of this issue includes a set of articles examining contracted modal constructions, a phenomenon that has emerged as a particularly vibrant area of enquiry within cognitive linguistics and Construction Grammar (e.g. Lorenz Reference Lorenz2013a; Daugs Reference Daugs, Hilpert, Cappelle and Depraetere2021, Reference Daugs2022). These three contributions, each offering distinct theoretical and empirical insights on modal contractions, can be understood as forming a coherent progression from formal to functional considerations. On the formal dimension, researchers continue to grapple with fundamental questions regarding the granularity of constructional representation: specifically, at what point contracted variants achieve sufficient phonetic and morphosyntactic autonomy from their source constructions to warrant status as independent constructions in their own right (Krug Reference Krug2000, Reference Krug, Bybee and Hopper2001; Lorenz Reference Lorenz, Sommerer and Smirnova2020). These questions take on particular significance when we consider that the synchronic variation we observe between contracted and full forms may represent a snapshot of diachronic constructional specialisation in progress (Traugott & Trousdale Reference Traugott and Trousdale2013). Turning to meaning, while recent scholarship has made significant strides in documenting semantic and collocational distinctions between contracted modals and their full-form counterparts (Lorenz Reference Lorenz, Hasselgård, Ebeling and Ebeling2013b; Nesselhauf Reference Nesselhauf and Hundt2014; Flach Reference Flach2021), there remains considerable scope for exploring the full constructionist conception of encyclopaedic meaning. This broader perspective encompasses pragmatic dimensions and, crucially, the social indexicalities that contracted forms carry in actual usage contexts (Levshina & Lorenz Reference Levshina and Lorenz2022; Morin, Desagulier & Grieve Reference Morin, Desagulier and Grieve2024; Leclercq & Morin Reference Leclercq and Morin2025). By attending to these understudied aspects of meaning, we can appreciate just how fine-grained constructional differences can become, even between such formally subtle variants as contracted and full modal forms.

In her contribution ‘Breaking free from the be going to / gonna dichotomy: A study of variation in an emerging English modal’, Leela Azorin examines the be going to / gonna paradigm through a morphosyntactic analysis of two corpora: the Santa Barbara Corpus of Spoken American English (Du Bois et al. Reference Du Bois, Chafe, Meyer, Thompson, Englebretson and Martey2000–5) and a Twitter corpus of climate change discourse (Littman & Wrubel Reference Littman and Wrubel2019). The study identifies eight distinct variants beyond the traditional binary, including gunna, gon’, and the ‘hypercontraction’ I’ma/Imma, where the first-person pronoun, copula and modal have fused into a single morphological unit. The analysis employs four distributional criteria: subject choice, presence or absence of be, negation patterns and following verb types. Results show that be is more frequently contracted or elided with gonna than with going to, with elision rates highest in the most contracted variants like gon’ and monosyllabic forms [gə] and [nə]. The hypercontracted I’ma variant displays categorical constraints, occurring exclusively with first-person subjects due to morphological incorporation of the pronoun. The study finds that certain variants derive from gonna rather than from be going to (notably gon’ and I’ma), suggesting that gonna functions as an independent source for further grammaticalisation. Distributional differences emerge between forms: gonna appears without a following verb three times more frequently than going to, which the author suggests may indicate metadiscursive functions in conversation. Additionally, gonna shows stronger associations with contracted be forms and exhibits distinct collocational preferences, particularly with motion verbs such as go and come. Through this formal analysis, Azorin argues that variants traditionally characterised as ‘phonetic realizations’ or ‘non-standard spellings’ represent morphosyntactically distinct constructions at various stages of autonomisation. The findings support a Constructionist perspective where formal variation reflects cognitive differentiation, with gonna achieving sufficient independence to generate its own network of variants.

In their contribution ‘Negate me not, negate me never: Cross-varietal distributional skews in modal negation from a diachronic perspective’, Robert Daugs and Ulrike Schneider examine the negation and contraction patterns of will and would through a diachronic analysis of British and American English fiction from the nineteenth and twentieth centuries. The study employs Configural Frequency Analysis (CFA) to investigate nearly one million tokens extracted from multiple corpora, including COHA, Nineteenth-Century Fiction (Karlin & Keymer Reference Karlin and Keymer1999–2000) and the British National Corpus (BNC, 1995). The analysis tracks the emergence and spread of contracted forms (’ll, ’d) alongside their full counterparts, with particular attention to their interaction with different negation strategies (not, n’t, never). Results reveal that while contraction rates increased similarly in both varieties following typical S-curve patterns of language change, the contracted forms developed distinctive distributional preferences that suggest functional differentiation rather than simple phonetic reduction. Most strikingly, the study finds that ’ll and ’d strongly avoid negation with not/n’t but show marked attraction to never, with contraction rates reaching 70–75 per cent for ’ll never and 35–40 per cent for ’d never by the late twentieth century. CFA uncovers specific collocational patterns: ’d emerges as strongly associated with first-person subjects and emotion verbs (I’d like, I’d wish), while retaining will in its full form with third-person subjects and stative verbs (it will be). The hypercontracted won’t dominates negative contexts while ’ll not remains marginal, particularly in American English. Through this configurational analysis, Daugs and Schneider argue that these patterns represent not mere pronunciation variants but ‘emancipated sub-schemas’ with distinct syntactic environments and modal meanings, supporting a network-based model of linguistic representation where associative links between elements become differentially entrenched through usage.

Finally, in their contribution ‘Register as a source of non-equivalent contracted constructions: going to and gonna in British English’, Olaf Mikkelsen and Cameron Morin echo Azorin by investigating the relationship between the modal constructions going to and gonna, through a corpus-based analysis of British English data from the LiveJournal blogging platform (Speelman & Glynn Reference Speelman and Glynn2012). The study employs both Collostructional Analysis and a Behavioural Profile Analysis based on a logistic regression model of blind annotations, assessing semantic, pragmatic and social meaning factors alongside processing constraints. Working with 8,331 tokens from informal personal online narratives spanning 2002–12, the authors operationalise meaning across three dimensions: semantic (communicative function such as ‘future intention’ vs ‘future prediction’), pragmatic (temporal proximity, speaker certainty, contingency) and social (topic of discourse, degree of formality). Results reveal that register formality emerges as the only significant meaning predictor for the alternation, with gonna strongly associated with informal contexts while semantic and pragmatic variables show no significant effects. The distinctive collexeme analysis shows relatively low association measures overall, though gonna exhibits preferences for personal-sphere verbs (hang, pick, drink) and emotion verbs (love, hate, scar), while going to associates with more formal verbs (become, attempt, provide). The regression model achieves acceptable discrimination (C=0.76) with minimal lexical effects, confirming that the alternation is driven by register sensitivity rather than verb-specific constraints. Through this analysis, Mikkelsen and Morin argue that social meaning constitutes an intralinguistic predictor central to constructional meaning, validating the proposed Principle of No Equivalence (Leclercq & Morin Reference Leclercq, Morin, Sommerer and Hartmann2023) whereby formally distinct constructions must differ semantically, pragmatically and/or socially. The findings support the constructionhood of contracted modal forms and demonstrate that ‘colloquial’ status is not peripheral but central to gonna’s meaning as a distinct construction, emphasising register-based variation as a crucial mechanism for meaning differentiation between formally similar constructions.