The co-editors of this volume, Dunstan Brown, Marina Chumakina & Greville Corbett, are members of the Surrey Morphology Group at the University of Surrey. This book can thus be considered a theoretical manifesto of this well-known centre of linguistics, which mostly focuses on issues of theoretical morphology and morphosyntax.
The ambitious title of the volume points to the framework which aims to elaborate a set of parameters and criteria that will allow scholars of different linguistic frameworks to compare the data they obtain. This important task has its roots in what the leader of the Surrey Morphology Group, Greville Corbett (Reference Corbett, Scalise, Magni and Bisetto2009: 129), determined as the Correspondence Problem: ‘as typologists we need to be able to justify treating features and their values as comparable across languages. This is not straightforward, and yet a good deal of typology, including enterprises such as the World Atlas of Language Structures, depends upon it’.
To put it differently, it is extremely important for linguists to be able to decide whether ‘similarly named features in different languages’ (1) refer to (essentially) the same entities. The aim to achieve more rigorous understanding and definitions of the theoretical key concepts is a feature of a ripe science, which linguistics undoubtedly had become by the second half of the 20th century, when the amount of accumulated information has surpassed a critical mass. This task is certainly not a novelty in our science in general and in the domain of morphology and syntax in particular, and has been undertaken within a variety of frameworks. Suffice it to mention such seminal works as Nida (1946/Reference Nida1949), focusing on the notion of word and other morphological concepts (and still remaining a classics), and, more recent, Mel'čuk (Reference Mel'čuk1982, 1993–Reference Mel'čuk2000).
This ambitious task suggests a well-elaborated research program, which the Surrey Morphology Group has developed during the last two decades, eventually becoming one of the world's most prestigious linguistic research groups. The present volume makes a further step towards this important theoretical goal, thus being an important event in linguistics.
The first chapter, written by Dunstan Brown & Marina Chumakina, ‘What there might be and what there is: an introduction to Canonical Typology’, conveniently summarizes the basic features of the approach called ‘Canonical Typology’ (CT). By and large, CT is one of the theoretical approaches developed within Prototype Theory, where, in contrast to more traditional (Aristotelian) categorizations operating with necessary and sufficient conditions (as, for instance, in Mel’čuk's work), categories are characterized as being closer to or further away from the prototype (or canonical ideal), thus being more or less prototypical (canonical). This leads to a scalar notion of categories. In accordance with these prerequisites, the key concepts of CT include: (i) the base, which delimitates the domain of the category or categories under study; (ii) criteria, which serve to characterize instances of a given category actually attested in languages as more or less canonical; and (iii) the canonical ideal of a category. The authors take as an illustration the category of agreement, for which the canonical criteria include C1 ‘Controller present’ (that is, cases where the controller of agreement is absent in the sentence instantiate less canonical agreement), C2 ‘Controller has overt expression of agreement features’ (for instance, the Russian examples on pisal ‘he wrote’ and ona pisal-a ‘she wrote’, with the subjects on/ona ‘he/she’ overtly marked for gender, instantiate more canonical agreement than ja pisal ‘I (masculine) wrote’ and ja pisal-a ‘I (feminine) wrote’, with the subject ja ‘I’), C3 ‘The agreement controller is consistent in the values it takes’, and C4 ‘The part of speech of the controller should be irrelevant’ (for a detailed discussion of these criteria and further examples, the reader can be referred to Corbett Reference Corbett2006). They further explain, in terms of a Boolean lattice, how these criteria can be used to order attested instances of a category: the canonical ideal of a category meets all the criteria (C1/C2/C3/C4), while the situation where one of these criteria is not met (e.g. there is no overt expression of agreement features on the controller of agreement), C2/C3/C4, C1/C3/C4, C1/C2/C4, and C1/C2/C3, implies going down one level below; and so on. Here, I am afraid, the authors somewhat oversimplify the situation. Presumably, criteria may differ considerably in their ‘weight’ so that, for example, if C1 is ‘heavier’ than C2 (see below on their interdependency), C1/C3/C4 should be considered more canonical than C2/C3/C4 (albeit being on the same level of non-canonicity); perhaps even C1/C2 may be more canonical than C2/C3/C4, etc. Furthermore, note also that some criteria may be not entirely independent from each other. Thus, in case of the agreement criteria mentioned above, C2 (‘Controller has overt expression of agreement features’) obviously applies only in cases where C1 (‘Controller present’) holds, which again seems to point to their unequal weights.
Finally, the authors introduce a few useful principles of CT, such as Recognizability Precept (canonical ideals must be close to our traditional/intuitive understanding of the category (9)) or Venus Precept (the best known and readily available examples of a category are not necessarily the most canonical (9–10)).
The identification of the three types of issues relevant for CT (i.e. the definition of the base, criteria, and the canonical ideal of a category) ensures a very clear outline of the book and makes the aims of each chapter easily identifiable in terms of this trichotomy. Thus, Oliver Bond (‘A base for canonical negation’) defines the domain of linguistic negation in terms of the contrast between the state of affairs in communicated vs. alternate reality. He further establishes 18 criteria for canonical negation, which are divided into three groups: (i) structural criteria (such as structural symmetry, that is, formal parallelism between the negative and affirmative constructions – for instance, in languages such as Ket or Russian, where the only morphosyntactic difference between the negative and affirmative sentences is the presence of a negative particle; free negator (typically, negative particle) and segmentable negator (contrasted, in particular, with non-segmentable negators in the case of portmanteau morphemes)); (ii) applicability criteria (obligatory, productive, etc.); and (iii) semantic-pragmatic criteria (wide scope, focus on binary contrast, etc.). In addition, the author dwells on a few important theoretical issues, which include distinguishing between two types of Canonical Typology; the same dichotomy must, presumably, apply to typology in general. The first type, called ‘exploratory CT’ (quite an infelicitous term, since every typology is supposed to explore evidence from natural languages), establishes criteria ‘on empirical grounds’ (24), while another type, ‘retrospective CT’, focuses on reviewing the results of existing typology. In other words, one might say that the former type is oriented from data to theory, while the latter proceeds from theory to data. However, I think that it is a mistake to sharply oppose to each other these two types of typology. What the author presents as two alternative types of typology should rather be considered two stages of typological work that might be called empirical and theoretical. Of course, in any particular moment of his/her research, a typologist may focus on one or another stage of research (and, accordingly, this particular part of his/her research may be published in the form of an article), but, ideally, both methods should be combined (and even recurrently applied) within the same research. This combined approach to typology, when the typological description of a linguistic category is built on the basis of empirical evidence obtained from languages (‘exploratory’, or empirical, stage), then checked against new evidence, which leads to new theoretical revision (‘retrospective’ stage), and so on, in the form of ‘shuttle motion’, is vividly described in Nedjalkov & Litvinov's (Reference Nedjalkov, Viktor P., Shibatani and Bynon1995) excellent overview of the St Petersburg/Leningrad Typology Group.
Greville G. Corbett (‘Canonical morphosyntactic features’) focuses on the CT-approach to the most basic morphosyntactic categories, case, person, number and gender (which he calls ‘features’) and arrives at an interesting hierarchy of canonicity of these categories, comparing them in terms of their interaction with canonical parts of speech. According to this hierarchy, number, as the most canonical ‘feature’, is placed before the three others, which show several deviations from the canon. This fact calls for a cognitive explanation (is this due to the fact that number more directly refers to extra-linguistic phenomena?). There is another interesting observation in this chapter, which requires a short comment. The author notes the non-canonical character of the neuter gender in the Surselvan dialect of Romansh on the basis of the fact that it only surfaces in the verbal agreement with the sentential subject, the demonstrative pronoun quei ‘that’ and the impersonal pronoun igl ‘there’, and concludes that ‘the adjective has access to all three values of the gender feature [i.e., masculine, feminine and neuter – LK], while the noun has access to two only’ (63). Here it will be in order to remind the reader that there are two categories of gender – the classificational, or lexical, gender, which normally is a (lexical) feature characterizing substantives, and the inflectional gender, manifested, in particular, in adjectival and verbal agreement. Of course these two categories are closely connected with each other, so that the former triggers the manifestation of the latter, but it is not quite correct to merge them into one category with two different accesses. This important theoretical distinction was very clearly posited, in particular, in the Russian linguistic tradition by Zaliznjak (Reference Zaliznjak1964) and adopted, in a different terminology (‘controller gender’ vs. ‘target gender’), by Corbett (Reference Corbett1991: 150–154; see also, among many others, Beard Reference Beard1995: 106), but is, often and quite regrettably, neglected in contemporary linguistics.
Nicholas Evans (‘Some problems in the typology of quotation: a canonical approach’) discusses in his chapter criteria of canonical direct and indirect speech, offering a new typology of quotation and extending the traditional dichotomy direct/indirect speech by a new member, ‘biperspectival speech’, ‘in which constructions simultaneously represent two distinct viewpoints’ (88) – that of the original speaker and that of the reporter. Linguistic phenomena involving biperspectival speech include logophoric pronouns (referring both to the speaker in the reported speech event and to the third person in the primary speech event) and complex tenses (for instance, in many Western Indo-European languages) referring to two reference points (as in John said he would be at work).
Irina Nikolaeva (‘Unpacking finiteness’) elucidates morphological, syntactic and semantic criteria of canonical finiteness. Again, as in Chapter 1, one may ask if different criteria have equal ‘weight’ – for instance, should we not rank the criterion of subject agreement higher than (many) other morphological criteria, such as tense marking?
Andrew Spencer & Ana Luís (‘The canonical clitic’) offer an elegant definition of clitics as linguistic forms that are halfway between words and affixes, meeting canonical form criteria for the latter (in fact, only one such criterion is mentioned: ‘An affix consists of a monomoraic CV syllable’ (128)) and distributional criteria of the former. This combined set is supplemented by an additional form criterion: ‘A clitic is prosodically unspecified and hence is prosodically dependent on some other adjacent element’ (130).
Anna Siewierska & Dik Bakker (‘Passive agents: Prototypical vs. canonical passives’) address a very interesting methodological issue: Can a feature be considered as a criterion of canonicity even in spite of its low frequency? Such is the case of the passive with an overt agent, which is much less frequent than agentless passives, yet the former is considered closer to the prototype by many scholars. One convincing reason for taking passive with an overt agent as more canonical is neatly formulated at the very beginning of the article: ‘the property in question distinguishes the given construction from canonical realizations of all or most other constructions’ (153). The second reason is that ‘the property in question tends to coincide both across and within languages with yet other properties which may be seen to be more canonical of a given construction’ (154). The bulk of this paper (which is the longest in the volume) is dedicated to substantiating this claim for passives with agents on the basis of a detailed scrutiny of all relevant aspects of passive.
Martin Everaert (‘The criteria for reflexivization’) outlines a number of features of canonical reflexives, which fall into four groups: the properties of the binder (antecedent), the properties of the bindee, the morphosyntactic encoding, and the domain of the binder–bindee relation. One of the important conclusions formulated in this chapter is that we have to make a strict distinction between the notions of reflexivization and anaphor.
Irina Nikolaeva & Andrew Spencer (‘Possession and modification – a perspective from Canonical Typology’) arrive at an important theoretical result. On the basis of the CT-analysis of rich language evidence, they place possessor constructions of the type the girl's hand and attributive expressions such as (the) tall girl as ranking along the same scale of canonicity, where inalienable possessives are considered to represent the canonical ideal, ranking over alienable possession, modification by noun and attributive modification.
The concluding chapter, by Scott Farrar, ‘An ontological approach to Canonical Typology: Laying the foundations for e-linguistics’, outlines further perspectives of CT, heavily depending on computer processing of linguistic data and storing them in accordance with a General Ontology for Linguistic Description (GOLD). The chapter gives some details of logical formalisms used for achieving this task, discussing also several issues of linguistic modelling in GOLD.
The book closes with a cumulative list of references and indices.
The volume is a valuable overview of the research in one of the influential contemporary linguistic theories at the intersection of linguistic typology and grammatical theory. It can also serve as a helpful introduction to the theory of grammatical categories, offering definitions and discussions of several key concepts of morphology, syntax and linguistics in general, and drawing attention to a number of important methodological issues. Both students and professional linguists will undoubtedly benefit from reading this book.