The study also highlights the constructs of current linguistic theory, arguing for distinctive features and the notion 'onset' and against some of the claims of Optimality Theory and Usage-based accounts.
The importance of Henk Zeevat's new monograph cannot be overstated. [...] I recommend it to anyone who combines interests in language, logic, and computation [...]. David Beaver, University of Texas at Austin
EDITORS: Rappaport Hovav, Malka; Doron, Edit; Sichel, Ivy TITLE: Lexical Semantics, Syntax, and Event Structure SERIES: Oxford Studies in Theoretical Linguistics PUBLISHER: Oxford University Press YEAR: 2010
Berit Gehrke, Department of Translation and Language Sciences, Universitat Pompeu Fabra
The volume brings together papers related to the work of Anita Mittwoch, most of which were presented at the workshop ‘Syntax, Lexicon, and Event Structure’, held in her honour at the Hebrew University of Jerusalem in 2006. It contains 15 contributions distributed over three parts (I: Lexical Representation, II: Argument Structure and the Compositional Construction of Predicates, and III: Syntactic and Semantic Composition of Event Structure), preceded by material including an introductory chapter by the editors. The volume is completed by references and three indices of names, topics, and languages. Data from various languages are discussed, such as English, Hebrew, Hungarian, Greek, Japanese, Blackfoot, Dutch, as well as two different sign languages employed in Israel.
In the introduction, the editors state that despite the wide range of topics covered by the papers in this volume, they all share the same research question, namely to determine the division of labour between lexical semantics, compositional semantics, and morphosyntax in the representation of events, as well as the nature of cross-linguistic variation in this area. After providing a short review of the main ingredients of event descriptions, such as lexical aspect, telicity, incrementality, argument structure, voice, viewpoint aspect, tense, mood, and habituality, they summarise the chapters and establish connections between them as well as their relation to Mittwoch’s research. The concluding section gives an overview of Mittwoch’s contribution to linguistics, starting from her 1971 dissertation, which anticipated many topics in the literature on events, such as the correlation between incremental themes and the possibility of object omission or parallels between incremental theme verbs, change-of-state verbs and motion verbs in combination with goal phrases. Particular papers by Mittwoch are discussed, which address topics that are directly picked up by some of the authors in this volume, such as optional intransitivity (Landman & Rothstein), homogeneity (Landman & Rothstein, Mittwoch), the interaction of aspectual class and temporal when-clauses (Mittwoch), habituality (Boneh & Doron), the different effect of bare plural and mass arguments on accomplishments and achievements (Borer), and cognate object constructions (Horrocks & Stavrou).
PART I: LEXICAL REPRESENTATION
In ‘Reflections on Manner/Result Complementarity’, Malka Rappaport Hovav and Beth Levin develop their hypothesis from previous works that (non-stative) verbs either lexicalise manner or result, but not both, mainly discussing data from English. They assume canonical realisation rules for the association of a root’s ontological categorisation with an event schema (made up of combinations of the predicates ACT, CAUSE, BECOME). They propose that each root has an ontological categorisation, chosen from a fixed set of types, including manner and result, and that roots are integrated into event schemas either as arguments or as modifiers of the event predicates involved. In particular, manner roots are argued to modify ACT, whereas result roots are arguments of BECOME. They furthermore posit a constraint on lexicalisation, according to which a root can only be associated with one primitive predicate in an event schema, either by modifying or adjoining to it, from which it follows that a root cannot lexicalise both manner and result. They argue that the notion of result should not be equated with the notion of telicity, but rather with the notion of scalar change. In contrast, manner verbs are argued to specify non-scalar change, i.e. change that cannot be characterised in terms of an ordered set of values of a single attribute. They suggest that the notion of scalar vs. non-scalar change and thus of result vs. manner, has a direct parallel in the motion domain, where we find complementarity between manner and path (i.e. scalar change) verb roots.
In ‘Verbs, Constructions, and Semantic Frames’, Adele E. Goldberg postulates that meanings are relativised to frames, combinations of a word sense’s profile (what is asserted) and the background frame of a word (what is presupposed). She reviews two recent proposals for constraints on verbal meanings and argues that they are both too strong and should merely be seen as tendencies. To argue against Croft’s (1991) constraint, according to which verbs can only describe simple events or complex events in which two subevents are causally linked, she discusses verbs that profile two subevents that are not causally related (e.g. ‘blanch’ asserts the two events of immersing food in hot water and then in cold water). As counterexamples to Rappaport Hovav & Levin’s manner/result complementarity, she addresses verbs that apparently specify both, such as ‘scale’ or ‘climb’ (but see Levin & Rappaport Hovav, to appear, for a way to integrate the latter into their system), or verbs of cooking and verbs of creation, e.g. ‘scribble’. Instead she proposes the Conventional Frame constraint, according to which a verb’s meaning can involve two or more subevents, only if these are related by a semantic frame. In addition, she posits that argument structure constellations (‘constructions’), can also be associated with semantic frames, and that the meanings associated with verbs and with constructions can combine into one meaning to evoke novel events which do not have to comply with the Conventional Frame constraint. Goldberg’s careful discussion of particular counterexamples to the two constraints under discussion raises the important issue that particular notions, such as cause, manner, or result, have to be defined precisely in order for the constraints to work. However, it is not clear that they warrant rejecting the constraints altogether. For example, the two distinct subevents in ‘blanch’ might not be causally related, but they are in a much tighter relation than she wants us to believe, since they have to involve the same theme, be temporally adjacent, and cannot appear in the reverse order. ‘Scribble’, in turn, is actually not a counterexample under Rappaport Hovav & Levin’s definition of result as scalar change, since the change involved is not necessarily scalar. Finally, the Conventional Frame constraint proposed instead might shift the problem to another unclear question, namely what are possible semantic frames?
In ‘Contact and Other Results’, Nomi Erteschik-Shir and Tova Rapoport extend their previous account of argument structure alternations with two types of contact verbs (e.g. ‘hit’ vs. ‘smear’), to include a third type (e.g. ‘splash’). Starting from the assumption that a single verb can project various structures and that, given a universal inventory of atoms, M(anner), S(tate), and L(ocation), verbs can at most specify two atoms, M and two kinds of result (S or L), they argue for different make-ups of lexical atoms in the otherwise uniform lexical representations of contact verbs. The projection possibilities themselves are taken to be constrained only by the principle of Full Interpretation, according to which the interpretation of each atom has to take place within its local V projection, and projected structure requires the availability of an uninterpreted atom. Hit-verbs are analysed as involving M (forceful means) and L (point of contact). Such verbs can project a change-of-location structure, with M an adverbial modifier and L projecting as a (null) preposition (‘The car hit the wall’), as well as an agentive causative structure with an overt goal predicate, with M an adverbial modifier and L modifying the theme DP (‘Jane hit the ball to the other side of the field’). Smear-verbs are argued to involve M (smear manner) and L (surface contact). A complex cause structure is derived when L modifies an overt preposition ‘on’ (‘We smeared mud on the wall’), whereas a simple change structure (*‘Mud smeared on the wall’) is argued not to be acceptable because the M atom would remain uninterpreted, as it requires an agent (a cause). In a second causative structure (‘We smeared the wall with mud’) L modifies the theme subject of the central coincidence predicate ‘with’. Finally, splash-verbs are taken to specify the dispersal of a plurality of particles, and to involve only one atom, L-pl(ural) (splash-shaped surface contact). The fact that single-component splash-verbs can project a change structure (‘Mud splashed on the wall’) follows naturally; however the complex cause+change structure (‘We splashed the wall with mud’) is more surprising. The authors claim that this structure is possible because of L’s plurality.
In ‘The Lexical Encoding of Idioms’, Martin Everaert defines an idiom as a “conventionalized linguistic expression which can be decomposed into potentially meaningful components and exhibit co-occurrence restrictions that cannot be explained in terms of rule-governed morphosyntactic or semantic restrictions” (p. 81). Being concerned with the nature of a lexicon in the I-language sense, he investigates the question whether idioms, whose conventional nature makes them an object of E-language, could be part of this lexicon. He extends the generative theory of lexical representation by proposing that lexical items can be specified not just for C(ategorial)- and S(emantic)-selection, but also for L(exical)-selection, which imposes co-occurrence restrictions between lexical heads. For example, to account for the meaning of ‘kick the bucket’, Everaert proposes among the literal subsense(s) of ‘kick’ another subsense, say meaning (4) (abbreviated in the following as kick4), which in combination with a particular subsense of ‘bucket’, say meaning (8) (bucket8) and with a definite determiner, means ‘die’, as L-selectionally specified for kick4. Similarly, the lexical semantics for ‘bucket’ specifies several subsenses, including bucket8, which is L-selectionally specified as meaningless in the context of kick4. However, all subsenses of ‘kick’ or ‘bucket’ share the same C- and S-selectional specifications, which captures the generally acknowledged fact that in idioms, the morphosyntactic properties of their constituting parts are the same as under the non-idiomatic readings (e.g. irregular past tense or plural morphology, the lexical aspect of the verb), or that idioms allow for modification or replacement of parts of them.
PART II: ARGUMENT STRUCTURE AND THE COMPOSITIONAL CONSTRUCTION OF PREDICATES
In ‘The Emergence of Argument Structure in Two New Sign Languages’, Irit Meir investigates the argument structure mechanisms of two sign languages that started developing in the 1930s, namely Israeli Sign Language (ISL) and Al-Sayyid Bedouin Sign Language (ABSL). She compares three different age groups for each language in the way (context-free) single actions depicted in video clips are signed to another person, who in turn had to identify the action in a picture-verification task. Meir observes that the older generations pursue communicative avoidance strategies, for example by employing one-argument clauses, which obliterate the need to mark argument structure (e.g. WOMAN SIT; GIRL FEED for an event involving a girl feeding a woman), or the (less common) “body as subject” strategy, in which the subject (even if it is not first person) is identified with the signer. Both strategies are assumed to be costly: the inflation of verb forms makes the discourse loaded and redundant, whereas the second strategy confounds grammatical person with syntactic role. Grammatical strategies to encode argument structure are argued to gradually develop in younger generations, with ISL showing preference for verb agreement, which is fully developed in the third group, and ABSL for word order (SOV order is predominant by the third generation), without verb agreement. Verb agreement in sign languages takes place when the arguments’ R-loci, i.e. the points of space which their referents are associated to, are incorporated into the verb form by pointing or eye gaze. Meir also identifies intermediate steps to mark verb agreement in first- and second-generation ISL signers. For example, some signers localise referents in space, without integrating these R-loci into the verb form, other signers use ‘auxiliary’ signs that move between the R-loci without this movement being part of the verb form, or they mark agreement only with the object, but not with the subject.
In ‘Animacy in Blackfoot: Implications for Event Structure and Clause Structure’, Elizabeth Ritter and Sara Thomas Rosen investigate a particular type of verbal morphology, so-called finals, in the Algonquian language Blackfoot. Different finals result in four verb stem classes: intransitive (in)animate (II & IA; the subject is (in)animate), and transitive (in)animate (TI & TA; the object is (in)animate). They propose that, rather than marking an aspectual or a lexical argument structure distinction, these finals are overt exponents of v, since they have both syntactic and semantic properties generally attributed to v. In particular, all and only transitive finals are shown to license a DP object, including unselected objects, such as benefactives, or so-called cross-clausal agreement. Semantically, TI, TA and IA finals are argued to impose an animacy restriction on the external argument and thus to theta-assign this argument, whereas II finals do not, and the respective verbs are unaccusative. They conclude that these properties make Blackfoot finals a mixed category with functional (object licensing, Case-checking) as well as lexical properties (theta-marking). Ritter & Rosen refute the alternative hypothesis that finals mark (a)telicity by showing that verbs that can appear with both transitive and intransitive finals, e.g. ‘eat’, behave alike with respect to standard telicity tests, no matter whether they appear with a transitive or an intransitive final.
In ‘Lexicon versus Syntax: Evidence from Morphological Causatives’, Julia Horvath and Tal Siloni compare morphologically derived causatives in Japanese and Hungarian. Based on evidence from Binding, negation, VP-ellipsis, and the interpretation of agent-oriented adverbs, data discussed in previous literature reveal that Japanese morphological causatives formed by the productive causative morpheme ‘-(s)ase’ are biclausal. Horvath and Siloni show, based on the same diagnostics, that Hungarian morphological causatives, which are derived by means of the productive suffix –‘(t)at/-(t)et’, are monoclausal. They argue that previous, uniformly syntactic analyses do not capture the facts because they either make the wrong prediction that productive morphological causatives are necessarily biclausal, or because they have to implement additional stipulations to account for the fact that the input can also be transitive and unergative verbs. Instead, Horvath and Siloni propose that languages like Japanese derive morphological causatives in the syntax, whereas languages like Hungarian derive them in the lexicon. They define a lexical causativisation rule that adds an agent argument to the theta grid of the input verb, and, if necessary, revaluates the lexically specified causative component of the Agent of the input predicate, [+c(ausative)], to [-c(ausative)]. From this account it follows that the former are biclausal, whereas the latter are monoclausal. To further support this account, they discuss Japanese data involving the causativisation of syntactically derived structures involving coordination or raising verbs, which is not possible in Hungarian.
Causatives are also addressed in ‘On the Morphosyntax of (Anti)Causative Verbs’ by Artemis Alexiadou, who posits four classes of verbal meanings represented by a root or core component: agentive (e.g. ‘murder’), internally caused (e.g. ‘blossom’), externally caused (e.g. ‘destroy’), and cause unspecified (e.g. ‘break’). She claims that in principle all but agentive roots allow the (anti)causative alternation, but languages fall into two groups whether they actually do so. In the first group, represented by English, only cause unspecified roots are shown to alternate, whereas in languages of the second type, e.g. Hindi and Greek, all but agentive roots do. Alexiadou proposes two anticausative/intransitive structures to be available, one without Voice and one with a non-agentive Voice, which lacks an external argument. Morphologically marked anticausatives are argued to always appear in the latter structure. She brings forward data from various languages (Greek, Hindi, Korean, Turkish, Japanese, Armenian) to support these two structures, which share the same pattern: with internally caused and cause unspecified verbs, the intransitive is basic and the anticausative is morphologically less complex, leading to the structure without Voice; with externally caused verbs, the transitive is basic and anticausativisation involves anticausative morphology (e.g. non-active morphology in Greek), leading to the structure with Voice. In the latter case, the morphology is often identical to passive morphology in the relevant languages, but the ungrammaticality of agentive ‘by’-phrases or modifiers shows that they are not passive. Alexiadou analyses anticausative morphology as a syncretism marking a valency reduction of some sort. It is furthermore argued that English only has the first structure, the one lacking Voice, but not the second structure, the one that is similar to a passive. This is proposed to follow from the fact that English lacks valency reduction morphology, and from an analysis of the English passive as structurally more complex (the passive morpheme sits in an additional aspectual projection, hence no valency reduction) than the passive in languages that possess valency reduction morphology (which sits directly in Voice).
In ‘Saturated Adjectives, Reified Properties’, Idan Landau addresses argument structure alternations with evaluative adjectives, for which a basic adjectival construction (BasA, e.g. ‘John was very generous (to Mary)’) alternates with a derived adjectival construction (DerA, e.g. ‘That tribute was very generous (of John) (*to Mary)’). The external argument of BasA is shown to appear as an optional PP in DerA, whereas the internal goal argument of BasA cannot appear in DerA (or only as an adjunct ‘towards’-phrase). DerAs are argued to necessarily involve a stage level interpretation, whereas BasAs also allow an individual level interpretation. Landau proposes two possible structures for BasAs, namely one that involves an event argument (for the stage level interpretation, deriving an event predicate) and one that does not (individual level interpretation, deriving a proposition). He argues that DerAs are derived from BasAs by two operations, the lexical operation Saturation (SAT), which unselectively saturates all the individual arguments of BasA, and the syntactic operation Reification (R), which introduces a new external argument that is construed as a realisation of the predicate. This analysis is argued to capture the fact that DerAs can only be derived from stage level BasAs, since R needs an event and cannot relate a proposition to an individual. He furthermore states that as a result of SAT, the arguments of BasA can only be expressed by (or doubled as) adjuncts (‘of’- or ‘towards’-phrases); a proper goal argument (projected by a more deeply embedded goal phrase) is assumed not to be possible because its projection would depend on an external argument, introduced by the adjectivalising head a; this would lead to ungrammaticality when leaving the external argument unsaturated, or such an external argument would clash with the adjectivalising head a-R, necessary to introduce R. Both operations are argued to be independently available, with SAT being involved in the derivation of verbal passives, passive event nominals, as well as alternations found with subject and object experiencer adjectives, and R introducing the external argument in nominals.
PART III: SYNTACTIC AND SEMANTIC COMPOSITION OF EVENT STRUCTURE
In ‘Incremental Homogeneity and the Semantics of Aspectual ‘for’-Phrases’, Fred Landman and Susan Rothstein make precise the notion of homogeneity that is required to pick out the predicates that can be modified by ‘for’-phrases. Homogeneity is commonly defined as a temporal notion: states are homogeneous down to instants, activities are homogeneous only down to (sufficiently large) subintervals, whereas accomplishments and achievements are not homogeneous. Landman & Rothstein argue that homogeneity down to subintervals cannot account for the fact that activities allow for pauses and gaps. Instead, they propose to define homogeneity with respect to events, as incremental homogeneity, which is the “incremental preservation of cross-temporal identity of an event, and of its event type, between the running time of the onset of that event and the running time of that event itself” (p. 236). It follows, they argue, that ‘for’-adverbials are allowed with states, since these are lexically constrained as being homogeneous down to instants, which is a stronger notion than incremental homogeneity. They note that activities might allow for gaps segmentally but not incrementally and analyse them as lexically constrained to be incrementally homogeneous. Accomplishments and achievements are argued not to meet the definition of incremental homogeneity, because their onsets are not events of the same type. Landman & Rothstein show that accomplishments and achievements in combination with bare plural and mass objects, but not in combination with determiner-noun phrase combinations, are acceptable with ‘for’-adverbials, and propose that this is so because bare plural and mass nouns are kind-denoting.
In ‘Event Measurement and Containment’, Anita Mittwoch addresses the semantics of ‘for’- and ‘in’-adverbials, which serve as common diagnostics for the distinction between atelic and telic predicates. She follows the common analysis of ‘for’-adverbials as extensive measure functions, with the presupposition that they apply only to homogeneous predicates. Such adverbials carry a scalar implicature, in the sense that ‘for an hour’ implies ‘not for more than an hour’, and they cannot apply to predicates that are already measured (in another domain) or quantised. Given that ‘in’-adverbials apply to predicates that are already quantised, Mittwoch argues that they cannot be measure phrases but only indirectly measure the event, by measuring the interval that contains the event. Furthermore, they involve a reversal of the scalar implicatures found with ‘for’-adverbials, in the sense that ‘in an hour’ implies ‘not in less than an hour’. She addresses particular restrictions on ‘in’-adverbials that do not hold for ‘for’-adverbials and argues that they follow from the assumption that ‘in’-adverbials are container measures and thus operate on a descending scale. For example, whereas ‘for’-adverbials are compatible with both upper and lower bounds (‘for at least / at most an hour’), ‘in’-adverbials require an upper bound (‘#in at least an hour’), and additionally involve relative shortness (‘in as little as an hour’ is more felicitous than ‘in as much as an hour’). The paper concludes with a comparison between ‘in’-adverbials and the ‘take’-construction (e.g. ‘It took her an hour to complete the essay’), as an alternative way of indirectly measuring the length of telic eventualities. Mittwoch shows that this construction is not subject to the aforementioned constraints that apply to ‘in’-adverbials and proposes that the ‘take’-construction is the unmarked means to (indirectly) measure telic events.
In ‘Draw’, Christopher Piñón argues that the verb ‘draw’ behaves differently from regular verbs of creation, such as ‘build’, and he provides a semantic analysis of three readings he observes with this verb, which are shown to be morphologically distinguished in Hungarian. Under the first reading, for which Hungarian uses the non-prefixed verb ‘rajzol’, a kind of drawing is produced but no particular object is involved. This reading (under neutral intonation) is only compatible with non-specific indefinites, and Piñón analyses this as a proper verb of creation. Under the second reading, expressed by the Hungarian prefixed verb ‘le-rajzol’ - ‘on-draw’, the drawing of some object is produced, which involves the copying of an image. According to Piñón, this is the meaning of a verb of depiction, and it requires a definite or a specific indefinite object NP. Finally, the third ‘draw’, for which Hungarian employs the perfective prefix ‘meg’ (‘megrajzol’), expresses that a drawing is made of an object based on a certain description of that object, which does not involve copying. Thus, the article provides a more fine-grained lexical semantics of verbs that have previously been classified under one label such as ‘verbs of creation’. An open question is whether it is necessary to postulate three distinct meanings for ‘draw’, rather than maintaining a single lexical entry and deriving the different readings from the composition of the verb with particular NPs ((in)definite, (non)specific) as well as particular aspectual information implicit in English but explicitly provided by the morphology in Hungarian, with particles like 'le' or 'meg' displaying systematic effects on various verb meanings.
In ‘Morphological Aspect and the Function and Distribution of Cognate Objects Across Languages’, Geoffrey Horrocks and Melita Stavrou investigate cognate object constructions (COCs) in Greek, Hebrew and English. They argue that Greek has transitivising COCs (TCOCs), which occur with verbs of all classes (unergative, unaccusative, (di)transitive), do not change the aspectual nature of the underlying verb, and in which the COs are fully referential arguments. They argue that Hebrew COCs similarly appear with all kinds of verbs and do not change the aspectual nature, but involve not arguments but rather activity and event nouns, commonly adjectivally modified. The status of the object as (non-)referential in a given language is diagnosed by its (in)ability to be questioned, to undergo passivisation, and to appear with all kinds of determiners, among others. In contrast, the core of English COCs are argued to only occur with unergatives and furthermore to have the potential to change the aspectual nature of the underlying verb (from non-terminative to terminative and thus to a telic VP). English COs are non-referential and non-argumental, and this construction is viewed as a telic alternative to the otherwise atelic VPs based on non-terminative unergative verbs. They propose that the aspectual change is mediated by a lexical rule, which is similar in nature to the lexical rule responsible for combining activity verbs or manner of motion verbs with secondary resultative predicates or goal phrases into a complex accomplishment predicate (e.g. ‘dance onto the stage’). In a cross-linguistic perspective it has been shown (e.g. Beck & Snyder 2001) that languages that have strong resultatives can also combine manner of motion verbs with goal PPs to derive an accomplishment. Horrocks & Stavrou propose that the ‘aspectual’ COCs of the English type pattern with these constructions, given that English, unlike Greek and Hebrew, has all three constructions to derive complex accomplishment predicates with activity verbs. They tie this to the fact that languages like Greek and Hebrew express grammatical aspect on verb stems, with every verb form being specified for grammatical aspect, so that these pairs of verbs have to be listed separately in the lexicon. It is argued that as a result the lexical aspect of an event description, which grammatical aspect operates on to derive the overall aspectual interpretation, has to be fixed at the lexical level, once and for all, and that it cannot be changed in the syntax by adding resultatives or similar phrases. A potential problem for their overall cross-linguistic generalisation is (most) Romance languages, which do not have strong resultatives or the complex motion predicates mentioned above (cf. Beck & Snyder 2001), but which have grammatical aspect only in past tense forms.
In ‘Locales’, Hagit Borer addresses post-verbal subjects in Hebrew, Italian and Catalan. Besides the well-known restrictions on the availability of post-verbal subjects (they cannot be external arguments or strong NPs), she makes the new observation that common post-verbal subject cases are rather limited and only occur with (a subset of the) achievement predicates. Furthermore, the events involved in these cases are interpreted as telic, despite their lacking a ‘quantity expression’ (they can only occur with weak nouns), which, by assumption, is otherwise necessary for a telic interpretation to arise. She proposes that post-verbal subject order is licensed in these cases by a covert ‘locale’ provided by such predicates, which is a locative associated with the location of the event. The locale existentially binds the event argument, which in turn binds its argument forcing it to be weak, yielding the post-verbal order. Borer proposes that a similar mechanism leads to the telic interpretation of achievements in the absence of a quantity expression in that the covert locale also licenses ASP-Q, which, by assumption, is necessary for a telic interpretation to arise. Her account correctly predicts that overt locales, such as ‘here’, ‘there’ (or Catalan ‘hi’), have a similar effect as the covert ones, in that they can license a postverbal position of the subject with other event and clause types, which usually do not occur with this word order. Also in these cases the subject must be weak; the overt locale has to be weak as well and is necessarily adjacent to the verb, unstressed, and may not be coordinated.
In ‘Modal and Temporal Aspects of Habituality’, Nora Boneh and Edit Doron discuss two different strategies to express habituality in English, Hebrew, and Polish, a simple form (e.g. ‘Yael worked in the garden’) and a periphrastic form (e.g. ‘Yael used to work in the garden’). Both are argued to express a state since habituals, by assumption, are stative, but to differ in the sense that the periphrastic form expresses a retrospective view on the state, to the effect that it is felt as disjoint from the speech time S, whereas the simple form allows the state holding at S. They propose that the disjointness effect with the periphrastic form arises from a scalar implicature, due to stronger forms to express that a habit continues until S, namely the English present perfect and the Hebrew simple present. They analyse the periphrastic form as a complex aspect, with the reference time R preceding the perspective time P. Even though habituality strongly correlates with imperfectivity, especially in the Romance languages, they show that habituality is independent of imperfectivity, which, by definition, involves the inclusion of R in the event time E, since perfective verb forms also allow for habitual readings (where E, the time of the habitual state, is included in R). Furthermore they argue that habituality has a modal character and that the modal operator in habituals is distinct from that in imperfectives, since it includes dispositionality. They define a modal operator Hab, which involves an initiation event and iteration in possible worlds and which adjoins to the VP, with this modified VP being the input to aspectual operators. With the simple form, Hab is taken to be the input to a perfective or imperfective operator, whereas with the periphrastic form they postulate a higher aspectual operator -Hab, realised as, e.g., ‘used to’, which predicates actualisation and ‘distancing’ from P. A main contribution of this paper lies in the dissociation of habituality from imperfectivity. Its shortcoming, however, is that it starts out with the promise to analyse habituality in English, Hebrew, and Polish, but in the end discusses data mainly from English and Hebrew, and it is not clear whether Polish works the same way. In addition, some claims about Polish are rather vague and questionable (e.g. that Polish does not have grammatical aspect but only lexical aspect, or that states cannot be perfective).
The papers in this volume address important and current issues in the semantics and morphosyntax of event predicates, employing different frameworks and covering a broad empirical basis. It will be of great service to scholars interested in the domain of events and could easily constitute the basis for a graduate course on this topic. The coherence of the volume is strengthened by the fact that similar topics are addressed from different points of view (e.g. lexicalisation constraints in a lexical, syntactic, or cognitive-cultural perspective; causativisation as a lexical or syntactic operation; the semantic basis for diagnostics to distinguish between different classes of event predicates; the verb-framed vs. satellite-framed typology and its extension to other empirical domains), which is acknowledged by cross-references in a number of papers. Finally, it achieves its goal to honour and recognise the work of Anita Mittwoch, which, as the editors note, ‘has stood the test of time’, with Mittwoch’s deep understanding of linguistic phenomena in the domain of aspectuality and temporality leading to the profound impact her work has had on the research in this area.
The volume is carefully edited, but what could have been improved is the uniformity of the single list of references, for example, by instructing authors to check whether a manuscript has been published in the meantime, by specifying if a particular paper or book has been reprinted, or by adding missing page numbers and other information. Instead, several references are repeated as manuscripts or just because they were not cited exactly the same, or because they were more or less reprinted in identical form.
Beck, Sigrid, and William Snyder (2001). Complex predicates and goal PPs: Evidence for a semantic parameter. In Proceedings of the 25th Annual Boston University Conference on Language Development, Vol. 1, ed. H.-J. Do, L. Domingez, and A. Johansen. Somerville, MA: Cascadilla Press, 114-122.
Croft, William (1991). Syntactic Categories and Grammatical Relations. Chicago: Chicago University Press.
Levin, Beth, and Malka Rappaport Hovav (to appear). Lexicalized meaning and manner/result complementarity. In The Subatomic Semantics of Event Predicates, ed. B. Arsenijević, B. Gehrke, and R. Marín. Dordrecht: Springer.
ABOUT THE REVIEWER
ABOUT THE REVIEWER:
Berit Gehrke is a postdoctoral researcher in formal semantics at the
Universitat Pompeu Fabra in Barcelona, currently investigating issues
related to the ontological foundation of events in verbal and non-verbal
predicates, as well as the semantics of participles and constructions
employing these. Her dissertation ‘Ps in Motion’ (2008, Utrecht)
investigates the semantic and syntactic role that P elements (adpositions,
prefixes, particles) with a spatial semantics play in the structure of