Introduction: Theory and Theories in Morphology
Abstract and Keywords
This opening chapter provides an overview of the aims, structure, and contents of the volume. It ties together the individual chapters by identifying common themes that run through the various theories of morphology presented in the volume. These are the place of morphology in the architecture of language, the degree to which it is independent from other components of the grammar, the basic units of morphological analysis, and the relation between morphology on the one hand and syntax, semantics, phonology, and the lexicon on the other. A brief summary of the literature on types of morphological theories helps the reader to become oriented to the landscape of frameworks. The chapter closes with an overview of the three parts of the volume and the individual chapters in each part.
Morphology, the grammar of words, has proved a rich and fertile ground for theoretical research. As a result, we are faced with a bewilderingly complex landscape of morphological terms, concepts, hypotheses, models, and frameworks. Within this plurality, linguists of different persuasions have often remained ignorant of each other’s work. Formalist and functionalist theories have run on mutually isolated tracks; theoretical approaches have not connected to insights from typology, psycholinguistics, and other fields—and vice versa. The research community is divided about basic matters, such as the central units of morphological description or the nature of morphological features and processes. Moreover, the proliferation of theories goes hand in hand with an increasing internal diversification, sometimes to the point where foundational principles slip out of sight.
This volume hopes to contribute to a greater unity in the field by providing a comprehensive and systematic exposition of morphological theory and theories. We have aimed to make it a helpful resource for those working within a specific framework and looking for a critical and up-to-date account of other models, as well as a comprehensive guide for those wishing to acquaint themselves with theoretical work in morphology, perhaps coming from other domains in linguistics or from related fields such as computer science or psychology. The book is intended to be informative and inspiring, and a lasting contribution to the field. We also hope that—in times of increasing scepticism towards theory, in morphology as in other areas of linguistics—it will serve to showcase the richness and value of theoretical thinking and modelling, and will encourage new advances in theoretical work.
1.1.1 About the volume
This volume stands in the long tradition of Oxford Handbooks in Linguistics and complements other recent volumes, in particular The Oxford Handbook of Inflection (Baerman 2015), (p. 2) The Oxford Handbook of Derivational Morphology (Lieber and Štekauer 2014), The Oxford Handbook of Compounding (Lieber and Štekauer 2009b), and The Oxford Handbook of the Word (Taylor 2015). It is kin to The Oxford Handbook of Linguistic Analysis (Heine and Narrog, second edition 2015) by focusing on linguistic approaches more than on linguistic facts, although a wide variety of data is addressed.
The closest relative to the present volume is Stewart (2016) on contemporary morphological theories. However, our book is an edited volume rather than a monograph, and the scholars working in the various frameworks are speaking in their own voice. In addition to the eminent contributors expected in a volume of this kind, many of our authors are up-and-coming linguists with a fresh look on classic and novel issues.
While the field is too diverse for a reference work to be exhaustive, we have attempted to cover a representative range of theories and have made a point of including very recent models, such as Canonical Typology, Construction Morphology, and Relational Morphology. Moreover, Part III of the volume connects morphological theory with various linguistic subfields, identifying the broader challenges and opening the dialogue where it is often lacking.
1.2 Morphological theories
Despite the evident, and often drastic, differences between theoretical approaches, the theories in this volume are united in the questions they seek to answer. This section briefly reviews a selection of time-honoured issues that have shaped the theoretical landscape over the years and that reappear in different guises in basically every theory.
1.2.1 What is the goal of morphology theory?
Morphology is the grammar of words. This includes the form and structure of words, their meaning, the relations between words, and the ways new (complex) words are formed. Depending on one’s views of what a theory of grammar should accomplish, the goal of morphological theory is either to account for all existing words or for all potential words of a language. As Aronoff famously stated in 1976 (17–18): ‘the simplest task of a morphology, the least we demand of it, is the enumeration of the class of possible words of a language’. Whether this goal has been attained by any of the theories on the market, or can be attained at all, is a matter of debate, since the working area of morphological theory is not easily delimited. For one thing, the word is notoriously hard to define (Haspelmath 2011, see also Arkadiev and Klamer, Chapter 21 this volume). Moreover, the field of morphology runs into other linguistic subfields, with fluid boundaries and shared responsibilities.
1.2.2 Where is morphology?
Morphology is famously called ‘the Poland of linguistics’ (Spencer and Zwicky 1998: 1), surrounded by neighbouring fields eager to claim the territory for themselves. Many theories, some of them represented in this volume, model the structure and behaviour of words in syntax and/or in phonology (e.g. Distributed Morphology, see Siddiqi, Chapter 8 (p. 3) this volume, or Optimality Theory, see Downing, Chapter 10 this volume). The countermovement is gathered under the term of lexicalism (Montermini, Chapter 7 this volume) and the motto ‘morphology by itself’ (Aronoff 1994), arguing that morphology needs to be recognized as a module, layer, or level of description of its own because it has unique, irreducible properties. Lexicalist approaches ask questions such as the following:
• What properties are unique to morphology?
• How does morphology interface with other types of linguistic structure?
The issue of interfaces, of course, only arises if morphology is granted its own identity, distinct from other areas of grammar. However, views on interfacing differ greatly depending on whether morphology is understood in a broad sense or a narrow sense.
In a broad sense, morphology spans the entire bottom row of Figure 1.1 (adapted from Jackendoff and Audring, Chapter 19 this volume). This row is the domain of the word. Morphology then contrasts and interfaces with the upper row, syntax, the phrasal domain (cf. §1.2.4). The horizontal arrows within the bottom row—connecting morphosyntax, morphophonology, and morphosemantics—represent morphology-internal links, since a word contains all these types of information.
However, morphology can also be understood in a narrower sense. Words carry sound and meaning. In addition, they may have a third level of structure, which Figure 1.1 calls ‘morphosyntax’, marked in bold. This level of structure houses all properties that cannot be subsumed under phonology or lexical semantics.1 This includes grammatical features, such as case, gender, or tense, as well as properties such as inflectional class, the heartland of ‘morphology by itself’. In some theoretical models, this layer also encodes the building blocks of words: roots, stems, and affixes. Morphology, as understood in this narrower sense, contrasts and interfaces with word phonology and word meaning.
Many controversies in morphological theory follow from explicit or implicit disagreements about the nature and place of morphology in the grammar. While most theories accept morphology in the broader sense, as the part of language that handles words, some deny the existence of a dedicated layer of morphological structure in the narrower sense (e.g. Cognitive Grammar, see Langacker, Chapter 17 this volume).
Additional complications arise from the various conceptions of morphological processes. Theories differ in whether they assume different rules for the grammar of words and the (p. 4) grammar of phrases. Also, a division between morphological rules on the one hand and the input/output of such rules on the other can lead theories to posit a morphology–lexicon interface. This contrasts with theories that place morphology in the (equivalent of the) lexicon, for example Word Grammar (Gisborne, Chapter 16 this volume), Construction Morphology (Masini and Audring, Chapter 18 this volume), and Relational Morphology (Jackendoff and Audring, Chapter 19 this volume).
1.2.3 Basic units and processes
What are the units that morphological theory handles? Again, we see widespread and fierce disagreement. Two prominent camps have arisen around the word-based and the morpheme-based views, arguing for the word and the morpheme, respectively, as the basic unit of morphological structure. The debate is often framed in principled terms (see e.g. Anderson, Chapter 2 this volume, or Stump, Chapter 4 this volume), but sometimes invokes more specific concerns, such as which entity comes closest to a stable and transparent 1:1 relation between form and meaning (see e.g. Langacker and Gaeta, Chapters 17 and 12 this volume, respectively). A complicating factor is the notorious difficulty to define either the word or the morpheme in a consistent and cross-linguistically applicable way. However, in view of the controversy surrounding the morpheme in particular, it is worth noting that the term is used widely and freely in descriptive linguistics as well as in psycho- and neurolinguistics, where it is found to be of value (see e.g. Schiller and Verdonschot, Chapter 28 this volume).
The chapters in the present volume show surprisingly little debate about the lexeme, which is a central unit in a variety of influential theories (e.g. Stump, Chapter 14 this volume). This notion is related to the difference between inflection and derivation, which itself is not easy to draw. While most theories make a point of distinguishing inflection and derivation/word-formation—some clearly specializing in one or the other—the nature of the difference is disputed, especially as to whether it is gradual or categorial (sometimes intermediate distinctions are made, such as between inherent and contextual inflection, Booij 1996). The issues scale up to the difference between morphology and syntax, and more generally between the grammar and the lexicon, since inflection is generally believed to be more relevant to syntax and on the whole ‘more grammatical’ than derivation. Within word-formation, certain types of compounds and lexicalized multi-word units further blur the boundaries between morphological and syntactic structures (see Arkadiev and Klamer, Chapter 21 this volume).
A further basic difference between frameworks is how they conceive of the relation between the units of morphological analysis and the processes that handle them. While units and processes are tightly wedded in many theories, with rules for specific affixes or individual feature structures, in others they are clearly separated. An example for the latter type is Minimalism (Fábregas, Chapter 9 this volume), some variants of which rely on a single general operation, Merge.
Other differences between theories are found in the way classes, features, and other properties are encoded. Some theories also seek to encode relations, from syntagmatic relations such as valency or agreement to paradigmatic relations such as those found in inflectional morphology.
(p. 5) 1.2.4 Morphology and syntax
Theories of morphology can be differentiated by the way they model the relation between morphology and syntax. Does the grammar of words involve its own module, with rules and representations distinct from the rules and representations of phrasal grammar? All extremes can be found: from assuming no difference at all (e.g. in Distributed Morphology, see Siddiqi, Chapter 8 this volume) to a strictly modular view in which morphology is encapsulated from syntax (e.g. in LFG/HPSG, see Nordlinger and Sadler, Chapter 11 this volume). For theories such as Construction Morphology (Masini and Audring, Chapter 18 this volume) or Relational Morphology (Jackendoff and Audring, Chapter 19 this volume), the difference lies not in the processes—morphological versus syntactic rules—but in the categories: morphology has stems and affixes, while syntax does not, and syntax has phrasal categories such as NPs and VPs, while morphology does not.
For those theories that do assume a split between morphology and syntax, the question arises how the two components interface. An often-cited assumption is that X0, the syntactic word, serves as the interface. This view runs into difficulties with complex words containing phrases, as in do-it-yourselfer (the No Phrase Constraint is discussed in Montermini and Fábregas, Chapters 7 and 9 this volume, respectively).
Other related points of debate, recurring in many theories throughout the book (see Lieber, Chapter 3 this volume, for an overview), are lexical integrity—the (in)ability of syntax to look into or manipulate word structure—and the issue of headedness, disputing the equivalence of syntactic and morphological heads.
1.2.5 Morphology and semantics
Another important issue in morphological theory is the relation between meaning and form. The canonical mapping is captured in the terms isomorphy, biuniqueness, transparency, compositionality, diagrammaticity (Gaeta, Chapter 12 this volume), or ‘the concatenative ideal’ (Downing, Chapter 10 this volume): each piece of meaning should correspond uniquely to a piece of form, and added meaning should go hand in hand with added form. A lot of what makes morphological theory interesting and hard has to do with divergences from this ideal.
The issue is pertinent to the divide between word-basedness and morpheme-basedness. Are there privileged units in which the relation between form and function is clearest or maximally stable? And if so, is the word or the morpheme a better candidate?
Violations of biuniqueness come in many guises. Well-studied phenomena are polysemy, homophony, and syncretism (cases of one form with several meanings), allomorphy, periphrasis, multiple or extended exponence (cases of one meaning expressed by several alternative or combined forms), plus a range of specifically paradigmatic mismatches, such as suppletion, overabundance, heteroclisis, and deponency (Stump, Chapters 4 and 14; see also Arkadiev and Klamer, Chapter 21, and Ralli, Chapter 24, all in this volume). In addition, complex words can display semantic non-compositionality, with unpredictable meanings showing up in individual words or as subregularities in clusters of words. While many theories set such quirks aside as lexicalizations, others make a point of including them, for example Construction Morphology (Masini and Audring, Chapter 18 this volume).
(p. 6) 1.2.6 Morphology and phonology
The interplay of morphology and phonology is another much-debated issue. Many theories in the generative tradition (e.g. Minimalism and Distributed Morphology, see Siddiqi, Chapter 8, and Fábregas, Chapter 9) model phonology as a spell-out component at the end of a syntactic derivational chain. This means that phonological information cannot play a role in the morphological operations themselves. Other theories (e.g. LFG and HPSG, see Nordlinger and Sadler, Chapter 11 this volume) argue that all information, including phonology, has to be available at the same time.
The most-researched interface phenomenon between morphology and phonology, however, is allomorphy, and almost every theory has something to say about it. The most pressing question with regard to allomorphy is whether variants of stems or affixes are computed from some underlying form or are listed and selected from memory. This brings us to the final major issue: the relation between morphology and the lexicon.
1.2.7 Morphology and the lexicon
Morphology is a part of grammar, and many theories make a principled distinction between the grammar and the lexicon. However, morphology is the grammar of words, and words live in the lexicon. This means that we have to ask whether morphology happens in the lexicon or whether the lexicon and the morphology are different domains, connected via an interface. Terminology is muddled here, and we often find different understandings of the same term, or different terms for the same notion. For example, Distributed Morphology has a vocabulary, which corresponds to the lexicon in other theories. Earlier generative theories distinguish a lexicon of morphemes and a dictionary of words (see ten Hacken, Chapter 6 this volume).
The distinction between lexicon and grammar is intimately related to the division of labour between storage and computation. This issue is especially pertinent to the chapters in Part III of this volume that discuss morphology in first and second language acquisition (Blom, Chapter 25, and Archibald and Libben, Chapter 26), in psycho- and neurolinguistics (Gagné and Spalding, Chapter 27, and Schiller and Verdonschot, Chapter 28), and in computational modelling (Pirrelli, Chapter 29). However, it is also relevant to morphological theory itself, which has to decide on the format of lexical representations and on the kinds of items assumed to be in the mental lexicon. Again, this is an area where word-based and morpheme-based theories clash. While the former expect the smallest entries in the lexicon to be word-sized (Blevins, Ackerman, and Malouf, Chapter 13 this volume, and Gisborne, Chapter 16 this volume), the latter posit entries for morphemes or even smaller structures (Siddiqi, Chapter 8 this volume). The crux is the modelling of regularly inflected word forms. Such forms are predictable enough to be handled by grammar, yet some degree of listed knowledge is necessary to choose the right form among alternatives, for example if the language has inflectional classes (Blevins, Ackerman, and Malouf, Chapter 13 this volume). Generally, models differ in the degree to which they embrace or reject redundancy in areas that can be handled both by lexical storage and by grammatical computation.
Last but not least, a major and problematic issue is productivity, the capacity to generate new complex forms with a particular structure. In contrast to syntax, where full productivity (p. 7) is commonly seen as the norm, morphology—especially derivational morphology—is rampant with semi-productive or unproductive patterns (see Hüning, Chapter 23 this volume). An important challenge for morphological theories lies in the modelling of such limited productivity. Theories that emphasize the generative capacity of the system commonly evoke constraints or filters that block non-existing forms (see e.g. Chapters 10 and 8 by Downing and by Siddiqi, respectively); others argue for built-in limitations in the system itself (Jackendoff and Audring, Chapter 19 this volume). A considerable degree of agreement is found in the modelling of blocking, where a well-formed but non-existing complex word (say, stealer) is impeded by an existing form with the same meaning (thief). Almost all theories that have something to say about blocking invoke a principle by which the specific properties of the listed form block the application of a more general rule.
1.2.8 Taxonomies of theories
In the linguistic literature we find various attempts to classify morphological theories. Two of them are repeatedly cited in the present volume. The earliest is Hockett’s (1954) ‘Two Models of Grammatical Description’, which distinguishes Item-and-Process from Item-and-Arrangement types of theories. Blevins, Ackerman, and Malouf (Chapter 13 this volume) explain the difference between the two morphemic models, contrasting them with their own Word-and-Paradigm approach.
The second classification is Stump’s (2001) well-known distinction of lexical versus inferential and incremental versus realizational theories, giving us a four-way taxonomy, which is laid out in Stump’s chapter ‘Theoretical issues in inflection’ (Chapter 4 this volume). Various theories presented in Part II of the volume explicitly position themselves on this grid.
A very recent classification is proposed in Stewart’s (2016) book, which sorts morphological theories along each of five axes, explicitly incorporating one of Stump’s classifications:
• morpheme-based vs. word-/lexeme-based
• formalist vs. functionalist
• in-grammar vs. in-lexicon
• phonological formalism vs. syntactic formalism
• incremental vs. realizational.
As some of the theories discussed by Stewart converge with those in the present volume, the reader is encouraged to consult the monograph for details.
Finally, Blevins’ (2006) distinction of constructive versus abstractive models is helpful due to its more nuanced take on the theoretical treatment of sub-word structures. The abstractive view, in particular, permits for a combination of word-basedness and word-internal structure, which might be an opportunity for consensus.
Generally, it should be kept in mind that theoretical frameworks can have different goals and rest on different foundational assumptions. While one theory emphasizes descriptive coverage or psychological plausibility, others stress computational implementability and/or architectural parsimony, that is, shorter descriptions and minimal machinery. Among the theories that seek parsimony, we find those that strive to minimize storage (these are clearly (p. 8) in the majority) and those that attempt to minimize computation. Such basic decisions have deep repercussions on the architecture of the model and on the items and processes assumed.
Finally, it should be noted that not all models presented in this book are bona fide theories of morphology. Some are in fact theories of syntax (e.g. Minimalism), and one (OT, see Downing, Chapter 10 this volume) is mainly a theory of phonology. However, each chapter illustrates the perspectives on morphology taken by these theories.
1.3 The structure of the handbook
1.3.1 Part I: Issues in morphology
Part I of the volume sets the scene. It starts with a brief foray into the history of morphology with a focus on North America (Anderson, Chapter 2). However, the journey begins in Switzerland, with the brothers de Saussure, Ferdinand and René, and their disagreement on the internal structure of words. While René saw complex words as concatenations of simple signs, later called morphemes, Ferdinand regarded the full word as the basic sign. To him, morphological structure emerged from inter-word relations. The morpheme-based view was perpetuated by Bloomfield (1933), who differentiated between a lexicon of primitives, on the one hand, and the rules of grammar, on the other. Full words came back into view with Matthews’ (1965), Aronoff’s (1976), and Anderson’s (1992) work, which reinstated the paradigmatic, relational perspective and found its most radical expression in Anderson’s ‘a-morphousness’ hypothesis, propagating morphology without morphemes. In addition to sketching the swing of the historical pendulum between word-based and morpheme-based models, the chapter shows the influence of Boas, Sapir, Harris, Chomsky, and Halle on the emergence of morphology as an independent domain in theoretical linguistics, and introduces some of the fundamental debates that have shaped the theoretical landscape in the following decades.
The next two chapters identify the central theoretical issues within the two morphological domains: word-formation and inflection. Lieber’s contribution (Chapter 3) on derivation and compounding also starts with a major historical divide, namely Item-and-Arrangement versus Item-and-Process types of theories (Hockett 1954). While the former makes morphology similar to syntax in assuming a hierarchical structure of minimal meaningful units, the latter emphasizes the importance of rules in deriving, or realizing, complex words. Here, morphology offers a variety of challenges. Do the rules of morphology have the same format as the rules of syntax? Can realizational rules, popular in modern theories of inflection, be fruitfully applied to derivation? The chapter continues with a discussion of interface issues between morphology and syntax, morphology and phonology, and morphology and semantics. It concludes with a number of hot topics such as headedness, productivity, blocking, affix ordering, bracketing paradoxes, and derivational paradigms.
In Chapter 4, Stump discusses theoretical issues in inflection. He singles out a number of fundamental points of disagreement between theories of morphology. These are: (a) what (p. 9) counts as the basic unit of morphological analysis; (b) what are the structures that belong to inflection; (c) the relation between concatenative and non-concatenative morphology; (d) the relation between function and form; and (e) the difference between inflection and other types of morphology. After outlining the issues, the chapter takes a position on each of them. As the general perspective of the chapter is inferential-realizational, Stump argues for paradigms and against morphemes, for rules of exponence and implicative rules, and for a unified treatment of concatenative and non-concatenative morphology. Morphology is argued to have its own domain in the grammar, distinct from but interfacing with syntax.
1.3.2 Part II: Morphological theories
Part II consists of concise but thorough accounts of the main theoretical approaches to morphology, both formalist and functionalist/cognitive, developed during the twentieth and early twenty-first centuries. Some chapters discuss clusters or families of models, but most are dedicated to one specific approach.
The first three chapters provide an overview of three clusters of theories: those commonly subsumed under the label Structuralism (Chapter 5), the transformational theories of early Generative Grammar (Chapter 6), and the lexicalist models of later Generative Grammar (Chapter 7).
Stewart (Chapter 5) identifies Structuralism as a formative period in the history of linguistics. It brought a re-evaluation of the theoretical and descriptive machinery inherited from antiquity and established linguistics as an autonomous scientific discipline. The central characteristic of the movement was the understanding that each language constitutes a system in itself which should be investigated empirically based on the distributional patterns of forms. This involved overcoming the focus on Indo-European, on culturally privileged languages, and on diachrony. The result was a flourishing of scholarly work on both sides of the Atlantic, with—among many others—de Saussure, Hjelmslev, Jakobson, Trubetzkoy, and Vachek in Europe and Boas, Whorf, Sapir, Bloomfield, Harris, Hockett, and Nida in North America. Important issues for morphology were the place of morphology in the architecture of the grammar, the identification and representation of morphological units and processes, and the interaction of morphology with other linguistic domains.
The 1950s to 1970s saw the rise of Generative Grammar. Ten Hacken (Chapter 6) discusses three seminal publications from this period, Chomsky (1957), Lees (1960), and Chomsky (1970), and—more briefly—two later publications, Halle (1973) and Jackendoff (1975), which are the focus of Chapter 7. The central innovations in early Generative Grammar were rewrite rules, including transformational rules, that promised to make complex grammatical structures computable. While mainly devised for syntax, the model was also applied to morphological structure. A lexicon was added to account for idiosyncratic properties of words, marking the beginning of the debate between storage and computation, still very much alive today (see Chapters 25–28). Other major issues of the time were the incorporation of constraints into the generative model and the place and role of semantics.
The history of Generative Grammar is continued by Montermini with Chapter 7 on the development of Lexicalism. The hallmark of lexicalist theories is the assumption that word-internal phenomena are situated in a distinct module, independent of syntax and phonology. For many theories, this included the belief that the grammar of words is not (p. 10) only separated, but also substantially different from the grammar of phrases. Montermini discusses two foundational publications, Halle (1973) and Jackendoff (1975), which can be seen as the first lexicalist models, although diverging fundamentally in their assumptions about the interplay of grammar and lexicon and the nature of the lexicon itself. The lexicalist spirit continued through Aronoff’s work on derivation and Anderson’s work on inflection, the latter stressing not only the division between morphology and syntax, but also the need to distinguish between inflection and derivation. The surge of lexicalist work from the 1970s onwards established morphology as a phenomenon ‘by itself’ and a self-respecting field of linguistic inquiry.
Chapters 8–11 describe models of a ‘formalist’ orientation. The direct inheritors of Chomskyan Generative Grammar are Distributed Morphology and Minimalism, while Optimality Theory and LFG/HPSG constitute radically different models.
Distributed Morphology (Siddiqi, Chapter 8) represents a countermovement to the Lexicalism described in Chapter 7: it is a theory of syntax that extends into the word by manipulating morphemes. The chapter motivates the general outlook as well as the more specific choice for a lexical-realizational, morpheme-based, Item-and-Arrangement type of model and outlines its various incarnations, depending on the syntactic theory of the time. Some variants distinguish a separate level of Morphological Structure, later abandoned. Complex words and phrases are built in two steps: the grammar constructs a complete derivation, which is then instantiated by Vocabulary Items and spelled out phonologically. This architecture makes a number of classic morphological issues—among others productivity, blocking, and allomorphy—appear in a different light, as is elaborated in the chapter.
Another theory that is actually a family of syntactic models is Minimalism. Fábregas explains the Minimalist views on morphology in Chapter 9. The name of the framework advertises its emphasis on a minimal grammatical component, as most constraints on language are seen as located either in Universal Grammar or in language-external systems, especially the Conceptual-Intentional (CI) and the sensorimotor (SM) system, or in the variable experience of individual speakers and learners. In its most minimal form, computations are done by a single operation, Merge. The chapter explains how the theory models lexical restrictions, grammatical categories, Aktionsart, and argument structure and discusses the rules of spell-out and the role of features.
Chapter 10 by Downing illustrates how Optimality Theory addresses the issues of prosodic morphology, specifically the non-concatenative phenomena known as reduplication, truncation, root-and-pattern morphology, and infixation. The model employs three types of constraints—faithfulness, markedness, and alignment—to determine the optimal form of a word or phrase. Constraint evaluation is demonstrated on a wide variety of languages, among others SiSwati, Japanese, Modern Hebrew, Samoan, Diyari, and Nupe. Important theoretical issues are (a) whether constraints can be stated generally or are specific to a certain morphological operation, construction, or morpheme, and (b) whether restrictions (e.g. on the size of the optimal nickname or the location of the optimal infix) follow from other properties, such as the stress type or syllable structure of the language.
Chapter 11 presents two distinct but related theories, Lexical-Functional Grammar (LFG) and Head-Driven Phrase Structure Grammar (HPSG). Nordlinger and Sadler briefly explain the architecture and the formalism of the two models, highlighting their strong lexicalist commitment, which states that word-internal structure is invisible to syntax. This (p. 11) perspective implies that both theories are compatible with a variety of morphological models, as long as the lexicalist stance is maintained. In LFG, the emphasis is on the way different formal structures across languages can map onto the same functional structure. Some variants of HPSG are similar to construction-based theories (cf. Masini and Audring, Chapter 18 this volume) by modelling derivational rules as lexical items, while inflection is often understood as being realizational. The chapter discusses a variety of phenomena, from case stacking to paradigms, stem space, and floating affixes, in a number of typologically diverse languages. Both theories are fully formalized and implementable in computational models.
In Chapter 12, Gaeta sketches Natural Morphology, a framework that strives to explain why morphological systems are the way they are and develop in the way they do. At the heart of the theory lies the notion of ‘naturalness’, understood as “cognitively simple, easily accessible (esp. to children), elementary and therefore universally preferred” (Dressler 2005: 267). Naturalness manifests itself in preferences rather than laws. Such preferences can be in conflict with each other and with other preferences—both typological and system-specific—resulting in cross-linguistic diversity. The chapter introduces the naturalness parameters (i) diagrammaticity (transparency); (ii) biuniqueness (uniform coding); (iii) indexicality (proximity); (iv) binarity; and (v) optimal word shape and exemplifies how they bear on productivity, paradigm structure, and language change.
Chapters 13–15 form a loose cluster of allied models of the Word-and-Paradigm type. The general outlook is described succinctly in the contribution by Blevins, Ackerman, and Malouf, Chapter 13. A major cornerstone is the focus on paradigmatic relations among words, which other models tend to neglect in favour of word-internal syntagmatics. Paradigmatic relations can take the form of inflectional paradigms or classes, but they are implicated whenever a word, or a cluster of words, is predictive of another. The Word-and-Paradigm (perhaps better called Item-and-Pattern) approach involves a broadly inclusive view on the size and granularity of morphosyntactic items, as it is “defined less by the units it recognizes than by the relations it establishes between units” (§13.3). That said, the word might be a privileged unit, both in its stability of form and function and the mapping between them, and in the degree to which it predicts other words. Formalizations of Word-and-Paradigm models use the mathematics of information theory to calculate the entropy of a given paradigm cell and the reduction of uncertainty effected by another cell or cluster of cells. The chapter closes with the unique perspective on learnability and cross-linguistic variation invited by the information-theoretic perspective.
In Chapter 14, Stump presents his influential Paradigm Function Morphology, an inferential-realizational theory, which means that it rejects the listing of morphemes and the accumulation of properties by stringing morphemes together. Instead, the model assumes a Paradigm Function that operates on stems and cells of inflectional paradigms to induce the realization of each cell, that is, the phonological form of the fully inflected word. The model employs an explicit and rigorous formalism based on property sets and functions. The chapter lays out an earlier and a later variant of the theory and illustrates the basic functions. As the theory emphasizes its inclusive coverage, the second half of the chapter is devoted to non-canonical inflectional morphology, as manifested in defectiveness, syncretism, inflection classes, and deponency. The chapter closes with a brief look at derivation and the various interfaces between morphology and other domains.
(p. 12) Network Morphology, outlined by Brown in Chapter 15, has much in common with PFM—centrally the inferential-realizational orientation—but differs in its architecture. As its name suggests, the model assumes an inheritance network containing lexemes and generalizations over their properties. Its aim is to model inflectional systems by developing the most parsimonious network that contains all information necessary for inferring the correct form for each inflected word. This means determining the right level for every generalization (e.g. about patterns of syncretism or stem allomorphy) and ordering properties such as number or case in such a way that queries about a particular inflected form are guided to the place in the network where the answer is encoded. The model is formalized and computationally implemented with the help of the DATR notation (Evans and Gazdar 1996). The chapter explains central notions such as default inheritance, underspecification, and generalized referral and shows the application of the model in a number of case studies, including a diachronic case.
Word Grammar, discussed by Gisborne in Chapter 16, shares many traits of realizational models like PFM and is network-based like Network Morphology, but differs radically in the entities it models. In line with the cognitive orientation of the theory, nodes in a Word Grammar network encode linguistic knowledge directly and declaratively, requiring no procedures or algorithms. The network encodes three types of information: linguistic structure of various kinds (the nodes), the relations between nodes, and certain attributes that specify the relations (e.g. realization, base, variant, or part). Inflected and derived forms are represented in full. Morphemic structure is encoded indirectly via relations between forms that share parts. Generalizations, including those normally expressed as features, are captured by means of default inheritance. The chapter also discusses the difference between inflection and derivation, the interfaces between morphology and the lexicon and morphology and syntax, and comments on phenomena like productivity and syncretism.
Word Grammar forms a bridge to the more cognitively oriented models in Chapters 17–19. The first and most venerable is Cognitive Grammar by Langacker (Chapter 17). Including this theory in the volume might seem surprising, as it only recognizes two types of structure—semantic and phonological—and excludes morphological structure. Yet, the model allows for the expression of morphological units and patterns, both in individual words and as generalized constructional schemas. The perspective is explicitly usage-based: any unit of structure is abstracted from production or perception events and entrenched through recurrent use. Larger structures appear as composites if their parts correspond to (parts of) other structures. Stems can be distinguished from affixes in that affixes are dependent items that need other structures to be manifested. However, analysability of complex items is a matter of degree and can change over time. The theory provides a unified account of language structure, within which morphology is not highly differentiated, but seamlessly integrated.
Construction Morphology (Masini and Audring, Chapter 18) is the morphological theory within the framework of Construction Grammar. It shares a number of properties with Cognitive Grammar, especially its usage-basedness and the notion of constructional schemas. However, it assumes morphological structure as an independent layer of information. The central unit of analysis is the construction, intended as a sign, a form–meaning pairing. Constructions can be fully specified, in which case they correspond to words, or (p. 13) they can be partly or fully schematic. Schematic or semi-schematic constructions are the counterpart of rules in more procedural models, since they serve as templates for the creation of new words. All constructions are situated in a network which combines the lexicon and the grammar into a continuous and highly structured environment. As the same basic architecture is assumed for morphological and syntactic constructions, the model has a specific affinity with in-between phenomena such as multi-word units.
The newest theory in the volume is Relational Morphology (Jackendoff and Audring, Chapter 19), an account of morphology set in the framework of the Parallel Architecture (Jackendoff 2002). The model is a sister theory of Construction Morphology, but differs by virtue of its radical focus on lexical relations, its inclusion of non-symbolic structures, and its formalism. Special theoretical attention is paid to unproductive patterns, which are regarded as more basic: productive patterns are patterns ‘gone viral’. Like all construction-based theories, but more explicitly so, the model is a theory both of morphology and of the rich internal structure of the lexicon. Moreover, it aspires to a graceful integration of morphology within a general and cognitively plausible model of language, and of language within other areas of cognition.
The survey of theories concludes with Canonical Typology (Bond, Chapter 20), which is special in not being a theory in the usual sense, but providing a methodological framework for a typologically informed understanding of linguistic phenomena and a better comparability of theoretical terms and concepts. Most of the work in Canonical Typology is on morphology and morphosyntax, especially inflection, with the closest ties to inferential-realizational models like PFM (Chapter 14). The method consists in the identification of a canonical core for a phenomenon and the possibility space of less canonical variants around it. Both the core and the possibility space are defined logically; establishing the actual population of the space by real-life examples is an independent, later step. The chapter outlines the method in detail and provides a wealth of references on the canonical approach as applied to a wide variety of phenomena.
1.3.3 Part III: Morphological theory and other fields
Part III of the volume is devoted to the interdisciplinary dimension. It presents observations and insights from other linguistic fields relevant for morphological theory, namely language typology (including creole languages), dialectal and sociolectal variation, diachrony, first and second language acquisition, psycholinguistics, neurolinguistics, computational linguistics, and sign languages. The chapters in this part do not discuss what the different theories of morphology have to say about the various fields (this should emerge—where relevant—from Part II), but illustrate how each field informs and challenges morphological theory.
Chapter 21 by Arkadiev and Klamer on morphological theory and language typology discusses the challenges that languages around the world pose for common theoretical concepts and terminology. This is especially true for morphology, since it is the domain where languages differ most, which stands in the way of cross-linguistic generalizations. Richly illustrated with examples and supported by a wealth of references, the chapter shows the difficulties associated with the notion of the word, the distinction between inflection and (p. 14) derivation, the deviations from biuniqueness in form–meaning mapping, the ordering of affixes, and various paradigmatic phenomena such as inflectional classes and morphomic allomorphy in stems and affixes. The authors conclude by arguing for greater collaborative efforts among typologists, theoreticians, and descriptive linguists in order to arrive at theoretically informed descriptions, dictionaries, and corpora, on the one hand, and typologically informed theories, on the other.
In Chapter 22, Luís carries on the typological theme with a survey of the morphology in creole languages. Creoles are often neglected in theoretical morphology, as their morphological systems are said to be poorly developed. The chapter refutes this assumption, showing the interesting diversity of morphological, especially derivational, patterns found in creole languages. These include affixes from both the superstrate and the substrate language, as well as novel morphological formatives, which gives interesting insights into the genesis of affixal morphology. While inflectional systems in creoles are indeed often simpler, languages do show complexities such as portmanteau morphemes, extended exponence, syncretism, allomorphy (including morphomic stem allomorphy), and inflectional classes. The chapter demonstrates that creole morphology is as interesting to analyse formally and discuss theoretically as is the morphology of non-creole languages.
The issue of diachronic change, pertinent to the creole languages discussed in Chapter 22, is addressed more broadly in Chapter 23 by Hüning. The chapter focuses on word-formation and discusses three major types of change: (a) the rise of new word-formation patterns by way of reanalysis, for example ‘affix telescoping’ or resegmentation; (b) the development of new affixes from lexical words through grammaticalization; and (c) the increase or decrease of productivity. Productivity proves especially problematic, being hard to establish synchronically, but even harder to assess diachronically. A general problem is the gradience that the ever-changing nature of language imposes on all entities, properties, and behaviour, making them difficult to capture in fixed theoretical categories and terms. The chapter closes with a plea for interdisciplinary, data-driven research, and a usage-based approach that is better suited to the emergent nature of language.
Variation from a synchronic perspective—with some additional discussion of pathways of change to complement Chapter 23—is addressed by Ralli in Chapter 24. The chapter identifies certain recurrent types of morphological variation in inflection, derivation, and compounding, with illustrative analyses of modern Greek dialect data. For inflection, patterns of special interest are overabundance, heteroclisis, and allomorphic variation in paradigms. In derivation, we find affix synonymy and affix competition. In the realm of compounding, the Greek data show puzzling doublets of left-headed (and exocentric) and right-headed (and endocentric) compounds with the same meaning. For all three domains of morphology the chapter stresses the importance of language contact as a trigger of variation and change, and as an explanatory factor in view of the often surprising variational patterns.
The volume continues with four loosely connected chapters on morphological theory and first and second language acquisition, psycholinguistics, and neurolinguistics. All four chapters share a common fundamental theme: the division of labour between storage and computation in complex words.
In Chapter 25, Blom outlines how data from first language acquisition can inform morphological theory. A central topic is the ‘past tense debate’ inquiring whether irregularly and regularly inflected English verbs are treated differently in processing, with full-form (p. 15) lookup for the former and computation from their parts for the latter.2 While the evidence is not conclusive, analyses of child language indicate a gradual acquisition curve with frequency effects both in acquisition order and in overgeneralization patterns, which suggests that lexical storage also matters for regularly inflected words. Results from language development in children with Specific Language Impairment or Williams syndrome, by contrast, do support a difference between the regular and irregular words. To date, the past tense debate remains unresolved. Deeper understanding can only be expected if individual and cross-linguistic variation is considered, as well as the interplay of morphology, phonology, and syntax and wider cognitive factors.
Archibald and Libben in Chapter 26 move the spotlight of attention to morphological theory and second language acquisition. The issues in this field are partly the same as in first language acquisition. What do error patterns tell us about linguistic knowledge? Which deficits are betrayed by a particular error? What makes certain structures difficult to acquire? Additional questions in second language acquisition concern the influence of the L1, the typical cognitive strategies of adult learners, and the representation of the bilingual lexicon and grammar in the brain. An important insight, also mentioned in Chapter 25, is that morphological errors need not represent morphological deficits. Instead, they may be caused by incorrect mappings of morphological knowledge to other aspects of linguistic competence, for example phonology. The chapter also problematizes the question of what constitutes morphological ability and presses the point that morphological knowledge cannot be investigated in isolation from other kinds of knowledge. In addition, scientific results are highly task- and methodology-dependent and may differ markedly for production and comprehension.
Chapter 27 by Gagné and Spalding broadens the view from language acquisition to psycholinguistics in general, focusing on the key question for morphology: the representation and processing of complex words in the mental lexicon. The central debate is whether complex words are stored in full or computed from their parts, or indeed both—in succession or in parallel. The chapter reviews a wide variety of psycholinguistic research from different experimental paradigms and concludes that there is strong overall evidence for the involvement of sub-word units in the processing of multi-morphemic words. However, the effects differ depending on frequency, on semantic transparency, and on whether the complex word is inflected, derived, or a compound. Sub-word units may have a facilitatory or inhibitory effect depending, again, on frequency and on the time window in the processing event. The chapter closes with an agenda for future work, emphasizing the need for a closer integration of experimental and theoretical morphology.
The fourth chapter in the cluster, Chapter 28, is Schiller and Verdonschot’s contribution on morphological theory and neurolinguistics. Neurolinguistics differs from psycholinguistics primarily in its methods: most of the evidence cited in the chapter comes from brain imaging (p. 16) studies using ERP or fMRI. Again, the main issue is the role of sub-word structure in the processing of complex words. The chapter provides a broad and detailed overview of recent research on language comprehension, that is, parsing, and language production, the less-studied perspective. Evidence from healthy speakers is discussed as well as studies on individuals with aphasia or other language disorders. The chapter presents a variety of experimental paradigms, from priming and grammatical violation experiments to lexical decision tasks and picture naming. Drawing especially on compound processing, the chapter argues for an important role of morphemic constituents, indicating morphological decomposition in both comprehension and production.
The volume continues with Pirrelli (Chapter 29) on morphological theory, computational linguistics, and word processing. The chapter reviews computational models of language processing such as finite state automata and finite state transducers, hierarchical lexica, artificial neural networks, and dynamic memories. Illustrations are given with the help of Italian verbal paradigms. A substantial part of the chapter is devoted to machine learning, both supervised and unsupervised. Each section concludes with a critical assessment of theoretical issues, pointing out ties to individual theoretical frameworks or to problem areas such as the interplay of storage and computation, the nature of representations, the encoding of general versus specific information, and notions such as entropy and economy. The chapter argues for an inclusive modelling of lexical and grammatical knowledge and highlights the mutual interdependence of word structure and word processing.
The final Chapter 30 by Napoli broadens the view from spoken language to sign language. The particular affordances and restrictions of sign languages pose considerable challenges to morphological theory. For example, signs can be uttered in parallel, adding a vertical structural dimension not found in speech. Moreover, sign phonology, in particular non-manual parameters, can be meaningful, which obscures the boundary between phonology and morphology. Other special properties can be attributed to the relative youth of sign languages, which limits the amount of grammaticalized morphology. Established theoretical notions are often hard to apply to sign, for example in identifying roots and affixes or distinguishing lexical categories. Compounding and affixation are notoriously hard to tell apart. On the other hand, there are morphological entities unique to sign, such as ion morphs: partially complete morphemes that need to be accompanied by a particular phonological parameter to yield a full lexical meaning. The chapter offers a broad overview of the issues and a wealth of references.
In conclusion, we hope that this handbook will serve as a guide through the jungle of theories in today’s linguistic morphology, and the phenomena they seek to account for. At the same time, we intend the volume to be helpful in fostering the dialogue among sub-disciplines that is much needed for a graceful integration of linguistic thinking. We hope that the book will be inspiring and useful to graduate students in linguistics as well as to scholars of various disciplines, from morphologists wishing to acquaint themselves with neighbouring or competing models to specialists from other subfields of linguistics.
(1) Note that the term ‘morphosyntax’ is used differently here than it is used in the typological literature, where it denotes morphological structure relevant to syntax (e.g. in agreement).
(2) The terms ‘single route’ and ‘dual route’ are used in this connection; these terms also appear in Chapters 27 and 28. However, the reader should be aware that they are not always used in the same sense. Dual route is often associated with different processing mechanisms for different types of word (e.g. in Blom, Chapter 25, and Schiller and Verdonschot, Chapter 28). However, the term can also mean different processing strategies for the same type of word (e.g. in Gagné and Spalding, Chapter 27). Evidence in favour of parallel lookup and computation for various types of complex word would support a dual route theory in the latter sense, but not in the former.