This book, which aims to provide the state-of-the-art in research on exponence, grew out of an international network named ‘Core Mechanisms of Exponence’ that organized regular meetings and brought together a group of morphologists and phonologists working around this theme. The term exponence refers to the relation between abstract morphosyntactic information and the realization of this information in the phonology. Consequently, one of the recurring themes in the book is the division of labour between morphology and syntax on the one hand and phonology on the other. Jochen Trommer observes in his introduction that despite all the differences in approach and theoretical assumptions, there are two more or less hidden points of convergence. First, most researchers agree that non-concatenative morphology and reduplication should follow from a so-called Item-and-Arrangement model, rather than from an Item-and-Process model. Since the work of McCarthy (Reference McCarthy1981) Item-and-Process models slowly have disappeared from the main stage and the development of autosegmental models and Optimality Theory has made it possible to successfully analyze non-concatenative patterns and reduplication in terms of Item-and-Arrangement. The book ends with a contribution by Patrik Bye & Peter Svenonius who aim to bring this development one step further by trying to eliminate, as is also done by Generalized Template Theory (McCarthy & Prince Reference McCarthy, Prince, Kager, van der Hulst and Zonneveld1999), reduplicative templates altogether and to radically derive non-concatenative morphology from independent syntactic and phonological properties of the language.
A second point of convergence noted by Trommer is the ‘triumph of realizational morphology’ (3). By this he refers to the by now generally accepted idea that morphology is the spell-out (realization) of abstract morphosyntactic information by markers that are themselves often underspecified for such morphosyntactic information.
For a reviewer it is almost commonplace to say that it is impossible to do justice to all the individual papers in such a collection. But in the present case, with nearly 500 pages all by top experts in the field comprising lengthy papers with often the most detailed and subtle analyses of complex data spread over 12 chapters, two of which present new models (Bermúdez-Otero and Bye & Svenonius), it is truly an impossible task to give proper attention to all the individual papers. Thus, I will have to refrain from discussing each chapter separately, but I will try to sketch some general lines in the research that seem to recur in many contributions of this book and that may determine the research agenda for the next years. Some of these questions have a longer history in the field, others are quite new and can only be formulated now that we have a better understanding of certain empirical data and have the notions needed to address these issues.
One of the older issues concerns the question, already alluded to above, whether all morphological types of exponence can be reduced to the simple concatenation of morphological elements. In their contribution, ‘Non-concatenative morphology as epiphenomenon’, Bye & Svenovius make the radical proposal that everything that makes non-concatenative morphology ‘special’ should follow from independently motivated syntactic and phonological properties of the language. One of the ideas is that syntax obviates the need for morphological subcategorization and morpheme-specific alignment constraints. So, in this way, part of the work needed for exponence is done by syntax, simplifying the morphophonological part of the job, making it possible to eliminate constraints that are not independently motivated. Building on important insights from Saba Kirchner (2010), the aim is to show that reduplication also follows from independently motivated phonological constraints. Saba Kirchner shows that copying of segmental material in the phonology is sometimes preferred over epenthesis of a ‘default’ vowel. The vowel in the input then does ‘double duty’; it has two output-correspondents: the ‘original’ one in the same position as in the input and a copied one in the epenthetic position. This structure violates a constraint integrity that mitigates against ‘double correspondences’. By ranking this constraint below the markedness constraint that forces the epenthesis and the *Dep constraint that requires all segments in the output to be dependent on the input segments, copying emerges as the optimal phonological solution. Now, if reduplication can be made to follow from this type of copying, there is no need for ‘reduplicants’ as templates or for specific constraints to define the base-reduplicant correspondence as in the standard approaches to reduplication. A segmentally underspecified syllable, or foot or mora, can be suffixed and the possibility of having two correpondents results in the ‘copying’ of the segmental information. More difficult to incorporate in this framework are cases of truncation. Bye & Svenonius suggest that truncation is not really a serious threat though since it seems to be quite marginal if it exists at all. However, Birgit Alber & Sabine Arndt-Lappe in their contribution argue quite strongly for a restoration of truncation as a real means of exponence. In fact this issue goes back to Anderson (Reference Anderson1982), where subtractive morphology is presented as problematic for Item-and-Arrangement models of morphology. Alber & Arndt-Lappe distinguish two types of truncation. The first type is templatic in the sense that the output conforms to some specified template. These cases, that were part of the empirical motivation for the theory of Prosodic Morphology (McCarthy & Prince Reference McCarthy and Prince1993), involves hypocoristics of names as English Pat and Trisch from Patricia and there is some discussion whether this should be considered real cases of morphology. The reason is that these templatic cases of truncation always have affective or other ‘social’ meaning components such as appreciation or depreciation but not new referential meaning.
However, setting aside this issue, such forms can be derived in the theory of Prosodic Morphology. In general, a high-ranking MaxIO dominates some ‘size restrictor constraint’, which in turn dominates MaxBT. The latter constraint requires that all elements of the base have a correspondent in the truncated form, but since the size restrictor dominates this constraint, this can only be satisfied up to the maximal amount of segments allowed by this restrictive template. However, in order for such analyses to work, there should be a ‘trunc-template’, something that Bye & Svenonius explicitly try to avoid. The second type involves patterns of subtraction in which the truncated part (rather than the resulting form, as in the templatic cases) is a well-defined portion of the base. Such patterns pose even more difficulties for an Item-and-Arrangement theory of morphology, which explains why many authors have tried to reanalyze such cases as ‘normal’ additive morphology.
Another issue that recurs in this volume is one of the central topics in the contribution by Ricardo Bermúdez-Otero. It concerns the division of labor between phonology and morphology. The author shows that the current theory is underdetermined and leaves far too many options in simple analytic problems. For example, in Spanish the theme vowel in the second conjugation is normally -e, but in several cases unexpectedly an -i shows up in its place (beber ‘drink’, bebedor ‘drinker’, bebible ‘drinkable’). We now may assume that this -i is due to a rule that changes the theme vowel in the condition of the morpheme -ble (i.e. a morphologically conditioned rule) or, as in the theory of Distributed Morphology, we could propose that it is a readjustment rule triggered by the morphosyntactic context. Alternatively, we could also propose a floating autosegmental feature [high] that is part of the underlying representation of the affix -ble and docks onto the nearest available vowel of the stem -ble attaches to. Or, we could assume that the floating feature [high] is part of the second conjugation and that the affix -ble triggers a rule that associates this feature with the nearest available vowel. A completely different option would be to pose two different underlying representations (rather than a rule that derives the one from the other) and that there is a competition for insertion of these two different underlying forms. Etcetera. This wealth of analytic possibilities needs to be restricted for obvious reasons and Bermúdez-Otero proposes the following four hypotheses of which at least the first three restrict the possible analyses severely, thus contributing towards a more explanatory theory of exponence:
(i) Morph Integrity Hypothesis
Morphology is not allowed to operate directly upon elements of the phonological representation.
(ii) Indirect Reference Hypothesis
Indirect reference prevents phonological constraints from directly referring to morphological information.
(iii) Phonetic Interpretability Hypothesis
Derived phonological representations must be phonetically interpretable. This forbids the presence of morphosyntactic diacritics in phonological output representations.
(iv) Cycle Hypothesis
Morphosyntactic constituents define domains over which the phonology applies iteratively, starting with the most deeply embedded domains and moving progressively outwards.
As Trommer observes in his introduction, the paper by Bye & Svenonius is a possible implementation of this program in the domain of non-concatenative morphology. It is my impression that these four hypotheses are shared by most contributors of this volume and as such might be seen as programmatic for a shared future research agenda.
In several contributions there is attention to what we may call special cases of exponence. These special cases involve polarity, zero-exponence, iconicity, and truncation. Polarity is discussed in two chapters, one by Paul de Lacy and the other by Dieter Wunderlich. The term refers to a situation in which two affixes appear in morphosyntactically opposing contexts. A potential case is the marking of number and case in class II masculine nouns in Hindi. The plural oblique case is marked as -a:. Plural is marked by -e in the non-oblique cases, whereas oblique is also marked as -e in the non-plural cases. The affix -o marks the non-oblique, non-plural cases. Wunderlich argues that this particular case can be accounted for by assuming the existence of a kind of super-feature F that comprises the set {obl, plur}. The affix -e is specified as [–F1, +F2] meaning that it marks those morphosyntactic contexts in which exactly one marked feature (the plus value of either oblique or plural) occurs. In this way polarity can be understood in the same way that other morphosyntactic features are realised in the phonology, reducing the ‘special’ cases considerably. However, it seems that not all cases of polar morphology can be thus analysed. There are such cases as ‘full reversal’; in such cases one and the same affix marks the presence of two marked feature values or none.
Another such ‘special’ case is haplology. Andrew Nevins gives a typology of phenomena that all could fall under the umbrella of haplology but since they occur in different stages of the derivation the properties of haplology differ accordingly. Just to mention one example, Nevins assumes, following Richards (Reference Richards2010), that in the first stage of exponence structurally adjacent DPs cannot be linearized since they only differ in their phonology. At this particular early stage (i.e. ‘before spell-out’), the phonological form of the DPs cannot be addressed, and therefore linearization fails. However, at a later stage, when vocabulary items are inserted, haplology is phonologically-sensitive.
(1) the cats' feet are dirty
(2) Katz's deli
The example in (2) shows that the haplology in (1) occurs between two affixes rather than just being a phonological phenomenon. Nevins proposes that in (1) there is a zero-morpheme expressing the possessive, which takes the place of the normal possessive sibilant suffix. Given the perspective offered by Bye & Svenonius, we may wonder whether a description in terms of ‘double-duty’ of the -s affix would also be possible and maybe less stipulative.
I fully realize that the above only scratches the surface of this wonderfully rich and inspiring volume. Apart from the contributions mentioned above, the volume contains chapters by Eulalia Bonet & Daniel Harbour on contextual allomorphy, by Adam Albright & Eric Fuβ on syncretism, by Jochen Trommer on zero-exponence, by Sharon Inkelas on reduplication, and finally by Laura Downing & Barbara Stiebels on iconicity. The volume is a must-read for everyone interested in the relation between phonology and morphology. It summarizes the state of the art in research of exponence and provides many new potentially fruitful lines for future research.
