Combining formal and functional approaches to variation in (morpho)syntax: Introduction to the special issue

1 The formal–functional divide and the status of variation

Although the long-standing formalist–functionalist debate has certainly been one of the most contentious issues in linguistics (and beyond), it is based on an extreme abstraction, a strong simplification of a much more complex reality. Instead of a dichotomous categorization into ‘formalism’ and ‘functionalism’, we are dealing with a continuum of more formal or more functional elements of linguistic theorizing (for an example of an alternative classification of approaches to the study of language into ‘externalism’, ‘emergentism’, and ‘essentialism’, cf. Scholz et al. 2023). It is also a history of reciprocal misunderstandings and misconceptions, which goes back at least to the 1960s or 1970s, if one focuses merely on the contemporary history of scholarship and functionalism as a countermovement to Chomsky’s unequivocally formalist generative framework. However, as Thomas (2020: 15, 28) convincingly shows, both approaches in their numerous modern versions build upon the ideas and work of many predecessors: traditions that, as to formalism, maybe could be traced back even until Pāṇini’s grammar of Sanskrit and, in the first half of the 20th century, to American scholars such as Leonard Bloomfield and the ‘post-Bloomfieldians’, whereas functionalism, having its roots rather in Europe, is strongly connected to the foundation of the Prague Linguistic Circle in 1926 by Vilém Mathesius, Nikolaj Trubetzkoy, and Roman Jakobson (cf. already their ‘Ten Theses’, published as the group’s manifesto in 1929, wherein they state: “language is a system of means of expression adapted to a goal”; Vachek 1967: 33, cited after Thomas 2020: 24). But, when attempting to give an outline of the history of science, it becomes evident as well that the attribution of many scholars and movements to one of the two camps is often doomed to fail (regarding, for example, Saussure’s structuralism or Louis Hjelmslev and the Linguistic Circle of Copenhagen, his older compatriot Otto Jespersen, and so forth; for a detailed discussion, see Thomas 2020: Chapter 2).

There are many, more or less marked differences between ‘formalism’ and ‘functionalism’ (although there seem to be less gaps between the two in phonology; cf., e.g., Carnie and Mendoza-Denton 2003; Haspelmath 2000). Both formalists and functionalists themselves may experience severe difficulties when asked to define the core of their approaches while they seem to be more confident in defining the positions of the respective other group (Curnow 2002: 506). Thomas (2020: Chapter 1 and her summary table in the appendix) gives an impressively comprehensive basic characterization of formalism versus functionalism. According to her collection, formalism (not only in linguistics) is generally associated with a focus on the internal structure and organization of the object of study, often by abstracting away from its surface appearance, and with a prioritization of form over content or context (“form transcends function”, Thomas 2020: 6). Even though the word ‘function’ itself has many facets and uses (cf. Nichols 1984), functionalism, on the other hand, is usually linked to emphasizing the role a phenomenon plays within its larger context, which determines its internal structure (“form derives from function”, Thomas 2020: 6; interestingly, the often cited motto “form [ever] follows function” goes back to an American architect, namely Louis Sullivan).

As Thomas (2020: 12–13, 32, 45, 51) rightly points out, modern functionalism is more diverse than formalism in its internal architecture (or, as Elizabeth Bates put it, “functionalism is like Protestantism: it is a group of warring sects which agree only on the rejection of the authority of the Pope”; cited after Van Valin 1990: 171): While contemporary formalism is centripetal, with a strong dominance of Noam Chomsky and his generative theory since decades, no single version of present-day functionalism can represent the whole. The development of the multiple versions of generative grammar from 1957 until today is usually subdivided into the stages of early transformational grammar (Chomsky 1957, 1965), principles and parameters theory (Chomsky 1984, 1986) and Minimalist Program (Chomsky 1995, 2000). The most important anti-Chomskyan but still generative (cf., e.g., Müller 2013: 59) varieties of formalist linguistics, from the late 1970s/1980s on, are Generalized Phrase Structure Grammar (GPSG; Gazdar et al. 1985) and, as a derivative thereof, Head-Driven Phrase Structure Grammar (HPSG; Pollard and Sag 1994) as well as Lexical-Functional Grammar (LFG; Bresnan 2001; Kaplan and Bresnan 1982); see, e.g., Thomas (2020: Chapter 3) and the literature cited therein for a first overview. Some exponents of the much more heterogeneous “galaxy of functionalism” (Graffi 2001: 389), on the other hand, are Lakoff’s (1963/1976, 1971) counterconcept of ‘generative semantics’, Halliday’s (1970, 2002 systemic functional linguistics, Dik’s (1978, 1989 functional grammar, Givón’s (1979, 1995, 2001 functionalist work, Foley and Van Valin’s (1984) Role and Reference Grammar, Hopper’s (1987) Emergent Grammar, Langacker’s (1987, 2008 cognitive linguistics, Bates and MacWhinney’s (1989) Competition Model, Goldberg’s (1995) Cognitive Construction Grammar, Croft’s (2001) Radical Construction Grammar etc. (see Thomas 2020: Chapter 4 for an outline of a selection thereof). Note that although many ‘functionalists’ accept both communication and cognition as fundamental influence factors, one could subdivide the approach into functional versus cognitive linguistics (cf., e.g., Siewierska 2013; for other classifications of modern functionalistic accounts, cf. Nichols 1984, who distinguishes between conservative, moderate, and extreme functionalism in her typology, as well as Newmeyer 1998, who separates ‘formal functionalists’ and then identifies external, integrative, and extreme functionalists as subgroups of functionalism proper).

Formalist frameworks are typically ‘system-based/rule-based’, whereas functionalist accounts are ‘usage-based’ (cf., however, Adli et al. 2015 for a more nuanced perspective, pleading rather for a continuum between system- and usage-based, with intermediate positions). Furthermore, formalist approaches are said to pursue ‘internal’ explanations, while functionalist approaches strive for ‘external’ explanations (for a concise overview of the discussion about what counts as an explanation in formalism vs. functionalism, cf. Thomas 2020: Chapter 6.3.1). Formalists thus focus on language as a system, shaped by (implicit) structural rules, constraints, principles etc. (whose exploration and explication are considered as the most important task of linguistics); their priority lies on theory construction and on attaining generalizations, characteristically by looking inward, i.e. at the inner structure of language (like a linguistic ‘engineer’, using Thomas’s term). A frequent criticism related to this approach concerns the attempts, in some cases, to integrate and explain new empirical facts merely theory-internally, on the basis of previously established theoretical concepts/constructs. In contrast, functionalists emphasize the central role of language used as a communicative tool (putting their efforts in the exploration of the relationship between language forms and functions); they prioritize language data and detailed records of its use in context (like a linguistic ‘collector’ or ‘curator’ assembling ‘cabinets of curiosity’, according to Thomas 2020: 111) and, normally, they search for explanations outside language (“under the assumption that languages are what they are because of the exigencies of human communication and cognition, or because of the external cultural environment in which language is used”; Thomas 2020: 6). Their focus on particularities and idiosyncrasies in combination with an extensive citation of language data earned them the criticism that, in some circumstances, data takes up more space than the actual analysis (Thomas 2020: 57, commenting on Givón 2001).

Some further, albeit partly too general, oversimplifying ascriptions to (generative) formalism, most of them gathered by and taken from Thomas (2020: appendix), comprise its understanding of language as a vehicle for thought/reasoning and of the ‘language faculty’ as a window into human nature, its assumption of a Universal Grammar (innateness hypothesis), its emphasis of autonomy and modularity in the organization of grammar (cf. the three autonomy hypotheses discussed by Newmeyer 1998: Chapter 2), the central role of linguistic competence, its openness to speculation, its predilection for abstraction (making use, e.g., of complex notational systems for the sake of a maximum of clarity and precision in the formalization of findings), its recourse to idealized, often engineered linguistic data^[1] instead of representing actual speakers’ actual usage (according to Fischer 2007: 54, for formalists “the system of grammar [is] more important as an object of study than the actual language data”), its preference of deductive methods, its pursuit of a rather narrow range of grammatical facts, its concentration on the sentence as the typical unit of analysis, the use of grammaticality/acceptability judgments as empirical evidence (i.e. native speakers’ intuition and introspection), and the formulation of generalizations in absolute terms. In comparison (and in addition to the traits already mentioned above), functionalism has been characterized, sometimes in an overdrawn manner as well, by its fundamental understanding of language as performing social-interactive functions, its principled rejection of UG (in favor of learning, ‘nurture’ instead of ‘nature’), its strong believe that all parts of language are integral to each other, its affinity to linguistic performance, its commitment to analyzing real language data and representing diverse data, its use of inductive methods, its exploration of the totality of language and neighboring phenomena, highlighting discourse as the level of analysis, its preference for deep analyses of particular cases (in the form of case studies, using corpora etc.), the formulation of generalizations as statistical probabilities, and many more (for these and other characteristics, summarized in the form of bullet points in her appendix and discussed in detail throughout her monograph, see Thomas 2020). Notice that this rather simplistic presentation is for illustrative purposes only.

One of the central motivations of the present volume is the question whether such black-or-white attributions are appropriate and up-to-date or if we are rather dealing with a continuum of formal and functional, as I had already suggested at the beginning of this introduction: Is ‘formal(ist)’ really identical to system-/rule-based, theoretical approaches, using introspection as the main method of collecting data, abstracting away from variation?^[2] And is ‘functional(ist)’ then equivalent to usage-based, using quantitative-empirical and statistical methods, assuming mostly language-external causes for change and variation etc.? First, I want to stress that, despite the fact that the methods applied to study language(s) can (and, depending on the respective subdiscipline and research question, even need to) be quite diverse, theoretical modeling of any kind must be empirically well-founded. At least today, most formal-linguistic studies are based on questionnaire surveys, interviews with informants and/or the analysis of large language corpora. Second, a frequent misunderstanding about Chomsky’s (1965) ‘ideal speaker/hearer’ has caused a lot of incomprehension on the part of non-formalists. It is indisputable that collective (interpersonal) as well as individual (intrapersonal) variation or heterogeneity is an essential feature of every natural language, and the concept was merely thought of as an (at least initially) necessary abstraction. Different linguistic subdisciplines are concerned with different kinds of variation from different perspectives (cf. also Weiß, this volume). Today, for example, dialect syntax constitutes a well-established and very lively field of research, both within and outside of dialectology. Notice that it was generative syntax that, in the late 1980s/1990s, contributed with decisive impulses to the field. The growing interest in dialect syntax (cf., in great detail, D’Alessandro et al. 2010; Poletto 2000; Weiß 1998, 2004, 2017; Weiß and Strobel 2018) has to do precisely with the more and more central role of variation and the opportunity dialects offer to investigate intralinguistic microvariation, “the minimal units of syntactic variation” (Kayne 1996: xiii), as well as their localization within the language system (lexicon, syntax-phonology interface etc.). Further reasons highlighted by Weiß (inter alia, 1998 concern the limited reliability of data from standard languages (subjected to normalization/codification and systematic instruction), i.e. the fact that dialects are more natural than their standard(ized) counterparts in terms of first language acquisition, and the general advantage of significantly extending the data basis by including dialects (since many phenomena occur only in the dialects but not in the standard varieties of the respective languages).

Only recently, the view has emerged that both formal and functional perspectives are necessary to account for the full range of variation we can observe in language (‘hybrid’ models of language; see, e.g., the contributions in Adli et al. 2015). For a long time, however, formalist and functionalist approaches had been regarded as complete opposites in linguistic theorizing, with neither of the two approaches considering the other’s perspective in their work.^[3] Famously, Haspelmath (2000) titled his review of Newmeyer (1998) Why can’t we talk to each other? Indeed, formalists and functionalists may sometimes even insult each other as ‘fuzzies’ and ‘symbol pushers’, respectively, like in the case of Newmeyer’s fictitious characters Sandy Forman (MIT), a formalist, and Chris Funk (University of California at Santa Barbara), a functionalist. However, while Newmeyer’s imaginary dialogue does not have a positive conclusion, a later one by Carnie and Mendoza-Denton (2003), on occasion of their “interactive review” of Darnell et al. (1999), does: In the course of an invented conversation, where both play themselves (i.e. Carnie a “junior formal syntactician” and Mendoza-Denton a “junior variationist sociophonetician”), they discover a certain amount of compatibility (see also Thomas 2020: Chapter 6.2.1). So, is the formal–functional debate above all a/the “great(est) rhetorical divide/conflict” in linguistics (Newmeyer 2016: 129; cf. also Bülow and Vergeiner, this volume)? Most interestingly, in her survey of formalist and functionalist literature on their respective image of the other, Thomas (2020) proves that both camps seek to assimilate/reconcile the ideas of the other into/with its own positions (cf. her Chapter 6.2.2: How formal is functionalism?, alluding to Anderson’s 1999 approach of A formalist’s reading of some functionalist work, as well as her Chapter 6.2.3: How functionalist is formalism?, where she cites e.g. Bates and MacWhinney’s 1990 title of their commentary on Pinker and Bloom 1990, namely Welcome to Functionalism). In a recent paper, Bošković (2021/2022) assigns typology the role of “setting grounds for a potential rapprochement of the functional and the formal approach to language”.

There is no systematic investigation into the full range of grammatical phenomena and the extent to which formal and functional theorizing complement or even contradict each other. Thomas (2020: Chapter 5) compares formalism and functionalism with respect to some well-chosen long-running issues of linguistics concerning two syntactic phenomena as well as the ontogeny and phylogeny of language (word order, transitivity; first language acquisition; the origin of human language). Summarizing the range of positions, she reaches the conclusion that “formalists and functionalists sometimes study different language phenomena. […] [T]hey sometimes examine the same language phenomena but perceive them differently. In addition, formalists and functionalists sometimes approach the same language phenomena with different questions in mind” (Thomas 2020: 67–68). While she notices a significant gap and non-complementarity between formalism and functionalism in their accounts of word order, there is a certain overlap when it comes to transitivity, despite the notable contrast, and, in her opinion, functionalism “builds beyond” formalism here (Thomas 2020: 78–79). As for the debates about language acquisition and the origin of language, she sees an irreconcilable opposition/competition, a “fundamental incompatibility” (Thomas 2020: 79). The objective of the present volume therefore is to contribute to the following central questions: Is formal(ist) and functional(ist) theorizing complementary or even contradictory to each other, is one of them “redundant” as they both lead to the same explanation, or do the two perspectives treat two completely different aspects of language? Do we get different or similar results if we approach one and the same phenomenon from different (theoretical) perspectives? In particular: Can we explain linguistic variation using both perspectives, and what do we gain from this?

2 Contents of this volume

A large number of studies that assume a ‘hybrid’ approach investigate variation in English (and varieties thereof). Other Germanic varieties (including dialects) have remained under the radar so far. The present special issue makes an important contribution to filling this gap. Therefore, it brings together a total of seven papers that aim to explain (morpho)syntactic variation in other Germanic varieties than English (mainly German, Dutch, and Swedish) by uniting different formal(ist) and functional(ist) perspectives as well as, not least, by including also semantic and phonological factors. The diverse models and accounts covered in this volume include different versions of generative grammar, information-based morphology, Construction Grammar/Construction Morphology, natural morphology, and sociolinguistics. As it turns out, ‘formal’ and ‘functional’ are less different than assumed, and often it is rather the terminology that differs (cf., e.g., Reiner), while the underlying aim of describing and analyzing variation is essentially very similar.

Helmut Weiß’s contribution How to explain linguistic variation and its role in language change? examines the interplay between variation and language change in terms of cause and effect. His central claim is that variation can be the result of language change^[4] and that the reverse is also possible, i.e., that variation can trigger language change. The first scenario is illustrated by the pronoun cycle (the relation between variation in the strength of pronouns and their relative word order), the second one is exemplified by the development of prepositions into complementizers (e.g., German seit ‘since’). Weiß argues that a better understanding of the importance and role of variation in language change can only be reached if a proper distinction is made between the emergence of a variant (innovation/actuation) at the level of the individual and the diffusion (propagation) of a variant within the speech community. He maintains that the latter is treated by sociolinguistics (because the selection or replacement of a variant has often to do with the speaker’s identity, prestige factors, adaptation to the interlocutor(s), and so forth), whereas the former can only be investigated by formal linguistic approaches that posit abstract underlying structures (taking a look ‘under the surface’). In addition, Weiß claims that these two facets of language change proceed along different paths: While the emergence and further development of new variants often proceed in a cyclic fashion, the diffusion of an innovation typically can be described by an S-shaped curve (similar to the spreading of a mutation in a population). The paper thus does not only provide a discussion of the connection between linguistic variation and language change, but it also touches upon the division of labor between formal and sociolinguistic approaches to language change (see the paper for the reasons why sociolinguistics, as the study of the relationship between language and society, represents a type of functional linguistics). Weiß concludes that the two approaches are concerned with different aspects and hence complement rather than contradict each other: While grammar in a narrower sense (morphosyntax) changes in the way determined by principles expressed within formal theories (e.g., economy principles) and these principles explain why and how grammar changes, sociolinguistics can explain sociocultural aspects of variants as well as, for example, fluctuations in the rates of change (succession of stability and flux).

In their paper On the asymmetry of wh-doubling in varieties of German and Dutch, Gisbert Fanselow, Sjef Barbiers, Jessica M.M. Brown, Natasja Delbar, Sophia Nauta and Johannes Rothert present an empirical study on speaker variation concerning a special kind of wh-doubling (wh-copying) where one copy of the wh-phrase is more complex than the other. Provided that wh-doubling with simplex wh-words is accepted at all in the investigated varieties, the authors explore the question of which of the two extended versions is the preferred or only possible one: (i) wh … whXP (right-complexity), or (ii) whXP … wh (left-complexity). The aim of the study is to provide empirical evidence for the theoretical claim that the higher copy should be less complex, a claim that follows from assumptions of the copy theory of movement. In contrast to the prediction, their judgment experiments on the respective acceptability of the two versions, interpreted by a thorough statistical analysis, shows that both are in fact acceptable and that this can, but need not, be taken as a difference between dialects or groups of speakers. It would be interesting to search for authentic data as a further kind of evidence. The authors’ structural analysis of left-complexity constructions is based on the assumption that the more complex copy on the left did not get there by movement but was base-generated in a high dislocated position. Eventually, they discuss whether a copy and deletion approach would favor right-complexity and/or exclude left-complexity at all, focusing on two different versions of copy and deletion. As an outlook on a possible functional perspective and future research on language processing, Fanselow, Barbiers, Brown, Delbar, Nauta and Rothert point to the question whether left-complexity is easier to process than right-complexity or vice versa. Given the potentially lower processing load in the case of left-complexity constructions, where the full constituent can be identified immediately, left-complexity would be expected to be preferred over right-complexity. This does not only contradict the finding that grammars with a preference for left-complexity do not exist, but also the observable opposed tendency in natural language for heavily loaded information to appear later in the sentence.

Jackie Nordström’s article Semantic agreement and the Dual Model of Language deals with several agreement phenomena in, above all, Swedish (but, e.g., also in English and Russian) and the question of how these can be accurately modeled. More specifically, the paper concerns the problem of formal/grammatical and semantic agreement mismatches in Swedish predicative and passive constructions (among others, also so-called ‘pancake sentences’). Notice that in such constructions, the predicative adjective or passive participle in Swedish agrees with the subject noun phrase in gender and number. Nordström’s main point is that agreement is more than ‘narrow syntax’. As is widely known, in English a collective or committee noun in subject position allows the verb in both singular and plural (e.g., The committee is/are deciding on a solution). There is no doubt that semantic agreement, i.e. the fact that the verb can agree with the referents of the noun rather than with its grammatical number, poses a problem for any account built on the assumption that agreement between an argument and a verb/adjective is a purely syntactic operation, without semantic repercussions. The author argues that the simple ontology of interpretable–uninterpretable features assumed in the Minimalist framework is not adequate for handling such phenomena, and she sketches alternative analyses showing that agreement can and sometimes even must be given a functional-semantic explanation. She reaches the conclusion that person-number-gender affixes on verbs and adjectives are semantically interpretable even in languages such as English and Swedish because the PNG affixes convey meanings not overtly marked by the corresponding DPs. Furthermore, Nordström integrates her contribution into “a larger project that attempts to assign semantic functions to seemingly purely syntactic phenomena” and into her work on the ‘Dual Model of Language’ (Nordström 2022), which “offers a way of bridging the gap between the so-called formalistic and functionalistic grammars”, as against the traditional “syntactocentric” Y-model of generative grammar.

Tabea Reiner’s paper A constructionist analysis of gapping against the background of generative analyses illustrates how a comparison between a functional (constructionist) account and a formal (generative) account of the same phenomenon, namely gapping as one type of ellipsis (e.g., Gonzo ate the peas, and/but Lola [ate] the carrots; Aelbrecht 2015), can be conducted fruitfully. More precisely, Reiner compares mainly two analyses of gapping taken to be representative of a functional and of a formal approach, respectively: the theory of Goldberg and Perek (2019), which belongs to Construction Grammar (CxG), and the copy theory of Repp (2009a, b), which is situated within the Minimalist framework (MP). Gapping is an extremely well-described phenomenon with a vast literature on it, so that a selection was inevitable. After a general outline of the phenomenon in as theory-neutral terms as possible, Reiner shows in a detailed and equilibrated discussion that both accounts have their merits and potential shortcomings in the analysis or prediction of a number of empirical observations about gapping. Instead of arguing for one account or the other, she concludes that the functional and formal approaches have certain overlaps and that the main difference between them lies in the role and nature of (falsifiable) predictions they (can) make. While, contrary to a frequent assumption, both types of approaches have predictive power and provide generalizations, the success of the concrete predictions is shown to vary between them. From a more general perspective, Reiner highlights that formal approaches or MP make more far-reaching predictions that can be tested cross-linguistically, whereas functional approaches or CxG make language-internal predictions. She argues that cross-linguistic predictions are challenging for MP because their validation often is not successful and that CxG, in turn, faces the problem that basing predictions for individual languages merely on “communicative pressures plus conventionalization” is “so broad that it is vacuous”. Reiner concludes that the two approaches only compete within the small domain of anticipating novel data in well-described languages.

Oliver Schallert’s contribution Number fission from a formal and functional perspective provides a new view on the morphology of German modal verbs. Unlike in the present singular, they show phonologically unconditioned umlaut in the present plural (though not only there), with unclear functional motivation. The distribution of this umlaut and other morphological irregularization processes such as contractions and/or consonant mutations in German modal verbs (with a special focus on the two modals müssen ‘must’ and können ‘can’) is investigated by a thorough empirical analysis based on a very fine-grained dialectological dataset of 308 dialect grammars in total. The theoretical claim of the paper is that, synchronically, the observable number split within the paradigms of modals is no transcategorial number marking (as has been suggested in the literature) but a morphomic pattern, signaling nothing more than inflectional class coherence (‘distinctiveness’), since, in a substantial number of dialects, umlaut has been generalized to the infinitive or even to the past participle. Diachronically, however, for the first step of analogical extension of umlaut as a verbal plural marker, Schallert accepts the functional motivation of transcategorial umlaut (which resorts to analogy as a cognitive factor). In the modeling part, the author shows that the morphomic alternation can be captured by the formal word-based realizational model of Information-based Morphology (IbM), which was developed in the context of Head-Driven Phrase Structure Grammar (HPSG) and, due to several similarities, can be carried over to functional theories such as Construction Morphology (CM). Therefore, according to Schallert, “there is the chance of integrating the intuitive plausibility of functional explanations while maintaining a certain level of independence of morphological structure”. Umlaut alternations are then modeled on the level of stem hierarchies, not on the level of features: “[T]he morphology of umlaut has more to do with the organization of stem spaces than with feature signatures.” This approach disentangles the synchronic state as reported by the dialectological sources, and the diachronic spread of umlaut (which is corroborated by the hierarchies underlying the quantitative findings).

Lars Bülow and Philip C. Vergeiner adopt both formal and functional perspectives in order to explain recent empirical findings on a quite well-researched topic of verbal mood within their paper Explaining morpho-syntactic variation and change: The case of subjunctive II in the Bavarian dialects of Austria. The phenomenon has been dealt with in a number of studies from various angles and using different data from different time periods, proving that not only grammatical but also areal and social factors play a significant role. It is well-known that in the non-standard varieties of German in Austria, subjunctive II can be expressed both by synthetic and analytic means, but that the synthetic variants are being replaced by the analytic ones, especially by the würde-form. While the periphrastic variant with würde ‘would’ is quite generally and continuously on the rise, this is accompanied by a decrease in the frequency of use of the different synthetic forms as well as of the täte-periphrasis. Within synthetic formation, on the other hand, various -at variants (e.g., weak sōgat ‘[I, he/she/it] would say’ or mixed kamat ‘[I, he/she/it] would come’) are clearly preferred over strong forms (such as kam ‘[I, he/she/it] would come’). Previous research showed also that while synthetic structures are still widespread in the rural areas of Austria, they seem to be pushed back by the analytic ones in urban areas (Vienna and Graz). Building upon the results of such predominantly descriptive studies, Bülow and Vergeiner attempt to explain the present variation and change in the forms of subjunctive II in Austria from a (more) functional and (more) formal perspective, namely natural morphology (NM) and constructional morphology (CxM). Notice that, referring to Newmeyer (2003, 2016, they point out that “a formal explanation is not required to draw on a theory generally considered as formal, such as generative grammar” and that “[a]ccordingly, formal explanations can also be found in theories that are usually understood as functional”. They explicitly state that “this does not mean that [they] consider CxM a formal theory”. By uniting the two approaches in a complementary way in their theoretical model, the authors address the research desideratum of more theoretically grounded reflections on the phenomenon under investigation. In particular, they argue that NM can explain the extension of the -at suffix to strong and irregular verbs and that the application of key NM principles (constructional iconicity, uniformity, transparency) to the CxM framework allows to explain the spread of the analytic/periphrastic subjunctive II as the most iconic construction available and the easiest one to be acquired especially in urban areas that are characterized by language contact.

In their case study A modal account of syntactically non-integrated von wegen in contemporary German, Manuela C. Moroni and Ermenegildo Bidese address the question of how non-prepositional (thus, non-causal) instances of von wegen can be classified and integrated into Abraham’s (2020) theory of modality. Moroni and Bidese’s analysis is based on a total of 186 occurrences of von wegen in the two largest digital corpora Forschungs- und Lehrkorpus Gesprochenes Deutsch (FOLK) for contemporary spoken German and Deutsches Referenzkorpus (DeReKo) for written German. In addition, they take into account some historical attestations. The authors distinguish between an “illustrating/exemplifying” and an “opposing” type of non-prepositional von wegen in present-day German. Taking the classification of modal expressions based on evidentiality and epistemicity as a starting point, they show that both types encode a quotative meaning but only the second one exhibits an epistemic value. From a diachronic point of view, they argue that the illustrating use developed directly from the original preposition of pertinence (‘in regard to’), while the opposing use emerged from a contrastive accent on von wegen indicating that it relates to a proposition that is inappropriate or in contrast to the given context. Moroni and Bidese analyze the opposing von wegen as a non-canonical expression of epistemic modality (in the sense of Abraham 2020) and thus show that syntactically non-integrated constituents are able to express modality, which they identify as a third strategy of modalization (beyond lexical and grammatical modality expressions) that operates at the discourse level (rather than at the propositional level). Their paper therefore contributes to a more comprehensive theory of modality. Moreover, it demonstrates that an integration of non-prepositional von wegen into Abraham’s formal theory, where modality resides in the sentence structure and in a formal system of means of expression, can succeed if it is expanded to include a functional perspective, according to which modality can also arise through the interaction of semantico-pragmatic factors and the focalization of (opposing) von wegen. Following Axel-Tober and Gergel (2016) versus Aijmer (2016) and Newmeyer (2010, 2017, Moroni and Bidese’s understanding of formalism in modality research is based on grammaticalized expressions of modality (as described in Abraham’s account), whereas a functional view, according to them, starts from the semantics of modality and explores its different means of expression depending on the context, speech activities and text types involved.

In sum, all papers convincingly show that both formal(ist) and functional(ist) elements of theorizing can make a valuable contribution to explaining (morpho)syntactic variation, in synchronic as well as diachronic respects. The integration of ways of looking at a certain phenomenon or problem from the respective other camp is proven to provide a significant added value and should not be excluded a priori. An important precondition in order to increase intradisciplinary transparency and comparability is, of course, a shared (basic) terminology and as theory-neutral descriptions of the linguistic phenomena in question as possible. An eclectic approach that combines different perspectives has several advantages: Formal and functional (e.g., sociolinguistic) accounts may be concerned with distinct aspects of a phenomenon and hence complement rather than contradict each other; there may be a division of labor (cf. Weiß). Phenomena where not only grammatical but also areal and social factors play a crucial role (e.g., the variation and change of subjunctive II forms in Austrian dialects) even must be explained in a complementary way from both a formal and a functional angle (see Bülow and Vergeiner). Formal (generative) and functional (constructionist) models may show overlaps when it comes to one and the same phenomenon (cf. Reiner on gapping); both types of approaches provide generalizations and exhibit (a more or less successful) explanatory/predictive power. Similarly, specific language peculiarities (such as umlaut alternations in German modal verbs) may be captured both by formal and intuitively plausible functional theories, due to similarities between them (cf. Schallert). On the other hand, potential contradictions (e.g., between grammar theory, empirical findings, and different principles of language processing) must be reconciled in favor of a coherent explanation (as indicated by Fanselow, Barbiers, Brown, Delbar, Nauta and Rothert). As far as the interaction of linguistic subsystems is concerned, certain seemingly purely syntactic phenomena (such as agreement) can and sometimes even must be assigned a functional-semantic explanation (see Nordström). And the integration of non-canonical means of expression (e.g., German non-prepositional, opposing von wegen) of a certain category (here: modality) into an existing formal theory of grammaticalized expressions can succeed if one includes a functional perspective (such as the interaction between semantico-pragmatic factors and focalization), leading to a more complete theory of the investigated category (see Moroni and Bidese).