The phenomenon of predication is fundamental in language. It refers to the relation between a predicate and a subject. In a predicational copular sentence such as Mary is a doctor, the complex predicate is a doctor is denoted by two items: the functional verb be, which specifies features such as tense and aspect, and the nonverbal predicative part a doctor, which contributes lexical content to the complex predicate. Not only nouns but also adjectives and prepositional phrases can serve as nonverbal predicates. Cross-linguistically there is variation in the realization of predicational copular sentences: the predicate may occur with or without an indefinite article (e.g. German and French), it can have variable morphological marking (e.g. Russian), and it may be used with different copula verbs (e.g. Spanish and Modern Irish). In the last decade many analyses have been proposed for particular languages to account for such differences. The monograph by Isabelle Roy on nonverbal predication, which is a revised version of her doctoral dissertation at the University of Southern California, is a one-of-a-kind study that on the basis of a formal analysis of predication at the syntax–semantics interface in French develops a fine-grained cross-linguistic typology of nonverbal predication in general. The suggested ternary typological classification of predicational copular clauses as defining, characterizing and situation-descriptive on the basis of the semantic criteria maximality and density can account for language-specific differences more successfully than the common distinction between individual-level and stage-level predicates.
The book is organized into three parts, each of which consists of two chapters. The book begins with Part I, ‘Predicational copular sentences at the syntax-semantics interface’. Chapter 1 gives a brief overview of the book. Chapter 2 outlines the current state of the art on predicational copular sentences and introduces the main assumptions. Roy claims that predicational sentences have to be distinguished from other types of sentences with ‘be’ such as the identificational or specificational types since only sentences of the former type express a predication relation and have a non-referential post-copular expression that denotes a property. For the syntactic representation Roy assumes that the predication relation between the subject and the nonverbal predicate is established in the hierarchically organized small clause PredP as proposed by Bowers, among others, where the predicative expression YP serves as a complement of the head Pred.
(1) Johni is [PredP ti Predo [YP a doctor / sick / in his car]]
Contrary to previous suggestions in the literature Roy assumes that the copula itself does not establish a relationship between the subject and the predicative expression but rather is a raising verb generated in To that takes PredP as its complement. Semantically, Roy assumes a neo-Davidsonian event-based approach to predicates according to which nonverbal predicates express relations between individuals and states as a type of eventuality.
Part II, ‘Meaning and structures’, is the main part of the book. In Chapter 3 Roy introduces a typology of nonverbal predicates based on their interpretation. The three types of predication, namely situation-descriptive, characterizing, and defining, can be distinguished by potential questions:
(2) Mary is sick /in the garden. (What is going on?)
(situation-descriptive)
(3) Mary is a doctor/intelligent. (What/How is Mary?)
(characterizing)
(4) Mary is a doctor. (Who is Mary?)
(defining)
Situation-descriptive sentences differ from other types in that they do not ascribe a property to an individual, but describe situations. Situation-descriptive predicates can be any category except nominals. In characterizing and defining sentences, the property denoted by the nonverbal predicate is ascribed to an individual. A defining property, unlike a characterizing property, must be salient enough to ‘define’ an individual as a particular member of a class of individuals. Characterizing predicates may be expressed by nouns and adjectives. Roy shows (38) that in French, the difference between defining and characterizing sentences is grammaticalized in the presence vs. absence of the indefinite article with predicate nouns: while defining nominal predicates need an indefinite article, characterizing predicates are bare.
(5) Raymondestunacteur.(QuiestRaymond?) (defining)
RaymondisanactorwhoisRaymond
‘Raymond is an actor. (Who is Raymond?)’
(6) Raymondestacteur.(Qu'estRaymond?) (characterizing)
Raymondisactorwhat.isRaymond
‘Raymond is an actor. (What is Raymond?)’
Roy claims that the distinction between bare and non-bare nominal predicates cannot be reduced to the well-known distinction between individual-level and stage-level predicates since bare NPs may denote transient properties (e.g. acteur ‘actor’) as well as permanent ones (e.g. père de famille ‘a family man’). The distinction can also not reside in the type of the clause, predicational (with a bare NP) vs. identificational or identity (with a non-bare NP), since identificational and identity clauses are assumed to involve two referential NPs. Roy argues that non-bare NPs such as in (5) above are non-referential.
Further distinctions between characterizing and defining predicates show up in the impossibility of the defining predicate serving as a small-clause complement of verbs like considerer ‘consider’ or occurring in sentences with grammatically or lexically specified aspect, and in the availability of the so-called life-time effect in the past. Roy claims that these and other differences between the types of predicates are determined by aspectual properties of the state described by the nonverbal predicate. Contrary to the common view of states as atomic entities, Roy suggests differentiating types of states according to their internal aspectual properties. Roy assumes that defining predicates denote maximal states, i.e. there are no smaller states which could be proper parts of them. The other types of predicates, characterizing and situation-descriptive ones, are not maximal but are interpreted as having spatio-temporal subparts. Situation-descriptive predicates are distinguished from the characterizing (and defining) ones by the feature of density. Density relates to whether a predicate holds for every subpart of a particular eventuality or not. For example, the characterizing property of being a drunkard is not-dense since the state of being drunk need not hold for the subject at the moment of utterance – it allows for ‘gaps’. In contrast, the situation-descriptive property of being drunk is dense since it holds for every subpart of the eventuality – ‘gaps’ would render the predicate false. Predicate nouns always allow ‘gaps’, hence they are not dense. This explains why nouns never serve as situation-descriptive predicates.
Chapter 4 shows how the semantic distinctions between the three types of nonverbal predicates introduced in the previous sections map onto syntactic structures. Roy assumes that dense, non-dense, and maximal predicates differ in the internal syntactic structure. Dense predicates do not involve any functional heads. Non-dense predicates involve a classifier head Clo, which divides the eventuality denoted by the nonverbal predicate into subparts. Maximal predicates involve a number head Numo, which introduces a maximal quantity of an eventuality.
(7) [XP] dense(situation-descriptive)
(8) [ClP […]] non-dense(characterizing)
(9) [NumP […]] maximal(defining)
Roy claims that in French the numeral head of maximal predicates Numo is realized by an indefinite article. The assumption that maximal predicates are NumPs accounts for the fact that they carry independent number specification and do not agree with the subject in number. The incompatibility of maximal predicates in defining sentences with grammatically or lexically specified aspect is derived from the lack of AspP in this clause type. The eventuality argument provided by Predo, which in situation-descriptive and characterizing sentences is existentially bound by Aspo, is bound in defining sentences by the maximality operator in Numo. This leads to the maximal eventuality interpretation. Since the maximality operator is strong, NumPs are strong expressions. As such they must be interpreted in the restrictor of a sentence's Logical Form, i.e. outside the VP or PredP. Roy uses this assumption to explain the exclusion of maximal predicates from the PredP complement of lexical verbs such as considerer ‘consider’ and the availability of life-time effects in the past. Both phenomena are accounted for as effects of the obligatory raising of the strong predicate outside the PredP triggered by the strong maximality operator. However, the assumption that predicate expressions may be strong is not unproblematic. I will come back to this issue below.
In Part III, ‘Further analysis’, the author shows how the distinction of dense/non-dense/maximal states introduced with the example of French is grammaticalized in Russian (Chapter 5), Spanish (Chapter 6) and Irish (Chapter 7). Unlike in Russian and Spanish, nonverbal predicates in Irish do not map straightforwardly onto the three-way distinction. Table 1 summarizes the comparison between French, Russian and Spanish. In Chapter 8, ‘Conclusion’, Roy briefly recaps the main points of the monograph.
Roy's monograph is a very impressive, carefully worked-out, and clearly written study. While I consider many of Roy's findings very interesting and thought-provoking, I have some points which deserve more detailed discussion.
One point concerns locative modifiers as a test for dense (situation-descriptive) predicates. Roy claims that only dense predicates can be combined with locative modifiers under the reading where they serve to spatially locate the state denoted by the nonverbal predicate such as in (10a) and (10b). According to Roy, if locative modifiers occur with non-dense (characterizing) predicates such as in (11), they can only get a temporal interpretation instead of a locative one.
- (10)
(a) Paul was drunk (in the kitchen). (dense)
(b) The sky is blue (outside/in Morocco). (86) (dense)
- (11)
(a) Paul is tall (#in his car) (non-dense)
(temporal interpretation ok ‘Paul is tall when seated in his car.’)
However, Maienborn (Reference Maienborn2005) has convincingly shown that in predicational copular sentences locative modifiers never serve to spatially locate the state denoted by the predicate, but rather they can only serve as frame-setting modifiers restricting the overall proposition. Real locative modifiers with spatial meaning such as in (12) cannot occur in predicational copular sentences.
(12) The dress is (*on the clothesline) wet.(Maienborn Reference Maienborn2005: 292)
Unlike locative modifiers, frame-setting modifiers as in (10) and (11) above, being sentential modifiers, provide a semantically underspecified domain restriction for the overall proposition. Depending on the context, they can yield at least a temporal interpretation (for (10a), ‘When he was in the kitchen, he was drunk’), or an interpretation as subject restrictors (for (10b), ‘The sky outside/in Morocco is blue’). Since frame-setting modifiers may combine with dense as well as with non-dense predicates, such modifiers are not an appropriate density diagnostic. This becomes obvious in the classification of the predicates pregnant and dead (85, 86). Since they are incompatible with real locative modifiers, Roy classifies them as non-dense, at least in one reading. However, such predicates do not allow ‘gaps’ (episodes where the predicate does not hold). This suggests that they are dense. Hence, local modification is not a test for density and a reliable linguistic diagnostic for density is still needed.
Another point concerns the analysis of defining predicates such as in (4) as maximal eventualities in (9). According to Roy (101), since defining predicates refer to maximal states, they are strong. Strong NPs, however, are commonly assumed to be referential. Since non-verbal predicates are semantically predicates, they are non-referential. The relation between strongness and non-referentiality for nonverbal predicates therefore deserves more discussion.
The last point concerns the lack of comparison of the suggested typology with other typologies of nonverbal predicates such as Hengeveld (Reference Hengeveld1992) and Stassen (Reference Stassen2003). This relevant work is not even mentioned in the book. However, a comparison with these extensive typological studies could provide the author with an opportunity to discuss her own typology from a broader perspective.
Overall, despite the points mentioned above the book by Isabelle Roy makes a significant contribution to the investigation of nonverbal predication and will surely find many interested readers within the theoretical linguistics field as well as among those interested in French, Russian, Spanish, and Irish. Its fairly comprehensive scope and the overall good balance between technical detail and accessibility make it an important reference source not only for specialists but also for students.
