verbal roots
FUNCTION GLOSS ENGLISH CHINESE ITALIAN
1.4 Introducing the framework: a ‘first phase syntax’
Many proposals have been made in the literature to connect the morphosyntax and the semantics of the event structure in an intimate way (cf. Jackendoff 1990, Grimshaw 1990, Travis 2000, Lin 2004, Borer 2005, among others). The common idea behind these proposals is that the syntactic projection of arguments is based on event structure. Ramchand (2008) proposes a model where the events are decomposed in syntax, what she calls ‘first phase syntax’ (cf. also Butt & Ramchand 2005). As proposed in some lexicalist frameworks (e.g. Rappaport Hovav & Levin 199843), the event would be composed by a structural meaning and an idiosyncratic meaning, given by the lexical-encyclopedical content of the verb. In this framework the structural meaning is built compositionally from syntax and it includes only those aspects of meaning that are predictable and systematic, excluding many of the aspects that are traditionally included in the description of verbs, as theta roles and some kinds of semantic selection. The grammatically relevant aspects, according to this model, come from the interpretation of the syntactic structures in which verbs are inserted. However, if this approach departs from lexicalist approaches, it is also somehow different from constructionist approaches, since it posits that lexical items are provided of minimal syntactic information which allows them to be inserted in the syntactic structures.
Ramchand (2008) posits that the syntax of the event structure contains three important components: the causing subevent, the process subevent and the subevent corresponding to the resulting state. Each of these sub-events is represented with its own projection, ordered in a hierarchical embedding relation, as shown in (28):
43 According to Rappaport Hovav & Levin (1998), the meaning of a predicate is composed of two components: the structural component (‘event structure template’), which is the grammatically relevant component, and the idiosyncratic component (‘constant’ or ‘root’).
47 (28) initP (causing projection)
tu
DP3
tu
subj of ‘cause’ init procP (process projection)
tu
DP2
tu
subj of ‘process’ proc resP (result projection)
tu
DP1
tu
subj of ‘result’ res XP 4
This kind of structure, besides representing the complexity of an event, which can
be decomposed to a maximum of three subevents, also limits and captures the set of core argument roles, as defined by the predicational relations formed at each level. Each projection forms its own core predicational structure: the specifier position is filled by the ‘subject’ or ‘theme’ of a particular subevent, while the complement position is filled by the phrase that provides the content for that event. The procP (process projection) is the heart of the dynamic predicate (it represents change through time) and it is present in every dynamic verb. The complement position itself is complex and contains another mini-predication, with its own specifier and complement. In this way, the participant relations are built up recursively from successively embedded event descriptions and ‘subject’ predications. Note that the
proc head can either combine with a whole resP (result projection) to create a telic
pair or take a simple XP (DP, AP or PP) in its complement position, which does not determine its own independent subevent, but acts as a further modifier or descriptor of the proc event (cf. Ramchand 2008:46)44.
The initP layer introduces the causation event45 and licenses the external argument (‘subject of cause’ = INITIATOR46); the initP is present when the verb expresses a
44 Note that, when the complement XP position is not overtly specified, Ramchand (2008:62) assumes that this position is filled by a contextual variable.
45The causation event represents the external causative projection, which is responsible for the introduction of the external argument. To many respects, it is similar to the external argument that introduces v in recent literature (e.g. Hale & Keyser 1993).
46 Initiators are defined as the individual entities that possess the property denoted by the initiational subeventuality, which leads to the process coming into being (cf. Ramchand 2008:52). It is important to stress that INITIATION does not entail agentivity: the sun, in a sentence like the sun melted the snow, and the key, in the key opened the lock, are the initiators of the event.
causational or initiational state that leads to the process. Ramchand (2008:55) points out that the difference between pure ‘Causes’ and actual ‘Actors’ is that, while an ‘Actor’ is related to both initiation and process (which may also lead to a result, but not necessarily), i.e. it is an Initiator-Undergoer, a ‘Cause’ is a pure specifier of initiation, i.e. it is just an Initiator. Ramchand also observes that animate/human- referring DPs can be interpreted as volitional causers, willful controllers or experiencers of static or changing mental states; thus, according to Ramchand (2008:55), the psychological version of a pure cause is an ‘intentional initiator’ and the psychological version of an ‘actor’ is a volitional agent with continuous experiential involvement in the process47. Therefore, Ramchand does not make distinctions between different initiational heads in a feature-based sense and she also does not distinguish a causational head from an agent introducing one to account for different kinds of subjects. Different subjects can rather be accounted for in terms of the difference between Initiator and Initiator-Undergoer (an entity continuously involved in the process) and in relation to the encyclopedic content (either the verb’s lexical-encylopedic information or the referential properties of the DP participant, i.e. animate vs. inanimate).
The procP layer specifies the nature of the change or process and licenses the entity undergoing change or process (‘subject’ of process = UNDERGOER). The resP layer provides the telos or result state (‘subject’ of result = RESULTEE).
According to this proposal, there is a general combinatorial semantics that interprets this syntactic structure in a regular and predictable way. This means that the semantics of the event structure and event participants “is read directly off the structure and not directly off the information encoded by the lexical items” (Ramchand 2008:42). The possibilities for event combination correspond to two distinct types of event-event relation, which are part of the semantic ontology. The first relation is that of ‘causation/initiation’:
(29) e = e1 > e2: e consists of two sub-events, e1 and e2; e1 causally entails e2 (cf. Hale & Keyser 1993).
The idea is that the event position corresponding to a transitive verb such as eat can be decomposed into two subevents related by causation, where e1 is the causing or
49 instigating force and e2 is the event of something being consumed. See (30), from Ramchand (2008:42).
(30) eating (e) where e = e1 > e2: [cause-eat (e1) & process-eat (e2)]
The second important semantic relation between events is telic augmentation, i.e. the addition of a particular attained result (e.g. Rappaport Hovav & Levin 1998): accomplishment predicates (in the sense of Vendler 1967) consist of the two subevents of process and telos, respectively, in their representation (the initiational information may be present or not), and the process leads to the resultant state, creating a complex event, as the one shown in (31), from Ramchand (2008:43).
(31) ‘defuse the bomb’ (e) where e = <e1, e2>: [process-defuse (e1) & result-of- defusing (e2)]
The causal embedding is the only primitive in the combinatorial system which can create complex events of the same logical type (the difference between causational and resultative semantics is due to a different hierarchical order in the embedding). Subevents are not ontologically different from macro-events: they are simple processes or states. There are two general primitive predicates corresponding to the basic subevents: state and process (eventuality that contains internal change). Ramchand (2008) assumes that both the initiational eventuality and the result eventuality are states, and that their interpretation comes from the position in the hierarchical structure: in the init position the state is interpreted as causally entailing the process; in the res position the state is interpreted as causally entailed by the process.
Each lexical item specifies the syntactic relevant information (category labels or ‘tags’), which permit its insertion in the eventive structure. For example, in English the lexical entry for a verb such as push will be push [init, proc], while for a verb like
throw the lexical entry will be [init, proc, res]. Given the existence of this functional
sequence, the syntactic structures are freely built up by Merge, but they have to be licensed by the presence of specific lexical items. The lexical items simply Merge48
48 Since lexical items have more than one category label, Ramchand (2008) assumes that elements may Merge and project and then Remerge (cf. Starke 2001) at a later state of derivation. For example, the verb push has two features, [init, proc]. The verb push will Merge with a DP in its specifier position and project its [proc] label. Since it also has an [init] feature, it Remerges with procP, which now projects the init label. This new syntactic objects then Merges with the specifier to project an initP.
and project according to their category features. If Merge does not build a functional sequence that is correctly ordered and interpretable, then the derivation will crash (cf. Butt & Ramchand 2005). At the interface, the encyclopedic content of the lexical items is unified with the semantics given by the combinatoric system. A lexical item can only associate with a node that matches the category features it is listed with.
Ramchand (2008:97) assumes a ‘superset principle’, according to which a lexical item may be inserted to spell out a sequence of heads if its category signature is a superset of the sequence to be spelled out49: “The phonological exponent of a Vocabulary item is inserted into a node if the item matches all or a superset of the grammatical features specified in the node. Insertion does not take place if the vocabulary item does not contain all features present in the node. Where several Vocabulary items meet the conditions for insertion, the item containing less features unspecified in the node must be chosen” (Caha 2007; cf. also Caha 2008:259).
Ramchand (2008) uses the term ‘underassociation’ to indicate the use of a lexical item that bears a superset of the category features it actually spells out. However, she assumes that underassociation is subject to specific conditions and proposes the following tentative constraints: “if a lexical item contains an underassociated category feature, (i) that feature must be independently identified within the phase and linked to the underassociated feature, by Agree; (ii) the two category features so linked must unify their lexical-encyclopedic content.” (pp. 97-98).
In Table 3, you may see the different lexical types in English singled out by Ramchand (2008), with the specification of the participant relations.
Table 3 – Lexical types (Ramchand 2008)
TYPES PARTICIPANTS EXAMPLES