SE-LFG32 (21 May 2022, Oxford)

32nd South of England LFG Meeting

The 32nd South of England LFG Meeting, a student-oriented meeting for presentations and discussion of various topics from an LFG perspective, will be held on 21 May 2022 at the Oriental Institute, University of Oxford from 1:00-4:30 PM. Contact Joey Lovestrand with any questions (jl119@soas.ac.uk).

Agenda

1:00 - 1:45 John J. Lowe & Davide Mocci

The syntax of Sanskrit bahuvrīhis

Abstract: Lowe (2015) provided a formal analysis of Classical Sanskrit compounds, including bahuvrīhis of the adjective-noun type (e.g. dīrgha-karṇa 'long-ear'), within the LFG framework. In this paper we show that Lowe’s (2015) analysis cannot account for the full range of bahuvrīhi types attested in Sanskrit, including in the earlier Vedic Sanskrit; we offer a revised and extended account which covers all types of bahuvrīhi and related formations.

1:45 - 2:30 Frances Dowle

Agreement in Welsh: an LFG approach using Lexical Sharing

Abstract: In Welsh, there are two aspects of index agreement which are unusual. Firstly, many types of index agreement (including verb-subject, verb-object, preposition-object and noun-possessor agreement) cannot cooccur with lexical noun phrases (NPs) nor ‘independent’ pronouns, but must occur with ‘dependent’ pronouns. Sadler (1997) proposes that Welsh agreement obligatorily carries a PRED = ‘pro’ value for the relevant argument, which prevents agreeing heads from occurring with lexical NPs. This paper extends Sadler’s account to address the independent/dependent pronoun distinction, proposing separate lexical entries for the two classes of pronouns (rather than a single lexical entry with an optional PRED = ‘pro’ value). Secondly, in long distance dependencies (LDDs), what superficially appears to be third-person singular masculine marking occurs on a head when its argument is displaced. In the first LFG account of LDDs in Welsh, I argue that this is not agreement but a kind of morphological marking of LDDs, similar to that found with Irish complementisers (Asudeh 2004, 2012). The Welsh data provides further insight into the use of Lexical Sharing (Wescoat 2002) as a tool for analysing clitics. The dependent pronoun is phonologically clitic and is analysed using Lexical Sharing. The dependent pronouns appear to be sensitive to the functional information of their host in LDDs and complement clauses, which supports a syntactic rather than purely phonological account of clitics.

2:30 - 2:45 Tea Break

2:45 - 3:30 Jamie Findlay & Dag Haug

Managing scope ambiguities in Glue via multistage proving

Abstract: It is well known that Glue Semantics is prone to overgenerate scope possibilities in semantic representations. There is no commonly accepted way to deal with scope islands or constraints on modifier scope, which has undesirable consequences both on the theoretical and the practical side. In this paper we propose a new solution to managing scope ambiguities, by controlling the structure of linear logic proofs via a separate level in the LFG projection architecture. This offers theoretical advantages in the description of modifier scope and scope islands, and practical advantages in the implementation of sub-lexical meaning decomposition favoured in much recent Glue work (e.g. Asudeh et al. 2014). Our proposal has been implemented using the prover in the Glue Semantics Workbench (Meßmer & Zymla 2018), but would work equally well with any other linear logic prover, since it does not constrain the proof algorithm or add any new connectives or modalities to the logic.

References:
Asudeh, Ash, Gianluca Giorgolo & Ida Toivonen. 2014. Meaning and valency. In Miriam Butt & Tracy Holloway King (eds.), Proceedings of the LFG'14 conference, 68–88. Stanford, CA: CSLI Publications.
Meßmer, Moritz & Mark-Matthias Zymla. 2018. The Glue Semantics Workbench: a modular toolkit for exploring Linear Logic and Glue Semantics. In Miriam Butt & Tracy Holloway King (eds.), Proceedings of the LFG'18 Conference, 249–263. Stanford, CA: CSLI Publications.

3:30 - 4:15 Chen Xie

How do Syntactic and Phonological Representations differ? A Graph Theoretic Perspective

While representing syntactic structure as trees is a widely accepted practice, there is controversy regarding the representation of phonological objects. One school of thought (Anderson, 1985) maintains that syntactic and phonological representations share the same structural properties, i.e., being arboreal. Others have been attempting to exclude trees from phonology, arguing that phonological trees overgenerate (Neeleman and van de Koot, 2006; Scheer, 2004, 2013). In this talk, I examine syntactic and phonological representations from the perspective of Graph Theory (e.g., Bondy and Murty, 2008). For syntax, I discuss c(onstituent)-structure and f(unctional)-structure in Lexical Functional Grammar. I will show that c-structure is a connected acyclic graph (i.e., tree). By convention, constituent trees encode both dominance and precedence relations, which are mutually exclusive (Partee et al. 1990). Having these relations simultaneously falls foul of the definition of the tree, so we need to exclude precedence and encode it elsewhere. In contrast, f-structure is a direct cyclic graph (formally equivalent to the attribute-value matrix). It shares many properties of the tree but allows multi-dominance in the case of structure sharing (e.g., functional control or long-distance dependency). For phonology, I discuss two types of syllable structure. The classic representation divides a syllable into Onset and Rhyme, the latter dominating Nucleus and Coda. This representation, like f-structure, is a directed acyclic graph. Another type is a constituent-less structure entertained in the framework of Strict CV Phonology. It is an Autosegmental Phonological Representation, which is essentially a bipartite graph. Such classification allows us to examine the similarities and differences of linguistic representations precisely because it prevents us from being misguided by notational differences of these representations. Against this background, I review and refute arguments favouring a flat phonology, as they share a common problem: they impose external properties (recursion, binding, long-distance dependency, etc.) onto trees, and mistakenly conclude that phonology is flat due to its lack of these properties. Their problem is more conspicuous when we examine c-structure, a Graph Theoretic tree that does not necessarily have these properties. Further, I will show that, for Strict CV Phonology, excluding trees from its representation leads to a contradiction, since its representation is actually a collection of trees. Not only does my research demonstrate the infeasibility of excluding trees from phonology, but it stresses the importance of having a formal, mathematically precise definition for linguistic objects. Without such formalisms, we may have the risk to confuse notations of linguistic representations with the representations per se.

Selected References
Neeleman, A. and van de Koot, H. (2006). On syntactic and phonological representations. Lingua, 116(10): 1524–1552.
Scheer, T. (2004). A lateral theory of phonology: What is CVCV, and why should it be? De Gruyter Mouton, Berlin.
Scheer, T. (2013). Why phonology is flat: the role of concatenation and linearity. Language Sciences, 39: 54–74.

4:15 - 4:30 Planning the next SE-LFG meeting