Paradigmaticity and obligatoriness of grammatical categories

1. Introduction

The papers collected in this volume are the outcome of the presentations and discussions of the workshop “Paradigmaticity and obligatoriness: grammatical categories across languages” held at the 42nd Annual Meeting of the Societas Linguistica Europaea at the University of Lisbon, 9–12 September 2009, which united researchers working in the domains of grammar, morphology, typology, and grammaticalization, as well as historical linguistics.

The major objective of this gathering was a revision of the concepts of grammatical paradigm, paradigmaticity and obligatoriness and related issues. Although well-known and long-standing terms of grammatical description, they are by no means satisfactorily defined, nor are they sufficiently taken into account in most modern linguistic theories, except grammaticalization studies. In this latter branch of linguistic research, obligatoriness and paradigmaticity are treated as central concepts that are employed to solve some of the most pressing problems in grammaticalization studies. Therefore, it is time to take a fresh look at these notions, to check accepted definitions, to examine their interrelations with other relevant concepts, such as, for example, “constructions”, and to re-evaluate their possible applicability to theoretical as well as empirical questions in the field of grammatical categories and their evolvement under grammaticalization.

By collecting evidence about grammatical paradigms (in a broad sense of the term) in a number of languages and by comparing the results of synchronic and diachronic studies, the papers presented in this volume explore to what extent the notions of obligatoriness and paradigmatic organization may be applied to grammatical categories and to processes of language change connected with grammaticalization. Among the issues receiving particular attention in the contributions are the following:

1. the importance of the notions of paradigmaticity and obligatoriness for defining grammar;

2. the treatment of periphrastic forms/constructions in paradigms or as paradigms;

3. the correlations between paradigmaticity and obligatoriness;

4. the motives and conditions (structural, communicative, cognitive, etc.) as well as the successive steps of the development of paradigmatic structure in language-specific instances of grammaticalization;

5. regularities in the diachronic development of grammatical paradigms.

All the papers collected here, beside their specific topics, take up more than one of these issues. The contributions by Lars Heltoft, Lene Schøsler, Jens Nørgård-Sørensen and Silvia Luraghi pay a great deal of attention to questions 1 and 3. Question 2 is treated extensively by Lars Heltoft and Lene Schøsler, as well as by Liselotte Brems and Kristin Davidse and in the paper by Christian Lehmann, José Pinto de Lima and Rute Soares. The last mentioned two papers, as well as the one by Silvia Luraghi, discuss points 4 and 5 extensively with respect to the particular linguistic items treated. Thus, this volume presents a multifaceted and at the same time highly focused discussion of these issues, which yields a fresh evaluation of the notions of paradigmaticity and obligatoriness, as well as a couple of in-depth case studies of several languages and several grammatical categories.

Before turning to the specific questions and arguments of each contribution sketched in the third part of this introduction, the following section offers some general reflections on the notions under discussion, i.e. on grammaticalization, paradigms and obligatoriness and their connection, as well as on the concept of constructions.

2. Some central concepts

Paradigmaticity and obligatoriness figure prominently in grammaticalization theory. This becomes most obvious in Lehmann's parameters of grammaticalization, which are an indispensable instrument for comparing and measuring degrees of grammaticalization. As Table 1 below shows, it is parameters 3 and 5 which explicitly refer to paradigmaticity and obligatoriness, here under the label of paradigmatic variability. Although the parameters are widely known, some brief explanatory remarks may be useful.

The fundamental property relevant for breaking down the parameters is the autonomy of a sign, which by definition is high in lexical and low in grammatical signs. The relative autonomy of an item is measured on the syntagmatic and paradigmatic axis of language, and in relation to the three domains of weight, cohesion and variability (which leads to the six parameters given in Table 1). The weight of a sign concerns its distinction from the other members of its class. The more distinct it is, the less it is grammaticalized. The concept of cohesion captures relational aspects of a given sign and its neighbouring signs, i.e. the degree of dependence and coherence, which increases in grammaticalization on the syntagmatic as well as the paradigmatic axis. Variability, finally, refers to the mobility or shiftability of a sign with respect to other signs; it generally decreases with increased grammaticalization.

The two parameters relevant here are parameters 3 (paradigmaticity) and 5 (paradigmatic variability). Lehmann gives the following definition, contrasting paradigmaticity with its opposite, paradigmatic variability:

Although the parameters will not be discussed here in detail, one thing should be noted in this context. It is obvious even from the sparse description given above that the paradigmatic axis plays a double role in the structuring of grammatical categories. First, it functions as one of the two overall, general organizational principles of language; this is expressed in the juxtaposition of the syntagmatic and the paradigmatic perspective in each of the three domains. Second, the paradigmatic perspective is an essential feature of two of the parameters themselves, namely parameters 3 and 5 as mentioned above. This is an expression of the fact that paradigmatic structuring is an essential defining criterion of grammatical signs (and not just a general structuring device of language). It is this fact that is expressed in the concept of grammatical paradigm that has been canonical at least since Roman Jakobson. As this seems to be neglected occasionally in modern theorizing, it is worthwhile rendering this definition of a grammatical paradigm again.

Following this traditional view, a grammatical category by definition consists of a paradigmatic opposition between at least two elements. One of these (typically, the newly grammaticalizing one) constitutes the formally and notionally marked element, which is cast in opposition to the formally and notionally unmarked zero-element (which, in addition, is used as the neutralization stage of the opposition). Therefore, if any form/construction is grammaticalized, it by definition builds an oppositional pair with another item and, in virtue of this, is a member of a paradigm. In short: the paradigmatic integration of a sign refers to the fact that the members of a grammatical paradigm, which as a whole is constituted by a categorical value (the semantic and/or functional content of the paradigm), are closely linked to each other, and very often form intracategorial subgroupings. This is to say that the fact of being a member of a paradigm is the central feature and the very essence of the meaning of a grammatical sign. Paradigmatic structuring, i.e. defined, regulated and substitutive association, is part of the meaning of grammar.Footnote1

In recent years, the centrality and universal applicability of the notions of paradigmaticity and obligatoriness for grammar and grammaticalization have been challenged and relativized in various directions. Himmelmann (Citation2004) takes the position that the relevant criterion for analysing an observed change as grammaticalization is “context expansion” (i.e. “that the semantic-pragmatic usage contexts of the construction at hand have been expanded”), whereas “changes on the element-level (in particular erosion and fusion but also paradigm formation) are here considered epiphenomena which, among other things, depend on basic typological features of a given language” (Himmelmann Citation2004, 33, emphasis added).

The process that Himmelmann (Citation2004) calls “context expansion” is undoubtedly essential for grammaticalization, and it is captured with precision in Lehmann's grammaticalization parameters (see in particular parameter 5). Himmelmann's evaluation of paradigmatic restructuring as peripheral, however, is to be refuted, and seems to be based on a too narrow a definition of “paradigm” (cf. Diewald Citation2010, for an extended discussion).

Although Himmelmann's approach is unjustifiably restrictive, the suggestions put forward by Wiemer and Bisang (Citation2004, 4) proceed in the opposite direction. In an approach close to Hopper's concept of emergent grammar, the authors argue for understanding grammar “as a system of more or less stable, regular and productive form-function mappings”, which also means that “the field of grammaticalization in the above sense of a broader perspective is to be extended to all the processes involved in the diachronic change and in the emergence of such systems”. Here, the notion of paradigm as an inherent semantic and/or functional component of grammatical signs seems to be totally dissolved, with only the aspect of the general paradigmatic structuring remaining (see above).

In light of the enormous amount of knowledge gathered in grammatical research and grammaticalization theory, it is only sensible to opt for an intermediate position here that allows for gradience in the phenomena discussed. Thus, we follow authors such as Lehmann (Citation1982/1995) or Plungian (Citation1998, Citation2000/2003), who suggest treating both properties as gradual phenomena exhibiting different degrees of realization; that is, grammatical categories can form more or less clear-cut paradigms, and the choice among their members can be more or less obligatory. Lehmann (Citation1982/1995, 12), for example, shows that obligatoriness is a useful criterion, although it is not “an absolute one”, as “Something is obligatory relative to the context; i.e. it may be obligatory in one context, optional in another and impossible in a third context”. Among his examples are the different degrees of obligatoriness of the category of number in nouns in Latin vs Turkish. For the sake of clarity, the full quote is given here:

Diewald (forthcoming) tackles the phenomenon of gradient obligatoriness by distinguishing between two subtypes of obligatoriness: language-internal obligatoriness and communicative obligatoriness. Language-internal obligatoriness is steered language internally and is thus subject to formal triggers (such as concord, case selection, etc.). It is the most routinized, automatized part of grammar, which tends towards extremely abstract oppositions, and in the end may lead to the loss of independent grammatical distinctions (e.g. concord rules in noun phrases (NPs) in German). The term “communicative obligatoriness” captures the fact that many categories are obligatory in the sense that they have to be realized in the relevant position; that is the speaker cannot leave them unspecified if he or she does not want to produce incorrect utterances, but the choice among the paradigmatic members of the category is not determined by language-internal features but by the communicative intentions of the speaker. In diachronic grammaticalization terms, these typically represent more recent processes of change with less advanced degrees of grammaticalization.

Another important issue to be at least briefly addressed, as it is discussed in several papers, is the question of what types of formal realization or, in other words, what types of construction are valid realizations of grammatical categories, i.e. members of a paradigm? Opinions on this vary widely and cover the full gamut between strictly gram-based and highly inclusive, pattern-based models. Although it is undisputed that not only bound forms, but also periphrastic constructions may be members of grammatical paradigms (e.g. periphrastic tense constructions forming a paradigm with inflectional realizations), opinions diverge on the question of whether sets of constructions should be regarded as a paradigm even if there is no inflectional partner available. The issue gets even more complicated in those instances where there is no linguistic item that might be identified as the core of the construction and thus the carrier of the grammatical meaning, but where the construction holistically represents a paradigmatic value. Again, Himmelmann (Citation2004, Citation2005) takes a restrictive stance and excludes constructions lacking a distinguishable grammaticalizing element, for example the topological marking of grammatical distinctions (Tim calls Tina vs Tina calls Tim). This is stated in the following quotations:

Without going into the consequences of this type of restriction, which in its core is highly circular, it must suffice to state that this view is not supported here. Instead, the position favoured in this paper towards this matter, as well as in the other contributions to this volume, conforms to the constructional approach and allows for a holistic representation of grammatical meaning over a complex linguistic string.Footnote2

These topics, among others, are taken up and discussed with reference to linguistic data in the papers collected here. The following section gives an overview of their contents.

3. Contributions of the individual authors

This volume presents six original research papers that all deal with paradigmaticity and obligatoriness as crucial features of grammatical structure. Most of the papers investigate grammatical elements and constructions from a diachronic point of view, thus focusing on the dynamic concepts of paradigmaticization and obligatorification. The languages investigated include Danish, Old Scandinavian, Latin, French, Ancient Greek, Old Russian, English, and Portuguese.

The first three papers, by Lars Heltoft, Lene Schøsler and Jens Nørgård-Sørensen, are closely connected to each other in that they use a unified model of a paradigm. This model consists of a closed set of obligatory options that alternate in a slot which is defined by a “syntagmatic domain” and a “semantic frame”. On the expression side of a paradigm is the syntagmatic domain, i.e. the syntagmatic string and the set of forms involved. On the content side of a paradigm is the semantic frame, within which the content of the specific members of the paradigm is defined. In the papers, the model is applied both to represent the synchronic status of a particular paradigm in a language as well as to reconstruct diachronic changes within and between the paradigms of a language.

Lars Heltoft's paper “Paradigmatic structure and reanalysis: From NPs to DPs in Scandinavian” unites a theoretical discussion of the notion of paradigmatic structure with a detailed case study of the development of definiteness and determiner phrases in Scandinavian. The first part reflects on paradigms as part of grammar and grammaticalization theory. Relying on the assumptions of Danish Functional Grammar, according to which syntax organizes meaning, the author argues that paradigmatic grammatical structure is found not only in morphology, but also in syntax, i.e. in word order as well as in constructional syntax. He argues for a dynamic concept of structure; i.e. structure is seen as a precondition for usage, and at the same time as maintained through usage. According to this view, new structure usually emerges through reanalyses that normally lead to structures deviant from dominant reconfirming ones. Moreover, it is argued that paradigmatic change through reanalysis always involves semantic change. The case study presented in the paper explores the rise of Scandinavian determiners and determiner phrases during the late Middle Ages, using the text material from the manuscript of Sjælens Trøst (Consolatio animae) from around 1425. Three synchronic stages are detected, each of them characterized by a specific paradigmatic organization principle: stage one (Old Scandinavian), stage two (Consolatio animae) and stage three (Post-reformational Danish). It is argued that the modern language has developed an obligatory inflectional and/or constructional contrast between definite and indefinite NPs that was absent at stage one. The modern paradigm is heterogeneous compared with the old one, as the determiners are morphologically different. The modern paradigmatic oppositions between different syntagmatic types of constructions combined with inflectional subparadigms constitute a “hyperparadigm”, i.e. the complex system in which the contrast between inflectional forms is integrated into the contrast between constructions.

Lene Schøsler, in her paper “Organization and reorganization of a constructional paradigm: The case of two-argument, dative constructions from Latin to Modern French”, argues in favour of analysing constructions as being paradigmatically organized and thus as parts of grammar. A construction is defined here as a complex sign with recognizable internal syntax and with a paradigmatic relation to other complex signs belonging to a common grammatical domain. It is argued that, being parts of grammar, constructions naturally undergo change, which is manifested in synchronic variation, as well as in diachronic reorganization. This is demonstrated by presenting the results of an extensive diachronic corpus study of two-argument dative constructions through the following four periods: Latin, Old and Middle French, Renaissance and Classical French, and Modern French. It is shown that, from the outset, the two-argument dative pattern in French was not linked to specific content. During the sixteenth and the seventeenth centuries, the pattern was associated with a number of verbs with more or less clearly comparable features. After a series of paradigmatic reorganizations, the marked construction emerged. Hence, in the course of time, the two-argument dative pattern has grammaticalized as the only construction expressing a stimulus–experiencer relation. This is to say that the two-argument, experiencer dative pattern has become a construction in the sense defined in this paper. As a conclusion, the author states that more research on paradigms is needed, especially on constructional paradigms and on the exact relation between valency patterns and constructions.

The paper by Jens Nørgård-Sørensen, with the title “What languages must convey: The construction-based syntax of Old Russian”, is inspired by the statement of Roman Jakobson (1959/1971, 264): “Languages differ essentially in what they must convey and not in what they can convey”. The paper deals with obligatory grammatical oppositions, i.e. with the content distinctions that a language must convey. It is argued that these distinctions essentially differ from other linguistic sign distinctions by being paradigmatically organized. The paradigm is taken here as a core concept in the definition of grammar and grammaticalization. This theoretical position is tested in a case study of Old Russian syntax compared with that of Modern Russian. The author proposes that there occurred a change from construction-based syntactic organization to valency-based syntactic organization from Old to Modern Russian. Whereas in a valency-based syntax the verb is the organizing centre determining the number of arguments and their form, in a construction-based syntax the construction is the primitive unit and organizing centre, restricting the semantics of the predicate and determining the number and form of arguments. The differences between the Old and the Modern Russian syntax are formulated in terms of what must be conveyed in the respective type of syntactic organization. In Modern Russian, a particular verb normally determines a specific number and form of its arguments. That means that the speaker must choose a particular argument structure when using a particular verb. In Old Russian, on the other hand, the verb did not determine a specific argument structure. Instead, the argument structure was exclusively determined by the construction. That means that the speaker must apply a particular set of arguments (including the verb) when activating a particular construction.

In a similar vein, the paper “The extension of the transitive construction in Ancient Greek” by Silvia Luraghi deals with the interrelation of paradigmaticity and obligatoriness. Right at the beginning of the paper, the author states: “paradigmaticity involves obligatoriness”. This theoretical consideration is demonstrated by using the developmental case of the active/passive voice opposition in Ancient Greek. The author explores the change from a stage at which there was no active vs passive voice opposition to the stage in which it became paradigmaticized and obligatory. The grammaticalization process resulting in the passive voice is described as consisting of two steps: reanalysis of certain verbs as active, and actualization of the reanalysis by a consistent coding of the objects of all transitive verbs through the accusative. As a consequence, the grammatical category of voice establishes itself through paradigmaticization of the opposition active vs passive. Passive becomes increasingly obligatory, and its extension proceeds from prototypically transitive verbs with accusative objects, to verbs with lower degrees of transitivity, with non-accusative objects.

The periphrastic voice construction with the verb “see” in Portuguese is the topic of the paper by Christian Lehmann, José Pinto de Lima and Rute Soares “Periphrastic voice with ‘see’ in Portuguese”. The Portuguese “see” passive, or ver-se passive, is a construction made up of the reflexive form of ver plus the passive participle of a full verb that applies to the subject of ver-se. The paper focuses on paradigmatic relations of this construction to active constructions and to other periphrastic voices, as well as on its semantic constraints, synchronic as well as diachronic. The central claim is that this construction has been grammaticalized as a periphrastic passive voice in Portuguese. The actual auxiliary status of ver is tested using the data from corpora of present-day Portuguese. The use of this and related constructions with ver is illustrated with corpus examples. From a diachronic perspective, the grammaticalization process of the construction as a whole and of the verb ver in particular, is reconstructed through the history of Portuguese. The authors show that the syntactic paradigm of the ver-voices has been gradually integrated into the syntactic paradigm of the pre-existent voice system. Moreover, it is argued that the grammaticalization of the ver-se passive can also be seen as a process of paradigmaticization based on an analogical model.

Liselotte Brems and Kristin Davidse present work on “Complex subordinators derived from noun complex clauses: Paradigmatic oppositions and enrichment of the paradigm”. The authors explore the development of complex subordinators originating in “in + NP + that/of-clause” syntagms, for example in the hope that + clause, in the history of English. By applying the model of relevant contextual types in grammaticalization (Diewald Citation2006), the authors reconstruct three stages in the development of the complex subordinators: untypical, critical and isolating contexts. This case study demonstrates that the reanalysis of noun complement clauses into complex subordinators contributes new members to the general paradigm of subordinators. The emergent complex subordinators enter into systems of paradigmatic oppositions, i.e. “the source semantics of specific nouns licensing complements are refashioned into the semantic and formal oppositions making up hypotactic clause-combining”. The authors argue that nominal phrases that take complement clauses should be acknowledged as an important source of complex subordinators, as they are able to express various semantic relations, such as temporal, consequential and comparative. The complex subordinators resulting from these syntagms are shown to be gradually moulded by the whole system of paradigmatic oppositions of hypotactic clause-combining.