SE-LFG09 (03/11/2012, SOAS)

Meeting Agenda:

11:00-12:00: Prosodic encoding of discourse functions in LFG (Louise Mycock & John Lowe, Oxford)

In this talk, we will outline how Dalrymple & Mycock’s (2011) model of the syntax-prosody interface can be extended to capture an interface phenomenon not originally covered, namely the prosodic encoding of discourse functions.

According to Dalrymple & Mycock (2011) the string, conceived of as comprising a syntactic (s-string) and prosodic (p-string) form, represents the only locus of interface between prosodic and other grammatical information. Each p-string and s-string unit, in common with each p(rosodic)-structure and c(onstituent)-structure node, is associated with an ‘interface structure’ (chi-structure or epsilon-structure, respectively). Interface structures are posited to contain information that is relevant at multiple levels within LFG’s parallel architecture (e.g. clause type, the location of constituent boundaries). A principle of harmony requires interface-structure values associated with p-string units to match those associated with the corresponding s‑string units (and the c-structure nodes which dominate them), and vice versa.

We propose to include Information Structure category status in interface structures. When signalled prosodically, a category such as Focus will be the value of a L and/or R attribute at the level of chi-structure. Harmony requires the equivalent attribute-value pair to appear at epsilon-structure, which ultimately correlates with Focus being the value of the associated semantic structure’s DF attribute. We outline how this approach can be used to analyse the ways in which discourse functions are encoded prosodically and discuss some of its implications.

12:00-12:30: TheCOMP, sentential OBJ and XCOMP complex in Maltese (Maris Camilleri, Essex & Surrey)

This presentation takes Dalrymple & Lødrup’s (2000) paper as its starting point. From there I question and try to understand the nature of complement clauses in Maltese, at least the ones that are introduced by a complementiser li, which appears to function in the same way as English that. From the initial account provided here, given that this question has never been focused upon in Maltese, it seems that Maltese is a mixed language where it is not the case that all complement clauses bear a COMP GF or all are sentential OBJs. However, it looks as though all complement clauses introduced by li and which follow verbs bear an OBJ or OBJ_θ GF and not a COMP, at least following the criteria mentioned in Dalrymple & Lødrup, particularly the UBD criterion. Complement clauses that follow other lexical categories, such as Ns and As bear a COMP GF. In establishing the nature of the finite clauses introduced by li, the issue remains open in the case of control-like clauses, where an XCOMP or COMP could be possible. The reason for referring to these constructions as control-like is because this is a question in itself, i.e. whether some of the data really involve a control relation, or whether the string of verbs share one f-structure head, in which case they could be either co-heads or complex predicates. It is not easy to determine what the requirements are in establishing whether a control relation exists or not. This is especially problematic when a complementiser li needs to be obligatorily present, as is the case with the verb wiegħed ‘promise’, whose English counterpart is a SUBJ control verb. Another issue related with Maltese is that given that the language only has finite morphological clauses, and given the case that an XCOMP is typically related with non-finite clauses in LFG analyses, one might have to potentially search for distinctions between something being non-finite morphologically vs. being non-finite syntactically. If this leads to nowhere then one must have to say that in such language types, i.e. where infinitival morphology does not exist, a finite clause can bear an XCOMP GF.

12:30-1:00: Polysemy in Javanese verbal

suffixes: the case for a revised LMT (Charlotte Hemmings, SOAS)

In this talk, I will discuss the verbal suffixes –i and –aké in Javanese, which both have a range of causative, applicative and aspectual meanings and sometimes alternate with each other in constructions that resemble the dative and spray-load alternations. Though causatives and applicatives are sometimes grouped under the heading of valency increasing derivations, traditional analyses have tended to emphasise the differences between the two constructions. That is, whilst applicatives are treated as morphosyntactic alternations – affecting the linking of arguments to syntactic functions but not the meaning – causatives are viewed as morphosemantic, introducing an entailment of causation and thus altering the lexical conceptual structure of the predicate. Given this distinction, we are forced to ask whether this is a case of polysemy or homonymy. In other words, are we dealing with a single form that has multiple meanings or separate affixes that happen to have the same phonological realisation? I argue on the basis of cross-linguistic evidence that this is a case of polysemy and propose to account for the patterns using a predicate composition approach, following Alsina (1996) and Austin (2005).

Arka et al (2009) suggest a similar approach for the cognate suffixes –i and -kan in Indonesian. However, they suggest that argument fusion occurs on the basis of like thematic roles. I argue that a proto-roles analysis (whereby the suffixes introduce different entailments for their arguments) can better account for the range of meanings discussed above, whilst simultaneously avoiding the controversy surrounding thematic roles. I then look briefly at the implications that this has for LMT and outline a preliminary revision, following Zaenen (1993) and Singh (1992), that moves away from the intrinsic classification of certain roles and towards assigning features on the basis of relative number of proto-patient properties ([+/-o]) and relative level of embeddedness ([+/-r]).

1:00-2:30: lunch break

2:30-3:00: Constraints on Latin discontinuous NPs from an LFG perspective (Liselotte Snijders, Oxford)

In this talk two main issues concerning discontinuity of Latin NPs will be discussed from an LFG perspective. Even though it is often assumed that Latin word order is free and that therefore discontinuity is not subject to any constraints, it appears that the discontinuity of Latin NPs is in fact constrained. Two constraints on this type of discontinuity form problems for LFG as a theory of syntax. For the first constraint, one on discontinuous PPs, an exception to the principle of Economy of Expression is proposed. This paper claims that c-structure should be constrained more in LFG. A second constraint on Latin discontinuous adjuncts shows that the way in which LFG treats adjuncts (with a set notation) is problematic for a proper account of discontinuous adjuncts. This issue will be addressed in the talk, but unfortunately no real solution for it has been found.

3:00-3:30: Weak Crossover and the Direct Association Hypothesis (Prerna Nadathur, Oxford)

3:30-3:45: break

3:45-4:45: Modern Greek Polydefinites and Relative Clauses (Kakia Chatsiou, SOAS)

4:45-5:15: Discussion and planning for next time