Schreiner, Sylvia L.R. (In press). The syntax-/ interface. For A. Carnie, Y. Sato, and D. Siddiqi, eds. Routledge Handbook of Syntax. London: Routledge.

1. INTRODUCTION A number of phenomena important to our understanding of the structures and meanings of natural language lie at the juncture between the two. This overview considers the major phenomena at the interface between syntax and semantics/pragmatics, as well as the major theoretical questions that have arisen around these phenomena and around the interface itself. There is only an interface to talk about between syntax and semantics inasmuch as the two are considered to be separate components (as has generally been the case in the generative tradition). We can talk about this “interface” in at least two ways: on the one hand, we can talk about the location in a model of language competence and/or performance where the syntactic and semantic modules meet and interact. On the other hand, we can talk about phenomena that seem to be driven by both syntactic and semantic mechanisms or principles. Both perspectives will be considered here. Studies of phenomena at the interface seek to answer questions such as the following: Does each part of a syntactic structure play an equal role in determining the meaning? Which parts of the meaning have overt reflexes in the structure? Can the overall meaning be easily deduced from the summed meaning of the parts? And, which kinds of meaning are instantiated with a piece of morphosyntax, and which merely have a syntactic effect (i.e. on ordering relations, co-occurrence restrictions, limitations on movement, etc.)? Several approaches to overarching versions of these questions are discussed there. This article is structured as follows: Section 2 presents some of the major issues at the interface between syntax and semantics, with special attention paid to compositionality, theta theory, and functional heads; the final subsection is devoted to phenomena at the interface with pragmatics. Section 3 describes major models of the interface in syntactic and semantic theory, with the last subsection focusing on approaches to the interface(s) with pragmatics. Section 4 concludes and suggests avenues for future work.

2. ISSUES AT THE INTERFACE OF SYNTAX AND SEMANTICS Here I present some of the major topics that seem to occur naturally at the syntax-semantics interface, along with examples of work in each area.

2.1 Interpretation and compositionality The issue that underlies most if not all work at the syntax-semantics interface is how to arrive at the meaning of a structure. Many approaches have come to the same conclusion: that the structure is built first, and the meaning is then obtained from the structure in one way or another. This is the case in the Principles & Parameters framework in general (from Deep Structures, Surface Structures, or Logical Form), and in the Minimalist Program (from LF), but not, for instance, in Muskens’ (2001) non-compositional λ-grammar account (in which semantics is strictly parallel to syntax, rather than secondary to it in any way). In Lexical Functional Grammar, as well, syntactic and semantic levels of representation exist in parallel with mappings between them; meaning is read from the semantic representation. In mainstream generative syntax, the basic picture of the grammar has been one in which the syntax is responsible for building structures, and the semantics is responsible for assigning interpretations to those structures. In early views (the “Standard Theory”), syntactic Deep Structures were the input to the semantics. (In Generative Semantics, on the other hand, the interpretations were actually generated there.) In the “Extended Standard Theory”, semantic interpretation occurred at two points—once at Deep Structure, and once at Surface Structure (this was in response to issues with the interpretation of , as discussed below). This was followed by the move to an LF-input view. In much current Minimalist thinking, chunks of structure are interpreted piece-by-piece, e.g., at phase edges. Compositionality is the concept of assembling the meaning of a larger constituent from the meaning of its component parts via some principles of combination. A number of pragmatic or -level ( dependent) phenomena present problems even for non-strict interpretations of compositionality; it is difficult, for instance, to see how conversational implicatures or the meaning lent by sarcasm could be computed by the same mechanism that determines the interpretation of verb phrases. At the syntax-semantics interface, there are several levels at which compositionality might be expected to hold: with sentence-level modification like ; at clause level, from the composition of the external argument with the verb phrase; within the VP, to account for the composition of the verb with its internal argument; and within (the equivalent of) determiner phrases, adjective phrases, adverb phrases, etc. Depending on one’s theory of morphology, the syntax may also be responsible for producing the input to the lexical(- level) semantics—see e.g. Distributed Morphology (Halle & Marantz 1993, Harley 1999, etc.) for a view of morphology where word-building is done in the syntax. In formal semantics, Frege’s concept of semantic composition as the “saturation” of functions (i.e., as functional application) has remained in the fore, with Heim & Kratzer’s (1998) work being an important contribution. The concept of composition as functional application has been used in both extensional and intensional semantics. It is based on the idea that the meanings of words (and larger constituents) need to be “completed” with something else. (For example, the meaning of a transitive verb is incomplete without its direct object.) Sentence meanings are arrived at by a series of applications of functions to their arguments (which the functions need in order to be “saturated”). At the sentence level, the output is no longer a function but whatever the theory holds to be the meaning of a sentence—in extensional truth-conditional semantics, a truth value. Early formalisms based on (Montague 1974) worked from the perspective that each phrase level’s syntactic rule had a separate mechanism for semantic interpretation. Klein and Sag (1985) proposed that each constituent needing an interpretation was of a certain basic type; in their theory it was these types that had rules for interpretation rather than the syntactic rules themselves. Klein & Sag used the types of individuals, truth values, and situations to form their higher types; work in event semantics (following Davidson 1967) has also proposed a type for events. Other rules of composition have been introduced, such as modification (e.g., Heim & Kratzer 1998). This allows the meaning of intersective adjective phrases to be computed: it essentially lets us say that the meaning of brown house is the same as the meaning of brown plus the meaning of house. Non-intersective adjectives present some trouble for predicate modification. In addition to the mechanics of compositionality, theories of semantic interpretation differ in terms of how homomorphic they assert the syntax and the semantics to be—that is, how much of the interpretation is allowed outside the confines of the compositional meaning. Sentence meaning in strictly compositional theories (e.g., Montague’s 1970 approach) is derived only from the meaning of the syntactic parts and the way they are combined; in non-strictly compositional theories there are also rules that operate on the semantics itself, without a syntactic rule involved (as in some of Partee’s work, e.g. Partee & Rooth 1983).

2.2 Theta Theory The interaction between theta roles (e.g., external argument) and their associated thematic roles (e.g., agent) sits naturally at the syntax-semantics interface. The aim is to discover the connections between syntactic arguments and the semantic part(s) they play in sentences. The Theta Criterion has been the of much work in government and theory and its successors. The original formulation (Chomsky 1981) said that each theta role must be realized by one argument and each argument must be assigned one theta role. This was recast in Chomsky (1986) in terms of chains. An argument (e.g., a subject noun phrase in a passive) undergoes movement, and the coindexed positions it occupies before and after this movement make up a chain; the chain itself gets the theta role. The formal representation of theta roles also underwent changes—the early “theta grid” only represented the theta roles themselves with an indication of their status as internal or external, while later conceptions (e.g. as laid out in Haegeman’s 1991 textbook) also include argument structure information. Thematic roles and relations themselves have also received much attention. Work on “thematic hierarchies” (e.g. Larson 1988; Grimshaw 1990) attempts to explain the assignment of thematic role participants to their positions in the syntax. Dowty’s work (beginning with 1991) on proto-roles (Proto-Agent and Proto-Patient) was a reaction to the difficulty researchers were having in finding cross-linguistically reliable role categories. A notably different approach to defining roles is seen in Jackendoff’s work on thematic relations. Jackendoff (e.g., 1983) approaches the task from the semantics (or, rather, “conceptual structure”) side only—thematic relations are defined by conceptual structure primitives in different configurations. A number of researchers have also concerned themselves with the relation between thematic roles, argument structure/selection, and event structure. Krifka (1989) and Verkuyl (e.g. 1989) both propose aspectually specifying features to better account for particular thematic relationships (see Ramchand 1993 for an implementation). More recently, Ramchand (in her 2008 monograph) lays out primitives for decomposing verb meaning. She argues that in order to discover and understand thematic roles, we must first have the correct features that make up events, “since participants in the event will only be definable via the role they play in the event or subevent” (p. 30).

2.3 Functional heads Proposals about the nature, number, and location of functional heads have also been integral to the development of our understanding of the syntax-semantics interface. Notable work on parameterization and functional heads came out of the University of Geneva in the late 1980s (see the papers in Belletti & Rizzi 1996). Then, Ramchand’s (1993) dissertation on aspect and argument structure draws on data from Scottish Gaelic, investigating the relationship between the verb and its arguments in aspectual terms. Hers is both a semantic and a syntactic account; she argues for a different concept of θ-role-like labels, based on classes defined by event structure and Aktionsart. She motivates Aspect as a functional head in Scottish Gaelic, giving us the idea that the verb and its aspectual features need not be rolled into one complex bundle. Adger’s important (1994) dissertation on the relationship between functional heads and the interpretation of arguments focuses on the Agr(eement) head, while also calling for separate Tense and Aspect heads. Svenonius (1996) is concerned with the meaning and function of heads. He argues that lexical projections denote properties while functional projections denote entities; functional heads end up serving to connect the utterance to the discourse. Cinque (1999) has been highly influential for those concerned with the number and placement of functional projections. His arguments there rest largely on observations about the cross-linguistic requirements on the placement of adverbs in relation to their related functional material. (See also work adverb placement by Nilsen 1998 and ff.) Cinque observes that across languages we see a consistent ordering of classes of adverbs (e.g., temporal, aspectual, etc.); that we also see a consistent ordering of functional material that encodes concepts like tense, aspect, and mood; and, importantly, that the orderings of adverbs and functional material match each other from left to right. This leads him to suggest that adverb phrases are the specifiers of their corresponding functional projections. Based on these observations, he also suggests a very rich collection of functional projections—one for each adverb class—and proposes that these projections are always present in a language, even if all heads are not pronounced. Work on the inventory, structure, order, and parameterization of functional categories has extended into a research program of its own, Cartography (especially by Cinque, Belletti, and Rizzi; with a series dedicated to it, The Cartography of Syntactic Structures—Cinque 2002; Rizzi 2004b; Belletti 2004b; Cinque 2006; Beninca & Munaro 2010; Cinque & Rizzi 2010; Brugé et al. 2012; and Haegeman 2012).

2.4 Events, argument structure, aspect, and type issues Events and eventhood, , and type assignment lie at the heart of the syntax- semantics interface. Lexical aspect plays an undeniably large role in the semantics of event structure. Notions of durativity (whether an event has duration or is punctual), (whether an event has a natural endpoint or not) and eventiveness vs. stativeness largely define the semantics of events, and often arise as points of contact with the syntax. Discussions of argument structure beyond the theta theory considerations discussed above also naturally fall at the juncture of syntax and semantics. Grimshaw (1990), for example, argues that the thematic and aspectual information of argument structure should itself be structured in the form of prominence relations. She then proposes a new conception of external arguments, extending into the nominal domain. Levin & Rappaport Hovav (1995) approach these topics through the phenomenon of unaccusativity. They work to support Perlmutter’s hypothesis about the interfacing nature of unaccusativity: that it is represented in the syntax, and as such can affect other syntactic mechanisms; and that it is determined by the semantics: aspects of verb meaning determine whether or not unaccusativity arises. Tenny (1994) is another good example of work on aspect and argument structure that explicitly considers the influence of the syntax on , and vice versa. Tenny argues that the internal semantic properties of the event, which are spatiotemporal in nature, are what determine the syntactic characteristics of the verb describing the event and the verb’s arguments. Partee’s (1986/2003) influential work on type-shifting and the interpretation of noun phrases does not make explicit claims about the interface, but nonetheless makes an important contribution to our understanding of it. Motivating semantic types necessarily involves discussing the interaction between the elements involved and the syntax. For instance, Partee is looking to explain how we interpret a particular syntactic piece, the noun phrase. The question of what semantic type that syntactic piece is, and whether/when it shifts types, depends heavily upon evidence from how it interacts syntactically with determiners, quantifiers, etc.

2.5 Quantifiers and Raising The structure of sentences with quantified phrases has been difficult to fit into theories of semantic composition. Quantifiers stymied early generative linguists because it was not clear how they should be treated—they act neither like individuals (proper names) nor like sets of individuals. Theory (Barwise & Cooper 1981) established quantificational phrases as second-order sets; for Heim and Kratzer, quantificational DPs are “functions whose arguments are characteristic functions of sets, and whose values are truth-values” (that is, type <,t>) (1998:141). This works for quantified subjects, but quantified objects result in a type mismatch when composition is attempted. Proposed resolutions to this include type shifting, and raising the quantified object out of the VP (Quantifier Raising, proposed in Robert May’s dissertation, discussed at length in Heim & Kratzer 1998). This does not solve every problem related to quantifiers, but it does help us explain ambiguous scope readings of sentences with two quantifiers, and similar phenomena.

2.6 (Scalar) implicatures and related phenomena Scalar implicatures are often considered to be solely in the domain of pragmatics, as they depend on factors outside the sentence or discourse for their meaning. We might reconsider this, however, given their apparent similarities to polarity items, which seem to have real syntactic restrictions. Chierchia (2004) argues that scalar implicatures and negative polarity items are actually remarkably similar in their distribution (i.e., the kind of polarity context they are affected by). He finds that problems arise (e.g., with some quantifiers) for theories that take scalar implicatures to be computed fully after the grammar, and that these problems can be solved if we take implicatures to undergo processing by a kind of pragmatic “component” at multiple stages of the derivation, and not just “after” the grammar. He claims that there is a recursive computation, running parallel to the standard one, that brings in implicatures. This leaves scalar implicatures in the domain of pragmatics, but allows us to understand how they might have restrictions similar to purely “grammatical” phenomena.

2.7 Verbal phenomena Mood, modality, focus, force, middles, and lexical and grammatical aspect vary greatly from language to language as to whether and how they are instantiated syntactically. This makes for interesting work at the interface, bringing up questions of how compositionality proceeds in similar but distinct instantiations, or how best to structure these categories cross-linguistically. Zanuttini & Portner (2003) consider the syntax and semantics of exclamatives (e.g., “How big you’ve gotten!”). They conclude that two syntactic components, a factive and a Wh-operator, interact to produce the semantic features carried by exclamatives. Portner (2004) also brings up some important interface questions in his discussion of the semantics of imperatives and force. Comorovski (1996), after considering the structures and interpretations of interrogative phrases in various kinds of constituent questions and with quantifying adverbs, argues that the ability to front interrogative phrases out of questions depends on a factor that is semantic (and pragmatic) in nature, namely, a requirement that questions be answerable. Ackema & Schoorlemmer (1994) take middle constructions in English and Dutch as evidence for a “pre-syntactic” level of semantic representation that itself interacts with the syntax. They argue that a middle’s subject is in fact the external argument, and that the representation of the verb’s logical subject is at an earlier semantic level. Steinbach (2002) finds that middles in German share semantic properties with passives but pattern morphosyntactically with actives, and are morphosyntactically identical to transitive reflexives in the language. His account derives the in this syntactic form from the syntax-semantics interface; he argues that syntactic derivations of the middle cannot account for either the or the semantic properties of the sentences that are seen. Instead, a combination of independently motivated assumptions from syntax and semantics account for the different interpretations.

2.8 Issues at the interface with pragmatics Some facets of meaning do seem to fall primarily within the domain of pragmatics. Conversational implicatures, for instance, rely wholly on the discourse in which they are situated to draw their intended meaning. Some phenomena, however, do not fit so neatly into the pragmatic realm. Grice’s conventional implicatures, for example, were meanings that were not fully determined by context, but that were not fully within the category of truth-conditional semantics, either. The meanings of many functional items also seem to require input both from grammatical meaning and from pragmatics. And has several possible complex interpretations, depending on the sentence—in I hit the ball and ran to first base, the most felicitous reading is one that includes a notion of ordering between the two conjuncts. In a sentence like He left the door open and the cat got out, there is a notion of causation. These “additional” meanings are held to be pragmatic contributions. Cann, et al. (2005) discuss the notion of grammaticality versus acceptability by considering the use of resumptive pronouns in English. They give a syntactic account of relative clauses and construal that leads to the pronouns’ generation by the grammar, and then argue that the typical judgment of unacceptability by native speakers (in neutral, non-extended contexts) is due to pragmatic effects—they are only used when the speaker deems them necessary for a particular meaning to be conveyed. They are working within Dynamic Syntax and Relevance Theory, both common approaches at this particular interface. Some phenomena at the syntax-semantics interface have also been analyzed as interacting significantly with pragmatics. Polarity items (e.g. Chierchia 2004), scalar implicatures (ibid.), scalar quantifiers (e.g., Huang & Snedeker 2009), and clitics in Spanish (e.g., Belloro 2007), for example, have all been approached from a pragmatic standpoint. Work on the syntax-pragmatics interface has also turned up in the literature on bilingual and L2 speakers. Hulk & Müller (2000), for instance, propose that syntactic inter-language influence only occurs at this interface. Rothman (2008), concerned with the distribution of null versus overt subject pronouns, concludes that elements at this interface are more difficult to acquire due to their complexity. Studies like these bring us to interesting questions about the modularity of pragmatics and its interaction with the grammar, and establish new ways of thinking about whether pragmatics might be governed by the same rules as syntax and the rest of the grammar, or different ones.

3. MODELS OF THE INTERFACES The status of the syntax-semantics interface in a given syntactic model, and the phenomena that might be seen to reside there, will differ depending on how that model deals with the modules themselves. Here we briefly look at how the interface has surfaced in various ways in models both within and alongside or outside the generative tradition.

3.1 The Chomskyan Tradition In early (including the “Standard Theory,” as set up in Chomsky 1965), the semantics component interpreted Deep Structures, the direct output of the syntax. Later models of the grammar adjusted to this view, trying to make sense of the various scope phenomena affected by transformations that were supposed to happen after Deep Structure. In the revised model, the semantic components that dealt with scope instead interpreted the Surface Structures resulting from transformations, rather than the Deep Structures. More issues with scope and passivity (see e.g. treatment by Lakoff 1971) led to the creation of a further level of representation, Logical Form. This was first thought to be located after a second set of transformations that followed Surface Structure. In this model, the interface with the semantics was found in two places: part of the semantic component read meaning off of Deep Structures, and part off of Logical Forms. The model eventually developed was the “(inverted) Y-model”, with Deep Structure and Surface Structure in their previous positions, but with Logical Form (LF) and Phonetic Form (PF) as branches after Surface Structure. This was what the model looked like when the beginnings of the Minimalist Program (Chomsky 1993) were taking form at the start of the 1990s. In the Minimalist Program (MP) (Chomsky 1995), Deep Structure and Surface Structure are done away with as syntax-internal levels of representation, while LF and PF remain as levels of representation that interface with the conceptual-intentional and articulatory/perceptual systems, respectively. Spell-Out occurs after the syntax-internal operations of Merge and Move (later, Internal Merge); Spell-Out is the point at which syntactic structures get sent to LF or PF for interpretation. In some more recent minimalist work, including Chomsky’s (e.g., 2001) work on phases, there are no “levels of representation” per se, where syntactic structures are interpreted as wholes; rather, each step of a syntactic derivation is interpreted by the appropriate systems. This changes things: in these more derivational models, there is no longer a place that can be pointed to as “the” interface between structure and meaning. Instead, (partial) structures are interpreted at each spell-out or at the edge of each phase.

3.2 Generative Semantics A major departure from the patterns of early generative grammar was Generative Semantics (Ross 1967, McCawley 1968, Lakoff 1971). While theories within the Principles and Parameters framework have taken the view that the semantic component interprets the syntactic structures input to it, Generative Semantics took the semantic interpretation to be solely the product of (generated by) the Deep Structures themselves. While in interpretive theories the syntax and the semantics were left with a particular kind of independence from each other, this was not the case in Generative Semantics. Instead, one set of rules—transformations—applied to deep structure meanings to produce the syntactic forms seen on the surface. An important contribution of Generative Semantics to later theories, including some modern Chomskyan ones, was its approach to lexical decomposition—breaking down the meaning of verbs into component subevents. Decomposition of word (and especially verb) meaning featured in Dowty’s (1972 and ff.) approach to Montague Semantics, Jackendoff’s (1983 and ff.) Conceptual Semantics, and Pustejovsky’s Event Structure Theory (1988 and ff.). This has continued into the work on varieties of little v (DO, CAUSE, BECOME, BE, etc.) including Hale & Keyser (1993), Kratzer (1993), Harley (1995), and Distributed Morphology (starting with Halle & Marantz 1993), and extending to more recent work like Folli & Harley (2005). Wunderlich’s (1997 and ff.) strictly lexicalist Lexical Decomposition Grammar also grew out of the generative semantics tradition, and involves four levels of representation—Conceptual Structure, Semantic Form, Theta Structure, and Morphology/Syntax. In this model, semantic forms determine syntactic structure.

3.3 LFG Two of the more well-known non-Chomskyan phrase structure grammars are Lexical Functional Grammar (LFG) and Head-Driven Phrase Structure Grammar (HPSG). Lexical functional grammar (Bresnan & Kaplan 1982; see e.g. Bresnan 2001, Falk 2001) focuses primarily on syntactic relations, analyzing sentences in terms of both constituency (represented in c-structures) and grammatical functions (in f-structures). While the c-structure of a sentence is in the form of a phrase structure marker similar to what one would see in work within government and binding theory, the f-structure (or “attribute-value matrix”) contains a number of unordered valued features (such as ‘TENSE’, ‘SUBJ[ect]’, etc.). These structures exist in parallel, connected by correspondence functions. Because there is more than one syntactic level (c- and f- structures, but also a(rgument)-structure), there is more than one “interface” between syntax and semantics—presumably, there will be a correspondence function between the semantic structure and each level of syntactic structure (see the brief discussion in Falk 2001). Lexical Mapping Theory (LMT) was developed as a proposal about how to map thematic/θ roles to their syntactic realizations. LMT provides a mapping first between a semantic structure (“θ-structure”, possibly part of s-structure or another semantic/conceptual level) and a-structure, and then another mapping from a-structure to f-structure. Since f-structure already interfaces with c-structure, argument positions are able to be determined in this way. Glue semantics (Dalrymple, et al. 1993) is a theory of semantic composition and interpretation developed for LFG. The theory, which is “deductive” rather than compositional, assigns semantic interpretation to syntactic structures (f-structures) via the principles of (the “glue”). Since f-structures contain both complement and modifier information, and are unordered to help account for flexible- or free word order languages, composition via function application becomes impossible without some extra structure being imposed. Dalrymple, et al. (1993, p. 98) introduce “a language of meanings” (any logic will do) and “a language for assembling meanings” (specifically, the tensor fragment of first-order/linear logic). The f- structures and lexical items provide constraints for how to assemble word and phrase meaning (“lexical premises”). Rules in the logic stage combine those premises via “unordered conjunction, and implication” (ibid., p. 98) to yield sentence-level meaning; meanings are simplified via deduction rather than λ-reduction. Postlexical principles apply to map thematic roles to grammatical functions (ibid, p. 99). Glue semantics has been implemented for other syntactic theories as well, including HPSG (Asudeh & Crouch 2001).

3.4 HPSG HPSG (Pollard & Sag 1994; Sag, Wasow, & Bender 2003) is the successor to Generalized Phrase Structure Grammar (GPSG, Gazdar et al. 1985). GPSG established a system for computing the meaning of a sentence based on semantic tags on the syntactic structures, but otherwise dealt little with the syntax-semantics interface. Central to HPSG is a rich lexicon whose entries have feature structures with phonetic as well as syntactic (SYN) and semantic (SEM) features. This lays a natural groundwork for interface constraints and interactions. The syntactic part of the structure involves two important features, SPR (specifier) and COMPS (complements). On the semantic side, there are three features: MODE (type of phrase), INDEX (links to the situation or individual in question), and RESTR(iction) (an unordered list of conditions that must be met for the meaning to hold). It is through RESTR that thematic-like roles are specified. The entry for run, e.g., is tagged for a runner while give contains a giver, recipient, and gift; each of these roles carries an index that allows it to be linked to something else in the structure. Compositionality proceeds via RESTR; the Semantic Compositionality Principle states that “the mother’s RESTR value is the sum of the RESTR values of the daughters” (Sag, et al. 2003, p. 143), where summation means taking values in order. In addition, the Semantic Inheritance Principle (ibid., p. 144) ensures that MODE and INDEX values are shared between a mother and its head daughter. Argument structure and agreement are effected via the feature ARG(ument)-ST(ructure), which is separate from both SYN and SEM. The ARG-ST of a lexical head represents the values of the specifier plus its complements, in order (the “Argument Realization Principle”). Coindexation among SYN, SEM, and ARG-ST features account for binding facts, among other things.

3.5 Models from semantic theory Semantics in the generative tradition has been less overtly concerned with the exact positioning of syntax in the model of the grammar. Heim & Kratzer (1998) assume a phrase structure grammar approach to the syntax, but note that a number of different theories of syntax are compatible with the version of formal semantics they present. Since their approach to interpretation is “type-driven,”

it’s the semantic types of the daughter nodes that determine the procedure for calculating the meaning of the mother node. The semantic interpretation component, then, can ignore certain features that syntactic phrase structure trees are usually assumed to have. All it has to see are the lexical items and the hierarchical structure in which they are arranged. Syntactic category labels and linear order are irrelevant. (Heim & Kratzer 1998: 44)

Although their phrase structure trees are labeled and linearized, they note that “the only requirement for the syntax is that it provides us with phrase structure trees” (ibid.:45). However, not every kind of syntactic theory is compatible with Heim and Kratzer’s semantics; notably, a theory in which meaning is interpreted from both Deep and Surface Structures is incompatible with the semantic picture they present. Such a syntactic model would make the semantics interpret “something like pairs of phrase structure trees” (ibid.:47), i.e., something uninterpretable in the theory as it stands. While a bit further afield from the mainstream generative tradition, Conceptual Semantics (e.g. Jackendoff 1983 and ff.; Pinker 1989) still separates syntactic and semantic components, and also addresses the nature of the interface between the two. This approach locates the formulation of meaning in the Conceptual Structure, a level of representation located in Cognition (along with Spatial Structure) rather than in Language proper. The Conceptual Structure in turn interfaces with the syntax and phonology levels in Language. Some semantic frameworks fully outside generative linguistics like Cognitive Grammar do not include a concept of an interface between form and meaning components. In Cognitive Grammar (see especially Langacker 1987), the syntactic organization is constructed from the meaning itself; there is no autonomous “syntax” that a semantic component could interact with. Discourse Representation Theory (DRT), introduced in Kamp (1981), is a semantic framework whose first instantiations aimed especially to resolve issues dealing with tense and anaphora across multiple sentences in a discourse. DRT’s major innovations are for the most part semantic in nature—in particular, a level of mental representations called discourse representation structures; and the inclusion of discourse in the interpretation of meaning. However, DRT is also of interest because it seats these semantic innovations within a representationalist theory. Representations are built via a generative syntax and concomitant set of syntactic rules. If not for this DRT would be a dynamic theory of meaning, since meaning interpretation is seen as being “updated” as one moves through the discourse.

3.6 The interface with pragmatics Another important question is how, or whether, to distinguish in the theory between the work done by semantics and that done by pragmatics. The generative approach (at least on the syntactic side) has in general clearly separated syntax from the meaning component, but it has not spent much time addressing the question of where pragmatics comes in. The functions of pragmatics are often taken to exist as part of the semantic (or conceptual/intentional) system, or else pragmatics is pushed off on some other system within language or general cognition. In Minimalism, for instance, the interpretation of both semantic and pragmatic meaning is done at LF. In research on pragmatics, however, the interface between semantics and pragmatics has been the topic of a great deal of debate. The main strategy in this debate has been to take pragmatics as being fully separate from the grammar—the grammar deals with conventional or coded meaning, and pragmatics with nonconventional (inferential, contextual, non-coded) meaning. Alternately, the divide is taken to be between truth-conditional meaning (in the semantics/the grammar) and non-truth-conditional meaning (in the pragmatics). “The grammar” here is, in the cognitive view, the linguistic module, containing syntax and semantics and their corresponding rules. This picture of pragmatics as separate originates in Grice’s (1957 and ff.) pioneering work. Fodor’s (1983) take on modularity removes pragmatics from the language system (which is for him a domain-specific module). Pragmatics is instead “global,” part of the general computational system of the mind. For Fodor this is due to the fact that pragmatics requires contextual, domain-general information in order to be processed, rather than just the local information available to the language module. One important approach to pragmatics in this vein is Relevance Theory (starting with Sperber & Wilson 1981, and much work since then). Relevance Theory takes Grice’s work as a jumping- off point, but focuses on the idea that there are certain expectations of relevance when a discourse occurs. Some more recent work in Relevance Theory (Wilson 2005) proposes that pragmatics is actually a sub-module of a mind-reading module (again, non-grammatical in nature). Wilson uses the term module in a different sense than Fodor does, embracing a broader definition: “From an evolutionary perspective, the question is not so much whether the processes involved are global or local, but whether they are carried out by general-purpose mechanisms or by autonomous, special-purpose mechanisms attuned to regularities existing only in the domain of intentional behaviour” (1132). In this sense, she claims, mind-reading (i.e., utilizing a theory of mind) is modular. She claims that pragmatics is not just a special application of (generalized) mind- reading, but a sub-module of the specialized mind-reading module. It is perhaps not surprising that a consensus has not yet been reached as to how to characterize the semantics-pragmatics interface, nor as to exactly which phenomena fall on one side or the other. Even among those who are clear about locating pragmatics somewhere outside the grammar proper, there has been a good deal of debate as to what is explicable using the truth- conditional or conventional meanings of semantics, and what requires recourse to context and inference and therefore pragmatics.

4. CONCLUSION The interaction between syntax and semantics/pragmatics is necessary to explain a number of phenomena that have been important to the development of theories within and outside the generative enterprise. In this overview, we have looked at several possible approaches to the interface and discussed a number of important issues there, including the ever-present question of compositionality—how to extract meaning from the structures formed by the syntax. Other questions at the interface either seem to be best solved by an appeal to functionality on both sides of the structure/meaning divide, or simply have both syntactic and semantic components. The other issue discussed was the question of where pragmatics should be located in one’s model of language. We saw that a common tactic is to locate the semantics in the language “module”, while deriving pragmatic meaning outside that module. Depending on one’s theory of the cognitive and linguistic architecture, this might mean that the duties of pragmatics are accomplished in a separate (but not language-specific) module, or accomplished via a combination of domain-general processes.

4.1 Directions for future work As we move forward, one pattern of data seems particularly likely to lead to further interface phenomena of interest: categories that tend to vary between languages as to whether and how they are instantiated morphosyntactically (e.g., mood/modality/force/voice, as mentioned above). Comparing data across languages of this type for a particular piece of semantics can lead to insights about how we should be dividing up the semantic distinctions in the first place. For instance, if across languages one particular distinction of mood is instantiated in a manner conspicuously different from that of other moods, we have a strong hint that something is incomplete about our picture of the semantics of mood, or our understanding of the interaction between the syntax and the semantics with respect to that category. In the realm of aspect, for example, Coon (2010) observes that the perfective (as opposed to other aspects) has a tendency to be realized non-periphrastically across languages, and theorizes that this is due to the unavailability of a natural language expression for the time relation involved in perfectivity. In Reed (2012) I use the syntactic instantiations of aspectual distinctions in Scottish Gaelic to support a claim about the semantics of grammatical aspect—namely, that we should view the category as a bifurcated one, with (im)perfective-type aspects on one branch and perfect-type aspects on the other. Focus on this particular kind of interaction between syntax and semantics has already led to some very interesting discoveries, and is likely to lead to more.

REFERENCES

Ackema, Peter, & Maaike Schoorlemmer. 1994. The middle construction and the syntax-semantics interface. Lingua 93:1, 59-90. Adger, David. 1994. Functional Heads and Interpretation. University of Edinburgh, Edinburgh dissertation. Asudeh, Ash, & Richard Crouch. 2001. Glue Semantics for HPSG. In Proceedings of the 8th International HPSG Conference, ed. by F. van Eynde, L. Hellan & D. Beerman, 1-19. Stanford: CSLI Publications. Barwise, Jon & Robin Cooper. 1981. Generalized quantifiers and natural language. Linguistics and Philosophy 4:2, 159-219. Belletti, Adriana, ed. 2004b. Structures and Beyond. The cartography of syntactic structures, Vol. 3. New York: Oxford University Press. Belletti, Adriana, & Luigi Rizzi, eds. 1996. Parameters and Functional Heads: Essays in Comparative Syntax. New York: Oxford University Press. Belloro, Valeria A. 2007. Spanish Clitic Doubling: A Study of the Syntax-pragmatics Interface. State University of New York at Buffalo, NY dissertation. Beninca, Paolo, & Nicola Munaro, eds. 2010. Mapping the Left Periphery. The Cartography of Syntactic Structures, Vol. 5. New York: Oxford University Press. Bresnan, Joan. 2001. Lexical-Functional Syntax. Oxford: Blackwell. Bresnan, Joan, and Ronald Kaplan. 1982. Introduction: Grammars as Mental Representations of Language. In The Mental Representation of Grammatical Relations, ed. by Joan Bresnan, xvii-lii. Cambridge, MA: MIT Press. Brugé, Laura, Anna Cardinaletti, Giuliana Giusti, & Nicola Munaro, eds. 2012. Functional Heads. The Cartography of Syntactic Structures, Vol. 7. New York: Oxford University Press. Cann, Ronnie, Tami Kaplan, & Ruth Kempson. 2005. Data at the grammar-pragmatics interface: the case of resumptive pronouns in English. Lingua 115, 1551-1577. Chierchia, Gennaro. 2004. Scalar implicatures, polarity phenomena, and the syntax/pragmatics interface. In Structures and Beyond: The Cartography of Syntactic Structures, Vol. 3, ed. by Adriana Belletti, 39- 103. Oxford: Oxford University Press. Chomsky, Noam. 1955. The logical structure of linguistic theory. Harvard/MIT MS. Chomsky, Noam. 1981. Lectures on Government and Binding. Studies in Generative Grammar 9, Dordrecht: Foris. Chomsky, Noam. 1986. Barriers. Cambridge, MA: MIT Press. Chomsky, Noam. 1993. A minimalist program for linguistic theory. In The View from Building 20, ed. by K. Hale & S.J. Keyser, 1-52. Cambridge, MA: MIT Press. Chomsky, Noam. 1995. The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, Noam. 2001. Derivation by phase. In Ken Hale: A Life in Language, ed. by M. Kenstowicz, 1- 52. Cambridge, MA: MIT Press. Cinque, Guglielmo. 1999. Adverbs and Functional Heads: A Cross-Linguistic Perspective. Oxford: Oxford University Press. Cinque, Guglielmo, ed. 2002. Restructuring and Functional Heads. The Cartography of Syntactic Structures, Vol. 4. New York: Oxford University Press. Cinque, Guglielmo, ed. 2006. Functional structure in DP and IP. The Cartography of Syntactic Structures, Vol. 1. New York: Oxford University Press. Cinque, Guglielmo, & Luigi Rizzi, eds. 2010. Mapping Spatial PPs. The Cartography of Syntactic Structures, Vol. 6. New York: Oxford University Press. Comorovski, Ileana. 1996. Interrogatives and the Syntax-Semantics Interface. Dordrecht: Kluwer Academic Publishers. Coon, Jessica. 2010. Complementation in Chol (Mayan): A Theory of Split Ergativity. MIT, Cambridge, MA dissertation. Dalrymple, Mary, John Lamping, & Vijay Saraswat. 1993. LFG semantics via constraints. In Proceedings of the Sixth Meeting of the European ACL, 97–105. University of Utrecht. Dalrymple, Mary, ed. 1999. Semantics and Syntax in Lexical Functional Grammar: The Resource Logic Approach. Cambridge, MA: MIT Press. Davidson, Donald. 1967a. The logical form of action sentences. In The Logic of Decision and Action, ed. by Nicholas Rescher, 81-95. Pittsburgh, PA: University of Pittsburgh Press. Dowty, David R. 1972. Studies in the Logic of Verb Aspect and Time in English. University of Texas, Austin, TX dissertation. Dowty, David. 1991. Thematic proto-roles and argument selection. Language 67:3, 547-619. Falk, Yehuda N. 2001. Lexical-Functional Grammar: An Introduction to Parallel Constraint-Based Syntax. Stanford, CA: CSLI Publications. Fodor, Jerry. 1983. The Modularity of Mind. Cambridge, MA: MIT Press. Folli, Raffaella, and Heidi Harley. 2005. Flavors of v. In Aspectual inquiries, ed. by P. Kempchinsky & R. Slabakova, 95-120. Dordrecht: Springer. Gazdar, Gerald, Ewan Klein, Geoffrey K. Pullum, and Ivan A. Sag. 1985. Generalized Phrase Structure Grammar. Cambridge, MA: Harvard University Press. Grice, H. Paul. 1957. Meaning. Philosophical Review 66, 377-388. Grimshaw, Jane. 1990. Argument Structure. Cambridge, MA: MIT Press. Haegeman, Liliane. 1991. Introduction to Government and Binding Theory. Malden, MA: Blackwell. Haegeman, Liliane. 2012. Adverbial Clauses, Main Clause Phenomena, and the Composition of the Left Periphery. The Cartography of Syntactic Structures, Vol. 8. New York: Oxford University Press. Hale, Kenneth, & Samuel Jay Keyser. 1993. On argument structure and the lexical expression of syntactic relations. In The View from Building 20, ed. by Kenneth Hale and Samuel Jay Keyser, 53-109. Cambridge, MA: MIT Press. Halle, Morris, & Alec Marantz. 1993. Distributed Morphology and the pieces of inflection. In The View from Building 20, ed. by Kenneth Hale and S. Jay Keyser, 111-176. Cambridge, MA: MIT Press. Halle, Morris, & Alec Marantz. 1994. Some key features of Distributed Morphology. In MITWPL 21: Papers on phonology and morphology, ed. by Andrew Carnie and Heidi Harley, 275-288. Cambridge, MA: MITWPL. Harley, Heidi. 1995. Subjects, Events, and Licensing. MIT, Cambridge, MA dissertation. Heim, Irene, & Angelika Kratzer. 1998. Semantics in Generative Grammar. Malden, MA: Blackwell. Huang, Yi Ting, & Jesse Snedeker. 2009. Online interpretation of scalar quantifiers: Insight into the semantics–pragmatics interface. Cognitive Psychology 58:3, 376-415. Hulk, Aafke, & Natascha Müller. 2000. Bilingual first language acquisition at the interface between syntax and pragmatics. Bilinguilism: Language and Cognition 3:3, 227-244. Jackendoff, Ray. 1983. Semantics and Cognition. Cambridge, MA: MIT Press. Kamp, Hans. 1981. A theory of truth and semantic representation. In Formal Methods in the Study of Language, ed. by Jeroen AG Groenendijk, TMV Janssen, and Martin BJ Stokhof, 277- 322. Mathematical Center Tract 135, Amsterdam. Klein, Ewan, & Ivan A. Sag. 1985. Type-driven translation. Linguistics and Philosophy 8:2, 163-201. Krifka, Manfred. 1989. Nominal reference, temporal constitution and quantification in event semantics. In Semantics and Contextual Expression, ed. by R. Bartsch, J. v. Benthem & P. v. E. Boas, 75-115. Dordrecht: Foris. Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Volume 1: Theoretical Prerequisites. Stanford: Stanford University Press. Lakoff, George. 1971. On generative semantics. In Semantics: An Interdisciplinary Reader in Philosophy, Linguistics and Psychology, ed. by D.D. Steinberg and L.A. Jakobovitz, 232-296. Cambridge: Cambridge University Press. Larson, Richard K. 1988. On the double object construction. Linguistic Inquiry 19, 335–391. Levin, Beth, & Malka Rappaport-Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics Interface. Cambridge, MA: MIT Press. McCawley, James D. 1968. “Lexical insertion in a transformational grammar without deep structure.” In Papers from the fourth regional meeting of the Chicago Linguistic Society, ed. by B.J. Darden, C.-J.N. Bailey, and A. Davison,71-80. Montague, Richard. 1970. Universal grammar. Theoria 36, 373-398. Montague, Richard. 1974. Formal Philosophy. Cambridge, MA: MIT Press. Muskens, Reinhard. 2001. Lambda grammars and the syntax-semantics interface. In Proceedings of the Thirteenth Amsterdam Colloquium, ed. by R. van Rooy and M. Stokhof, 150-155. Nilsen, Øystein. 1998. The Syntax of Circumstantial Adverbials. University of Tromsø, Hovedoppgave, ms. (published in 2000 by Novus, Oslo) Partee, Barbara H. 1986. Noun phrase interpretation and type-shifting principles. In Studies in Discourse Representation Theory and the Theory of Generalized Quantifiers, ed. by J. Groenendijk, D. de Jongh, & M. Stokhof, 115-144. Dordrecht: Foris Publications. Republished in P. Portner & B. Partee (2003) Formal Semantics: The Essential Readings. Partee, Barbara H. & Mats Rooth. 1983. Generalized conjunction and type ambiguity. In Meaning, Use, and Interpretation of Language, ed. by R. Bäuerle, C. Schwarze, & A. von Stechow, 361-383. Berlin: Walter de Gruyter. Pinker, Steven. 1989. Learnability and Cognition. The Acquisition of Argument Structure. Cambridge, MA: The MIT Press. Pollard, Carl & Ivan A. Sag. 1994. Head-Driven Phrase Structure Grammar. Chicago: University of Chicago Press. Portner, Paul. 2006. The semantics of imperatives within a theory of clause types. In Proceedings of SALT 14, ed. by K. Watanabe and R.B. Young, 235-252. Pustejovsky, James 1988. The geometry of events. In Studies in Generative Approaches to Aspect, ed. by Carol Tenny, 19-39. Cambridge, MA: The MIT Press. Ramchand, Gillian. 1993. Aspect and Argument Structure in Modern Scottish Gaelic. Stanford University, Stanford, CA dissertation. Ramchand, Gillian. 2008. Verb Meaning and the Lexicon: A first-phase syntax. Cambridge: Cambridge University Press. Reed, Sylvia L. 2012. The Semantics of Grammatical Aspect: Evidence from Scottish Gaelic. University of Arizona, Tucson, AZ dissertation. Rizzi, Luigi, ed. 2004b. The Structure of IP and CP. The Cartography of Syntactic Structures, Vol. 2. New York: Oxford University Press. Ross, John R. 1967. Constraints on Variables in Syntax. MIT, Cambridge, MA dissertation. Rothman, Jason. 2008. Pragmatic deficits with syntactic consequences?: L2 pronominal subjects and the syntax–pragmatics interface. Journal of Pragmatics 41:5, 951-973. Sag, Ivan A., Thomas Wasow, & Emily M. Bender. 2003. Syntactic Theory: A Formal Introduction, 2nd Edition. Stanford, CA: CLSI Publications. Sperber, Dan & Deirdre Wilson. 1981. Irony and the use-mention distinction. In Radical Pragmatics, ed. by P. Cole, 295-318. New York: Academic Press. Steinbach, Markus. 2002. Middle Voice: A comparative study in the syntax-semantics interface of German. Amsterdam: John Benjamins. Svenonius, Peter. 1996. Predication and functional heads. In The Proceedings of the Fourteenth West Coast Conference on Formal Linguistics, ed. by JosÈ Camacho, Lina Choueri, & Maki Watanabe, 493-507. Standford, CA: CSLI. Tenny, Carol L. 1994. Aspectual Roles and the Syntax-Semantics Interface. Dordrecht: Kluwer Academic Publishers. Tesnière, Lucien. Eléments de Syntaxe Structurale. Paris: Klincksieck. Verkuyl, Henk J. 1989. Aspectual classes and aspectual composition. Linguistics and Philosophy 12, 39– 94. Wilson, Deirdre. 2005. New directions for research on pragmatics and modularity. Lingua 115, 1129-1146. Wunderlich, Dieter. 1997. Cause and the structure of verbs. Linguistic Inquiry 28:1, 27-68. Zanuttini, Raffaella & Paul Portner. 2003. Exclamative clauses: at the syntax-semantics interface. Lingua 79:1, 39-81.

FURTHER READING Levin, Beth, & Malka Rappaport-Hovav. 1995. Unaccusativity: At the Syntax-Lexical Semantics Interface. Cambridge, MA: MIT Press. This book focuses on establishing linking rules between syntax and lexical semantics, developing unaccusativity as a diagnostic for phenomena at that interface. Pelletier, Francis Jeffry. 1994. The principle of semantic compositionality. Topoi 13, 11-24. A good summary of various arguments for and against different versions of compositionality, from a semantic perspective. See also the reprint (Davis & Gillon’s 2004 Semantics: A Reader, OUP) for later thoughts on the topic from the author. Van Valin Jr., Robert D. 2005. Exploring the Syntax-Semantics Interface. Cambridge: Cambridge University Press. This is an updated introduction to Role and Reference Grammar, a model not discussed here but one intimately concerned with the interface, featuring lexical decomposition and its own set of thematic roles.