This book offers different papers focusing on the issue of identity within inflectional paradigms. It is thematically divided into three major sections. The first contains papers discussing identical (syncretic or homophonous) forms appearing within inflectional paradigms, written by scholars who are wary of appealing to the notion ‘paradigm’ within their morphological analyses. The other two sections focus on possible sources of identity effects: identity through shared morphological features as revealed through decompositional featural analysis, and identity through shared dependence on a certain base.

In chapter 1, ‘Introduction: Approaching inflectional identity’, Asaf Bachrach & Andrew Nevins lay the theoretical foundations uniting the disparate themes that permeate the different papers. Bachrach & Nevins raise two important questions about inflectional identity effects that must be addressed by paradigm-based theories: the asymmetry question (‘Why do inflectional identity effects go from some members in the paradigm towards others, and not vice versa?’) and the inclusion question (‘What is the set of relevant forms that learners put together into the set of inflectionally-related elements?’) (3). While summarizing each of the included papers in light of the major themes, Bachrach & Nevins express a well-motivated desire to establish ‘a rigorous formalization of the principles governing the formation of the paradigms and “mini-paradigms” used in invoking identity effects’ (7), and they convey a skepticism about the need to actually appeal to the ‘paradigm’ or ‘powerful inter-derivational output-based constraints which are predicated over paradigms’ (7) in morphological explanations.

In chapter 2, ‘Paradigms (Optimal and otherwise): A case for skepticism’, Jonathan David Bobaljik argues against the Optimal Paradigms (OP) theory of McCarthy (Reference McCarthy, Downing, Hall and Raffelsiefen2005), which seeks to account for disparities in the allowable stem forms of Arabic nouns and verbs by appealing to paradigm uniformity constraints. Unlike noun stems, verb stems in Arabic are restricted to ending in CVC] (and never *CV:C] or *CVCC]). OP attributes this disparity to the possible shapes of the inflectional affixes with which a given stem must be able to combine, by proposing ‘a species of output-output faithfulness constraints that place a premium on a stem keeping a constant shape throughout its inflectional paradigm’ (33). In the OP account, Arabic noun stems only have to combine with V-initial affixes and so may be of a variety of different shapes without creating words that either violate the syllabification constraints of the language or would violate paradigm uniformity in order to meet those syllabification constraints. Verb stems, on the other hand, take both V-initial and C-initial inflectional suffixes and so are limited to ending in CVC]. Bobaljik argues that Arabic is a less than ideal language for which to establish OP, however, since supposed OP effects cannot be distinguished from morphosyntactic category membership in that language. He presents data from Itelmen which suggests that it is in fact category membership, and not ‘paradigm uniformity’, that predicts when a certain phonological rule, i.e. epenthesis of schwa to break up consonant clusters involving a sonorant consonant, will overapply. Although the Itelmen data may also be amenable to an OP-style analysis, Bobaljik suggests that, contrary to what he calls the ‘thesis of category neutral phonology’, other phonological processes in the world's languages may also refer to syntactic category membership, and thus the necessity of referring to paradigm membership, as in the OP analysis of Arabic, has not yet been convincingly established.

In chapter 3, ‘Clarifying “Blur”: Paradigms, defaults, and inflectional classes’, Morris Halle & Alec Marantz argue against the No Blur Principle of Carstairs-McCarthy (Reference Carstairs-McCarthy1994), which holds that ‘within any set of competing inflectional affixal realizations for the paradigmatic cell, no more than one can fail to identify inflection class unambiguously’ (57). According to Halle & Marantz, Carstairs-McCarthy's No Blur analysis of Polish noun declensions depends on two additional assumptions: (i) that stems may only belong to one inflectional class (a principle that might or might not exist, but which in any case is distinguishable from and, crucially, relied upon by No Blur itself), and (ii) that there can be no accidental homophony among affixes within a given paradigm. The latter principle is intended to rule out ‘blurred’ paradigms like that in (1), where two homophonous affixes, -owi ₁ and -owi ₂, arbitrarily divide Polish nouns into two classes (1 vs. 2).

1. (1) Partial Paradigm for Polish Noun Inflection (adapted from 59, ex. (3.2))

   

No Blur is avoidable by allowing affixes to cross-classify noun stems, as in (2), where the dative (DAT) suffixes -owi and -u classify noun stems into two numerical classes (1 vs. 2), while the Locative suffixes -e and -u classify noun stems into two alphabetical classes (A vs. B) (64–65).

1. (2) Partial Paradigm for Polish Noun Inflection (60, ex. (3.4))

   

In Halle & Marantz's view, such cross-classification makes No Blur superfluous. The authors provide an alternative Distributed Morphology (DM) account of the Polish noun declensions, wherein they posit -u as a default Vocabulary Item (VI) (case affix) which gets inserted into ‘elsewhere’ environments but also into certain contexts where case features are deleted by Impoverishment.

In his contribution, ‘Paradigm generation and Northern Sámi stems’ (chapter 4), Peter Svenonius accounts for Northern Sámi stem alternations that involve a complex range of different, but phonologically predictable, interactions between stems and certain inflectional suffixes in the ‘Strong Grade’. Svenonius proposes that Strong Grade suffixes include an extra timing slot (i.e. a mora) infixed between the suffix itself and the stem. The realization of this morphologically-derived mora is based on the phonological environment resulting from the addition of the suffix to the stem. Other cases of morphological mora affixation are attested (cf. Samek-Lodovici Reference Samek-Lodovici1992), but the Sámi case interestingly involves a bare mora concomitant with an overt affix. By sketching the complex phonological rules necessary to give phonetic content to this extra mora in the Strong Grade (e.g. /l/ before /l/ and /h/ before /k/, but, unexpectedly, /k/ in the environment of /v/), Svenonius's discussion substantially adds to an emerging typology of morphological ‘mora augmentation’ (cf. Davis & Ueda Reference Davis and Ueda2006).

Artemis Alexiadou & Gereon Müller examine case syncretisms in Russian, Greek, and German in chapter 5, ‘Class features as probes’. They argue that instances of transparadigmatic syncretism can be explained with reference to natural classes of shared features, just as is assumed for intraparadigmatic syncretism. An example of the latter is nominative and accusative case, which are both [−oblique], allowing syncretisms to occur for these cases within a given paradigm (e.g. Russian nouns in Class III (which take Ø for both cases), Class IV (which take -o), and Class I inanimate nouns (which take Ø)). Contra standard accounts appealing to privative class features (e.g. ‘I’, ‘II’, etc.), however, Alexiadou & Müller claim that class features are also decomposable into abstract, non-interpretable binary features (e.g. [+/−α], [+/−β]). This allows for transparadigmatic syncretism, as with Russian Class I and Class IV nouns, which share the proposed [+α] and have syncretism for the singular dative (-u), genitive (-a), instrumental (-om), and locative (-e). In this analysis, case affixes are Vocabulary Items which must compete for insertion via the Subset Principle, where the affix with the most specific matching features is inserted, and where features are ranked according to the hierarchy of number>>class>>case. Unlike in DM, however, in Alexiadou & Müller's account the uninterpretable class features of noun stems act as probes for agreement with the matching features of an inflection marker (the goal). In order to avoid violations of either the Legibility or Inclusiveness Conditions of Chomsky (Reference Chomsky and Kenstowicz2000), these uninterpretable features must be deleted before syntax, a result which forces Alexiadou & Müller to claim that morphology precedes syntax, contra DM. One drawback to Alexiadou & Müller's proposal is that it does not seem able to account for accusative/genitive case syncretism in Russian animate plurals. Although the authors dismiss paradigm-based theories without argument in footnote 1 (103), they do resort to using the paradigm-based notion of ‘rules of referral’ in this situation, without addressing the possibility that such rules of referral, if necessary, might also be sufficient to account for case syncretisms generally.

In chapter 7, ‘A feature-geometric approach to Amharic verb classes’, Jochen Trommer utilizes a similar decompositional approach to class features in order to account for syncretisms in verbal paradigms in the Ethiopian Semitic language Amharic. In Trommer's account, the arbitrary features (~ α or β, as in Alexiadou & Müller's chapter) classify verb roots based on the final vowel and on whether gemination is triggered in certain paradigm cells. By organizing these features into a geometrical formation where each node can have only one node dominating it, Trommer is able to account for the limited number of attested classes. He goes on to argue that syncretisms across classes in derived contexts result from Impoverishment, where the insertion of zero Vocabulary Items into particular nodes deletes all class features above those nodes. This feature-geometric approach accounts for the attested data while ruling out possible but non-existent feature combinations that would be predicted to exist with unordered feature bundles.

In chapter 6, ‘On absolute and contextual syncretism: Remarks on the structure of case paradigms and on how to derive them’, Andrea Calabrese introduces a distinction between contextual syncretism, which involves a syncretism between cases that are morphologically realized in a given language, and absolute syncretism, which involves ‘the syncretism between a case that is morphologically realized in [a] language and a case that is not morphologically expressed in that language but that is morphologically expressed in other languages’ (163). Calabrese proposes that there is a universal set of case features and a set of ‘feature-changing operations’ that can apply to feature bundles as ‘repairs triggered by the negative constraints on feature combinations’ (178), wherein non-allowed combinations (within a given language) are adjusted to become allowed combinations via repair operations. Calabrese intends to constrain this model by appealing to comparative markedness, but his notion of ‘absolute syncretism’ and overall approach will have to be supported with a much more detailed survey of the typology of case syncretisms than Calabrese provides here.

John F. Bailyn & Andrew Nevins discuss Russian case inflection in chapter 8, ‘Russian genitive plurals are impostors’. Most case endings in Russian support two central hypotheses posited by Bailyn & Nevins: (i) that gender and class distinctions are neutralized in marked environments (e.g. with oblique case and/or in the plural), and (ii) that allomorphy is locally determined within the derivation (242). Previous analyses of the Russian genitive plural abandon one of these hypotheses, either by proposing that the genitive plural refers to gender features, which makes it unique with respect to other Russian cases where gender and class features are neutralized in the plural, or by proposing a transderivational analysis whereby the form of the genitive plural is somehow derived from the nominative singular. Bailyn & Nevins put forward a new analysis of Russian noun stems involving theme vowels visible in the nominative singular. Thus, the traditional analysis of kn'iga ‘book, nom.sg’ as /√kn'ig/ plus a nominative singular case ending /-a/ is replaced by an analysis of kn'iga ‘book, nom.sg’ as /√kn'ig/ plus a theme vowel /-a/ plus the case and number ending /Ø/. Theme vowels (a, o, or Ø) delete through regular phonological rules, either by suffixation of subsequent overt case endings (e.g. -e in the dative singular, yielding kn'ig'e), or after suffixation of yer (ъ), which yields truncation of the theme vowel so that the bare root appears in the genitive plural (e.g. kn'ig).

Adam Albright sums up the point of chapter 9 in his title, ‘Inflectional paradigms have bases too: Arguments from Yiddish’. He argues that the loss of word-final devoicing in Yiddish singular nouns resulted from paradigmatic pressure from corresponding plurals. Middle High German had two kinds of nouns ending with voiceless consonants: those that had a voicing alternation in the plural (e.g. wëc [k] ‘way’~wëge ‘ways’) and those that did not (e.g. druc ‘pressure’~druckes ‘pressures’). An account that proposes a historical change involving blanket demotion of word-final devoicing would incorrectly predict voicing of final consonants in both word types. Albright argues instead that a move toward paradigm uniformity, based on the form of the plural, explains the loss of the alternating word-final voiceless consonants. This analysis is supported by the fact that voicing was restored only to lexical roots (i.e. all of the suffixes of the language retain word-final voiceless consonants), and non-paradigmatically related forms of lexical roots, e.g. derived forms like avek ‘away’ (from /a+veg/ ‘way’), have word-final voiceless consonants as well. Although it may seem intuitively odd that the plural form would serve as an inflectional ‘base’ for the singular, Albright offers a reasonable hypothesis that ‘learners identify the part of the paradigm with the most information and focus on that form to learn the properties of words’ (300). In the case at hand, it was the plural that ‘most clearly exhibits lexical contrasts and extending the plural variant does the least violence to recoverability’ (300).

The tenth and final chapter is ‘A pseudo-cyclic effect in Romanian morphophonology’, by Donca Steriade. Steriade proposes that phonological alternations in Romanian involve inflection dependence, a situation where ‘segmental alternations are permitted in the derivatives of a lexeme only if certain inflected forms of that lexeme, its inflectional bases … independently display that alternation’ (313). Her analysis crucially relies upon a layered lexicon composed of at least a core lexicon containing ‘basic atomic entries (roots and affixes)’ and a derived lexicon ‘containing morphologically complex words, organized in successive layers’ (315). In Steriade's view, the phonological cycle is insufficient to account for inflection dependence, and she proposes instead that the derived lexicon makes available the full range of inflected forms as ‘reference terms in the generation of other complex forms’ (326). One issue for Steriade's proposal is its reliance upon a nebulous notion of ‘lexical relatedness’, her use of which approaches circularity. For example, a contextually-restricted palatalizing plural [vétʃ-] in the phrase [pe-vétʃ-] ‘forever’ (lit. ‘for ages’) is claimed to be lexically unrelated to an apparently related but non-alternating (i.e. non-palatalizing) form, [vak] ‘century, extended time interval’, on the grounds that the latter is non-alternating and so must be unrelated, cf. plural [vák-ur] ‘centuries’ (328).

In sum, this stimulating book presents a variety of perspectives on the issue of inflectional identity within paradigms. The range of novel theoretical proposals offered and the diversity of empirical data covered should be appealing to morphologists of all stripes.