Linguistic creativity and productivity have become active topics of research, especially amongst scholars who employ insights from Construction Grammar (CxG) in their analyses. The question of how creativity should be defined and operationalized has in particular emerged as a major point of discussion and debate (e.g. Hoffmann Reference Hoffmann2018, Reference Hoffmann2024; Bergs Reference Bergs2018; Bergs & Kompa Reference Bergs and Kompa2020), and the psychological, contextual and interactional factors that underlie creative language use have received a great deal of attention in recent work (e.g. Hoffmann Reference Hoffmann2018, Reference Hoffmann2020, Reference Hoffmann2024; Herbst Reference Herbst2018; Turner Reference Turner2018; Hartmann & Ungerer Reference Hartmann and Ungerer2023). These studies have substantially increased our understanding of linguistic creativity and productivity, but there are still many open questions about their mutual relationship as well as their connection to related phenomena, such as extravagance (Haspelmath Reference Haspelmath1999; Eitelmann & Haumann Reference Eitelmann and Haumann2022; Trousdale & Norde Reference Trousdale, Norde, Tanja and Turo2025). In this special issue, the authors investigate a variety of questions related to creative language use and productivity, each providing new insights into the ongoing discussion of the nature of linguistic innovation in the context of English.

From a theoretical perspective, all contributions in this issue are conceptually situated within the general framework of usage-based Construction Grammar, where the speaker’s total grammatical knowledge is described in terms of constructions, i.e. conventionalized form–meaning pairings, which come in various levels of schematicity (from fully specified constructions to abstract templates) and structural complexity (ranging from morphemes to syntactically complex constructions, such as the Caused Motion Construction, e.g. she sneezed the napkin off the table; Goldberg Reference Goldberg1995). Constructions are conventionally modelled in terms of a network structure, the constructicon, where the constructions themselves are depicted as nodes in the network, while the relationships between them are described in terms of links that connect the nodes together (e.g. Diessel Reference Diessel2019; Flach Reference Flach, Tanja and Turo2025). However, while this node-and-link model is well established in constructional literature, the precise role of links has only started to attract more attention relatively recently (e.g. Hudson Reference Hudson2015; Schmid Reference Schmid2020; Hilpert & Flach Reference Hilpert, Flach, Depraetere, Cappelle, Hilpert, De Cuypere, Dehouck, Denis, Flach, Grabar, Grandin, Hamon, Hufeld, Leclercq and Schmid2023). A particularly important question concerns what kind of information should be included in the nodes of the network (i.e. constructions), and what should be modelled by way of links between the nodes. In this special issue, Flach examines this question specifically from the perspective of productivity and creativity.

The studies included in this issue are based on a thematic workshop Creativity and Productivity in CxG organized at the University of Helsinki in September 2023 by Tanja Säily and Turo Vartiainen. In our call for papers, we asked the participants to reflect upon a number of questions pertaining to the theme of the workshop, such as the operationalization of linguistic creativity, factors pertaining to the innovation, actualization and diffusion of novel constructions, the conceptual mechanisms underlying linguistic creativity and productivity (e.g. conceptual blending, mismatch, coercion, entrenchment), and social and regional factors pertaining to them. A topic that was brought up in a number of workshop papers, as well as in the contributions to this special issue, was the division of linguistic creativity into F(ixed)-creativity and E(nlarging/extending)-creativity, a taxonomy introduced by Sampson (Reference Sampson and Hinton2016). According to Sampson, F-creativity refers to original usages that are nevertheless based on the existing inventory of constructions in the language, while E-creativity points to innovations that violate or extend the grammatical rules or constraints of the language and result in the creation of entirely novel constructions. Sampson’s proposal has formed a topic for lively debate in recent studies, and considering the often gradient nature of linguistic data (e.g. Traugott & Trousdale Reference Traugott and Trousdale2010), it is not altogether surprising that the categorical division of creativity into two types has not been accepted without criticism. For example, many authors have suggested that F-creativity and E-creativity should be reconceptualized as representing end points on a cline (e.g. Bergs Reference Bergs2018; Uhrig Reference Uhrig2018), and several articles in this issue engage directly with these concepts, pointing out a need for further theoretical amendment and making specific suggestions to this effect (Trousdale & Norde Reference Trousdale, Norde, Tanja and Turo2025; Laws Reference Laws, Tanja and Turo2025; Flach Reference Flach, Tanja and Turo2025).

Discussions of linguistic creativity are further complicated by the fact that the concept of creativity can be legitimately approached from various perspectives. The question of ‘what is creative’ is of course fundamental for any investigations of creativity, but additional questions, such as ‘who are the most creative speakers?’, ‘who consider an expression to be creative?’, ‘at what period of time is an expression creative?’ and ‘in which contexts is an expression creative?’, are also crucially relevant. These questions introduce conceptual and definitional complexities that are challenging to account for in any theory of language – and yet interspeaker variation, diachronic change, register variation and the pragmatics of interaction have long been highly relevant topics in usage-based linguistics. For instance, and as pointed out by Trousdale & Norde (Reference Trousdale, Norde, Tanja and Turo2025), linguistic creativity has typically been studied from a synchronic perspective, but it seems to us that, as a concept, creativity also includes an obvious diachronic element that is fundamentally important: while an expression can certainly be considered creative, or innovative, in synchrony (i.e. with respect to the present stock of grammatical constructions and variation within the linguistic community), the expression is also necessarily embedded in the history of that community – its degree of creativity is measured against what has been produced in the past. One consequence of this is the fleeting nature of creativity; when an innovation spreads through a community of speakers, it starts to lose its creative flair; or alternatively, what starts as E-creative may over time become F-creative (see Flach Reference Flach, Tanja and Turo2025). Furthermore, the speaker is not necessarily the sole locus of linguistic creativity, at least in spoken interaction. In a recent paper, Hoffmann (Reference Hoffmann2024) provides a model for constructional creativity (the ‘5C model’), where the interactional dynamics of the speech event are taken into consideration by acknowledging the relevance of both the ‘constructor’ and the ‘co-constructor’ (see also Bergs & Pentrel Reference Bergs, Pentrel, Tanja and Turo2025; Trousdale & Norde Reference Trousdale, Norde, Tanja and Turo2025).Footnote ¹ While the application of the 5C model remains to be tested in empirical work, the integration of cognitive and interpersonal aspects certainly seems promising.

In recent research, construction grammarians have also applied increasingly sophisticated methods in their analyses of constructional changes, and the empirical basis that has been accumulated on the topics of creativity and productivity has inspired scholars to extend their research questions to psycholinguistics (e.g. Hoffmann Reference Hoffmann2018), language acquisition (Goldberg Reference Goldberg2019), sociolinguistics (Säily et al. Reference Säily, Hilpert, Suomela, Buschfeld, Ronan, Neumaier, Weilinghoff and Westermayer2024; Säily & Vartiainen forthcoming; Silvennoinen Reference Silvennoinen, Tanja and Turo2025; Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025), regional variation (Hoffmann Reference Hoffmann, Buschfeld, Hoffmann, Huber and Kautzsch2014, Reference Hoffmann2020; Brunner & Hoffmann Reference Brunner, Hoffmann, Ngefac, Wolf and Hoffmann2022; Vartiainen, Callies & Liimatta Reference Vartiainen, Callies, Liimatta, Tanja and Turo2025), and deliberations of the domain-specificity of creativity (Trousdale Reference Trousdale2020), for example. These studies provide new perspectives to some of the questions outlined above: who are the most creative and productive speakers, and under what conditions should an expression be regarded as creative? For instance, regardless of whether one approaches creativity from a synchronic or a diachronic perspective, there is always a baseline against which creativity is determined (however this may actually be measured), and this baseline will necessarily shift when data from different varieties or different population groups are considered.

Our second focus in this special issue is on productivity. Chomsky (Reference Chomsky1964: 22) distinguishes between rule-governed and rule-changing innovation, and these have been taken as representing productivity and creativity, respectively (e.g. Bauer Reference Bauer2001: 93). They also notably resemble the concepts of F-creativity and E-creativity discussed above, and like them, have increasingly been interpreted as a cline (Bauer Reference Bauer2001: 66; Bergs Reference Bergs2018; Laws Reference Laws, Tanja and Turo2025). Productivity is regarded as a key component of constructions with open slots, such as the English future construction BE going to V (Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025): the wider the range of lexical items that can be used to fill the verb slot, the more productive the construction. In constructions with multiple open slots, the productivity of each slot can be analysed separately.

The notion of productivity has been extensively discussed in morphology. While early accounts argued for a qualitative, either–or difference between productive versus unproductive rules (e.g. Bauer Reference Bauer1983: 99–100), productivity later came to be conceived of as a quantitative, gradient phenomenon (e.g. Baayen Reference Baayen, Booij and van Marle1992). Baayen (Reference Baayen, Lüdeling and Kytö2009) identifies three facets of productivity, with operationalizations that enable them to be studied in text corpora. The first of these is realized productivity or the extent of use of the construction, which can be operationalized as the number of different word types used in the open slot (to use terminology from Construction Grammar). The second facet is expanding productivity, which estimates the rate of expansion of the construction or its share of the novel items occurring in the corpus as a whole, operationalized as the number of hapax legomena, or words used only once in the slot, divided by the total number of hapax legomena in the corpus. The third facet is potential productivity, which estimates the growth rate of the vocabulary used in the open slot of the construction; this is operationalized as the number of hapax legomena used in the slot divided by the number of all word tokens used in the slot.

In CxG, the concept of productivity has been used in a somewhat wider sense. In addition to the number of new or different lexical items used in the construction, the focus has been on variability in the kinds of types used: while some are, for instance, semantically quite coherent or similar to each other, others differ from each other in various ways (Barðdal Reference Barðdal2008: 27). The types can be thought of as unevenly covering a variation space, in which areas that are already well covered tend to attract new members more easily (Suttle & Goldberg Reference Suttle and Goldberg2011: 1254). Productivity is also involved in the spread of the construction to new areas or contexts, which in grammaticalization research is commonly known as host-class expansion (Himmelmann Reference Himmelmann, Bisang, Himmelmann and Wiemer2004: 32). One methodology developed for detecting such expansion comes from distributional semantics: the words occurring in the open slot of the construction in a diachronic corpus are automatically classified into semantic categories based on their contexts of occurrence, and the emergence of a cluster of words belonging to the same semantic category is taken as an indicator of host-class expansion (Perek Reference Perek2016). Outside the paradigm of CxG, similarly context-aware measures have been adopted from the field of information theory: for instance, Degaetano-Ortlieb & Teich (Reference Degaetano-Ortlieb, Teich, Reiter, Alex and Zervanou2016) assess productivity through the concept of surprisal.

Recent discussions in Construction Grammar have increasingly addressed variation in productivity and partial productivity, including individual variation and how speakers learn what not to say (Goldberg Reference Goldberg2019; De Smet Reference De Smet2020). Related to variation, diachronic change in productivity is regarded as an important component of constructionalization and constructional change (Traugott & Trousdale Reference Traugott and Trousdale2013; Hilpert Reference Hilpert2013; Perek Reference Perek2016), and recent work has expanded the focus from cognitive to social factors (e.g. Säily et al. Reference Säily, Hilpert, Suomela, Buschfeld, Ronan, Neumaier, Weilinghoff and Westermayer2024; Säily & Vartiainen Reference Säily, Vartiainen, Mahlberg and Brooksforthcoming). Methodologically, new solutions have been proposed to the problem that statistical measures of productivity are sensitive to sample size, facilitating the analysis of variation and change in productivity (e.g. Säily et al. Reference Säily, Hilpert, Suomela, Buschfeld, Ronan, Neumaier, Weilinghoff and Westermayer2024). The present issue makes significant contributions to these trends, increasing our understanding of both intra- and extralinguistic factors influencing productivity within the framework of CxG (e.g. Ceuppens & De Smet Reference Ceuppens, De Smet, Tanja and Turo2025; Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025). It also introduces novel methodological solutions based on permutation testing as well as a combination of Large Number of Rare Events (LNRE) modelling and the analysis of frequency spectra (Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025; Vartiainen, Callies & Liimatta Reference Vartiainen, Callies, Liimatta, Tanja and Turo2025).

Graeme Trousdale and Muriel Norde start this special issue by investigating questions related to E- and F-creativity, extravagance, conceptual blending, and constructionalization in light of English libfixes (e.g. pawfection, infotainment) and Dutch pseudoparticiples (e.g. besneeuwd ‘snow-covered’; ontmergd ‘demarrowed’). Following their earlier work (Norde & Trousdale Reference Norde and Trousdale2024), the authors split Sampson’s F-creativity into two types (F1 and F2), where F1-creativity relates to extensions that are fully sanctioned by an existing schema while F2-creativity refers to partial sanction. Alternatively, the authors propose that if the organization of constructions in the constructional network is viewed in terms of exemplar clouds, F1-creativity refers to the increased strength of an existing exemplar cloud, whereas F2-creativity relates to the inclusion of marginal members in the cloud. The authors provide a highly interesting discussion of the role of extravagance in creativity and language change, underscoring, for example, the interactional nature of creativity and change (i.e. that both the speaker and the hearer play a role in the creative process), and the gradient nature of extravagance, which has consequences on the conventionalization and productivity of the construction.

Alexander Bergs and Meike Pentrel examine three constructions that appear to be deliberately created as vehicles for witty and expressive, often scripted, communication: the X-much construction, the extrasentential not construction and the because X construction. Each of these constructions is a relatively recent innovation, and they recycle or blend structural elements of earlier, well-established constructions in new ways, with layers of pragmatic and intersubjective information that were not part of the older constructions. As a consequence, the constructions are particularly useful in genres associated with extravagant communication, such as movies, TV shows and various online genres. Bergs and Pentrel’s study deals with several interesting topics pertaining to linguistic innovation and the propagation and conventionalization of language change. For example, in their analysis of the constructions they emphasize the role of intersubjectivity, which resonates well with, for example, Hoffmann’s (Reference Hoffmann2024) point about the role of co-constructors in linguistic creativity. Furthermore, their focus on deliberately created expressive constructions raises questions about the spread and conventionalization of such constructions. Can they ever become conventionalized to the extent that they spread to new genres? Can they survive in the long term, or will they ultimately fizzle out once their extravagant flair has worn off? As the authors point out, there is certainly room for further empirical work in the future.

Jacqueline Laws explores the role of coercion, or a phenomenon where a semantic and/or structural mismatch between a construction and a somehow ill-fitting item is resolved in favour of constructional meaning, as a facilitator of productivity and creativity in complex verbs. To do this, she combines the coercion cline proposed by Audring & Booij (Reference Audring and Booij2016) with the cline between F-creativity and E-creativity proposed by Bergs (Reference Bergs2018) and Uhrig (Reference Uhrig2018). Her idea is that the senses of complex verbs that require no coercion in order to be interpreted represent F-creativity, or ‘pure’ productivity, whereas those that require a degree of coercion belong to the central section of the cline, or Extended-Fixed (ExFx) creativity. Finally, E-creativity is represented by extreme or ‘unruly’ coercion, with senses that exhibit unique argument structure patterns. Laws analyses this creativity–coercion continuum in a large-scale empirical study comprising an exhaustive list of complex verb senses extracted from the 1994 and 2014 versions of the Spoken British National Corpus, complemented by data from the Corpus of Contemporary American English (COCA) and the Oxford English Dictionary (OED). She finds that although F-creative uses lacking coercion form the majority pattern and E-creativity a tiny minority, ExFx-creative coercion appears to play an increasingly significant role over time, particularly with more complex argument schemas. Laws suggests that this could be due to the fact that English only has four main verb-forming suffixes: coercion is a helpful way of expressing a maximal range of meanings with this limited number of suffixes. She also argues that the widespread use of coercion in Present-Day English complex verbs means that it should be accounted for in constructional descriptions.

In her contribution, Susanne Flach examines questions related to productivity and creativity from a constructional network perspective by investigating the into-causative construction (e.g. they scared us into working harder) in COCA. As Flach points out, the construction stands out from other argument structure constructions in that the items used in the main verb slot (V1) are primarily associated with other constructions, which contributes to the construction’s ‘creativity effect’. From a CxG perspective, Flach makes an interesting distinction by defining creativity as ‘link-establishing or link-reinforcing language use’ and productivity as relating to the ‘quantity and quality of relationships between network elements’. By using sophisticated statistical analyses, such as Collostructional Analysis, Correspondence Analysis and LASSO regression on her data, Flach shows, for example, that the verb’s general flexibility (link-creating potential), as well as its higher flexibility across genres, is correlated positively with its association with the into-causative construction. Flach’s study seamlessly combines state-of-the-art statistical methods with important theoretical questions both from a CxG point of view and from the perspective of creativity and productivity by focusing on links instead of nodes in the constructional network, thus emphasizing the gradient and probabilistic nature of linguistic creativity and further problematizing the dichotomous split between E- and F-creativity. As Flach points out, ‘even F-creativity expands the constructicon, as every use alters the composition and internal structure of the network, if only by the addition of new links’.

Hilke Ceuppens and Hendrik De Smet focus on the semantic side of productivity. Like Barðdal (Reference Barðdal2008), they understand productivity as extensibility, but they argue that previous research has mostly considered semantics in terms of how it constrains combinatory extensibility, or how elements enter into new combinations in combinatorial space (Himmelmann’s host-class expansion). They draw our attention instead to semantic extensibility, or extension in semantic space, testing two hypotheses from Radial Network Theory that predict how semantic extensibility is constrained: (1) ‘senses stand in directional motivational relations to one another’ and (2) ‘frequency is one determinant of the likelihood that a sense can form the basis for semantic extensions’. Their evidence comes from three cases of semantic loss involving the items awful (adjective), about (preposition and adverb) and so (multifunctional) in the OED and corpus data. The evidence supports the hypotheses: Ceuppens and De Smet find that the core or sanctioning sense and the senses derived from it decline together, indicating motivational relations. Furthermore, this frequency decline coincides with a corresponding increase in a new core sense in the network, implying that frequency plays a role. The fact that the sense that becomes the new core fails to support the older senses while sanctioning its own peripheral senses indicates that sense relations are directional. These insights into the constraints on semantic extensibility provide a welcome complement to approaches based on distributional semantics in the study of constructional productivity.

Combining Construction Grammar with sociolinguistics, Olli Silvennoinen analyses the propagation (spread in the language community, e.g. across genres) and actualization (spread to new linguistic contexts) of the future adverb construction going forward in COCA, 1990–2019. He identifies significant periods of change using variability-based neighbour clustering and finds that the construction originates in business contexts but later expands to other domains; this expansion is correlated with a frequency increase. Based on metalinguistic commentary online and in the OED, the construction appears to retain its often negatively charged association with business and administration in the language community, and Silvennoinen argues that such social meanings should be incorporated in constructional descriptions. The emergence of the construction as seemingly fully actualized can be explained through a network-based account of multiple inheritance from higher-level constructions as well as analogical motivation from similar constructions. Discussing creative aspects of the construction, Silvennoinen insightfully concludes that ‘constructional innovations do not necessarily need to be formally or functionally extravagant to count as innovative, and to be subject to social evaluation and possible enregisterment by language users’.

Tanja Säily, Florent Perek and Jukka Suomela also take a sociolinguistic approach in their article, ‘Variation and change in the productivity of be going to V in the Corpus of Historical American English, 1810–2009’. The be going to V construction has been extensively studied in previous research, but the authors’ focus on productivity (in terms of both type frequency and host-class expansion), their use of cutting-edge methods, and their focus on the effect of both language-internal and language-external factors on the process of change enrich our understanding of the history of the construction. The language-external (social) factor examined in the article is gender, which the authors study with additional metadata generated with machine-learning approaches, while the language-internal factors include usages representative of the most recent and most innovative semantic extensions of the construction, expressed with mental verbs, inanimate subjects and in passive voice. The authors find that in two out of three cases (mental verbs and inanimate subjects) women use the construction more productively than men, which is in line with the findings in previous sociolinguistic research according to which women tend to adopt innovative usages more readily than men. Furthermore, a distributional semantic analysis reveals that men use verbs related to the better-established usages (e.g. caused motion) more productively than women, thus being more conservative in their usage. These results contrast with previous research on morphological productivity, where the tendency has been for a male advantage (e.g. Säily Reference Säily2014; Säily et al. Reference Säily, Hilpert, Suomela, Buschfeld, Ronan, Neumaier, Weilinghoff and Westermayer2024), which calls for more sociolinguistic research on the productivity of syntactic constructions. Future research will benefit from the types3 tool the authors have developed for analysing variation and change in productivity, as well as from the adaptation of its underlying methodology to distributional semantics.

Turo Vartiainen, Marcus Callies and Aatu Liimatta study the productivity of the Complex Modifier Construction (e.g. easy-to-use, better-than-expected, off-the-charts) in World Englishes. Focusing on Inner Circle varieties, West and East African as well as South-East Asian (SEA) varieties, the authors contribute to the ongoing debate about which factor has a greater influence on constructional productivity in World Englishes, the phase of the global variety in Schneider’s (Reference Schneider2007) Dynamic Model of Postcolonial Englishes or the typological profiles of the main substrate languages spoken in the different regions. Their data comes from the massive Corpus of Global Web-based English, and they compare the type frequency of the construction across the varieties using LNRE modelling. The authors find that the productivity of the construction is generally highest in the Inner Circle varieties, such as American and British English, which are also the most advanced varieties in Schneider’s model. However, they also discover that even though most of the African and SEA varieties are at the same phase according to the Dynamic Model, the productivity of the construction is much higher in the SEA varieties. The authors interpret this in terms of substrate influence: while the major substrate languages spoken in the African countries are mostly head-initial, favouring postmodifying structures, those in South-East Asia tend to be head-final and thus favour premodification, which facilitates the use of the Complex Modifier Construction in the SEA varieties. The authors therefore argue that language contact is a key explanatory factor affecting productivity in World Englishes, and it can even override the influence of the evolutionary phase of the variety. Methodologically, the main takeaway of the study is that comparisons of type frequencies predicted by LNRE models, which are sensitive to small differences in the distribution of the data, should be complemented by comparing the frequency spectra of the construction (how many types occur once, twice, etc.), which can also provide information on the degree of conventionalization or entrenchment of the construction.

In conclusion, the contributions to this special issue increase our understanding of constructional creativity and productivity in a number of ways. These include an enriched conceptualization of creativity, including F- and E-creativity (Trousdale & Norde Reference Trousdale, Norde, Tanja and Turo2025; Bergs & Pentrel Reference Bergs, Pentrel, Tanja and Turo2025; Laws Reference Laws, Tanja and Turo2025; Flach Reference Flach, Tanja and Turo2025), a wider focus on the semantic side of productivity (Ceuppens & De Smet Reference Ceuppens, De Smet, Tanja and Turo2025; Laws Reference Laws, Tanja and Turo2025; Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025), and rethinking what should be included in constructional descriptions and how this should be represented in the network model (Flach Reference Flach, Tanja and Turo2025; Laws Reference Laws, Tanja and Turo2025; Silvennoinen Reference Silvennoinen, Tanja and Turo2025). Furthermore, many of the studies focus on sociolinguistic and regional variation (Silvennoinen Reference Silvennoinen, Tanja and Turo2025; Säily, Perek & Suomela Reference Säily, Perek, Suomela, Tanja and Turo2025; Vartiainen, Callies & Liimatta Reference Vartiainen, Callies, Liimatta, Tanja and Turo2025) as well as on diachronic change in productivity and creativity, thus contributing to recent trends in CxG research. Finally, in addition to their theoretical contributions, the authors use and develop state-of-the-art methods that will significantly facilitate future research in the field.