MIT hosted the 33rd meeting of the North East Linguistic Society, November 8--10, 2002. One conference participant offered an apt characterization of the meeting: "many nice papers, and lots of unfinished problems, always the best kind". The unfinished problems, sharpened and enhanced considerably by the presenters, made for lively conversation throughout the long weekend. The invited talks by Donca Steriade and Arnim von Stechow pulled together and informed themes heard throughout the regular proceedings, and the special session on nonconfigurational languages, in honor of Ken Hale, showed that broad descriptive coverage and theoretical innovation can go hand-in-hand (at least in the hands of experts).
The MIT students who organized the conference are to be commended for putting on a flawless show. The huge amount of time and energy they no doubt poured into the conference made it all appear smooth and effortless from the outside.
The meeting seems fairly characterized as one concerned primarily with the syntax--semantics interface as conceived of in broadly transformational terms. On the whole, differences between the syntax sessions and the syntax--semantics sessions were slight, just a matter of emphasis and presentation. This is in large part because, within the transformational tradition, both syntacticians and semanticists have hit upon a common question: what happens when functional heads, the basis for Minimalist theorizing, are assigned nontrivial semantic denotations? In general, the strategy is to locate type-shifting functions in these heads, so that every shift in meaning corresponds to a local syntactic tree; the mapping from LFs to lambda terms is thus transparent, and it is the job of both syntacticians and semanticists to figure out what the interpreted objects look like. (The PF--LF-only view of the grammar means that many purely syntactic generalizations are stated over PF structures.)
The meeting's syntax was entirely Minimalist; there wasn't a functional structure, AVM, directional slash, or labelled edge on display. The phonetics and phonology work was significantly broader in theoretical orientation. In phonology, purely constraint-based, minimally derivational, and fully derivational viewpoints were on offer. And a comparatively high number of the "P" talks reported experimental results in phonetics. But overall, phonetics and phonology took a back seat, with both placed under the heading "Phonology", which occupied just two regular sessions out of nine. But the work reported on in these sessions indicates that both fields remain lively areas of research.
This review touches upon all and only the presented talks. A listing of
the alternates and poster presentations, along with their abstracts, is
available at the conference website:
Donca Steriade (MIT) delivered the Saturday evening invited lecture. 'Wellformedness conditions vs. lexical generalizations: The morpho-phonology of Romance agentives' was an ambitiously wide-ranging discussion of the morphophonology of (mainly) Romance agentives and paradigmatically related forms. Steriade traced changes in these forms over time, searching for the mechanisms that result in morphological reanalysis and, in turn, regularities and irregularities in and among paradigms.
Arnim von Stechow (Tübingen) drew conference participants
back to the site on Sunday morning, despite a late night for most on
Saturday. Irene Heim introduced him as an influential teacher,
pivotal in the education of most working semanticists. 'Binding by
verbs: Tense, person and mood under attitudes' imparted a sense for these
strengths. The lecture was an expansion of ("an extended footnote to")
Philippe Schlenker's 1999 MIT dissertation on indexical expressions and
intensional contexts. Von Stechow opened with an informative review of the
behavior of various intensional parameters inside attitude reports. He
cast this discussion within the descriptive apparatus of the Minimalist
Program, ensuring a close connection with his audience and the majority of
the other NELS syntax/semantics talks. (He also sprinkled the lecture with
impromptu remarks linking his work with papers presented during the
regular session.) Time constraints kept him from an in-depth discussion of
temporal adverbs and the status of Kaplan's Prohibition Against Monsters.
(A Kaplanian monster quantifiers over contexts in the object language.
Schlenker embraces monsters; von Stechow claims that the prohibition
holds in the face of Schlenker's data.) The handout provides the
arguments, and the associated paper is downloadable from von Stechow's
http://vivaldi.sfs.nphil.uni-tuebingen.de/~arnim10/. The work is
informative enough to work as part of a textbook, but also packed full of
Special session on nonconfigurationality
A highlight of NELS 33 was its special session on nonconfigurationality dedicated to the memory of Ken Hale. Hale's pioneering work on nonconfigurational languages, notable for its descriptive and theoretical richness, continues to define and inform current research. Hale formulated an articulated definition of 'nonconfigurationality', in terms of free word order, widespread pro-drop, and discontinuous constituents. He also infused this definition with a theoretical interpretation centered around the flat phrase structure of Hale 1983.
Norvin Richards (MIT) delivered the opening address for the special session. Richards situated Hale's work in linguistic theory, touching on some of the highlights of Hale's long and productive career (which continues: Hale and Keyser 2002!). Richards' eloquent remarks emphasized that Hale's work remains at the center of current debate, and is the subject of continued refinement. The special session's invited lectures were a testament to this; all three negotiated and critically reviewed proposals that stem directly from Hale.
Judith Aissen (UC Santa Cruz) opened with 'Bidirectional optimization in non-configurational contexts'. She showed how bidirectional OT informs issues of categorical semantic blocking effects in cases where the morphology does not determine only one structure as possible for a given string. The talk was characteristically rich in its factual coverage: Chamorro and K'ichee were the focus, but Navajo played a crucial role as well, in large part because Aissen offered an alternative to the proposal of Hale et al. (1977) that certain cases of overgeneration stem from a parsing strategy. Aissen traced the analogous restrictions in Chamorro and K'ichee to interpretability, and made this precise using an OT model in which outputs are triples consisting of a phonetic form, a semantic representation, and a syntactic structure. Sets of these objects undergo both interpretive and expressive optimization. Interpretive optimization fixes as optimal a semantic representation for a given input phonetic form. This in effect filters the candidate set for expressive optimization, which determines the surface realization.
Mark Baker's (Rutgers) lecture 'Agreement and dislocation: New evidence from Mapudungun' began with an endorsement of Hale's basic descriptive characterization of nonconfigurationality, using it to establish Mapudungun as a nonconfigurational language. But, as is well known, Baker parts with Hale on the issue of the structures involved, favoring a view in which fully lexical phrases are adjuncts to a structured core containing pronominal arguments. Baker used parallels with dislocation structures in English to argue in favor of this full-nominals-as-adjuncts perspective for Mapudungun. These preliminaries were in the service of his argument that apparent topicalization in Mapudungun involves a nondemotion passive: the "topicalized" element occupies the specifier of TP. This accounts for why apparent topicalizations pattern with passives and against true extraction structures in nearly every respect. Theoretically, it means that the agreement features determined by the topicalized phrase need not be located on A-bar heads. Rather, nondemotion passive can be analyzed as involving agreement features on the Case-marking head T, in line with Baker's earlier proposal that agreement features appear on all and only Case-marking heads (the basis for the claim that agreement absorbs Case, forcing pronounced arguments to be realized in noncase positions wherever there is robust agreement).
The starting point for Mamoru Saito (Nazan), in 'On the role
of selection in the application of merge', was Hale's configurationality
parameter, first presented at NELS 12 (Hale 1982).
the parameter to reflect current Minimalist theorizing: "Merge implies
selection. Selection implies Merge". English enforces both conditionals,
whereas Japanese denies both. (Chinese is subject only to the first.)
Saito used this parameter to explain phenomena in Japanese ranging from
("radical") reconstruction across syntactic island boundaries to the
gradable markedness of double accusative clauses. A notable feature of
Saito's proposal is that it took the core insights of Hale's work, adapted
them to the derivational model that dominates theorizing at present, and
showed that the new perspective answers many of the objections of earlier
critics. It was a fitting note on which to close the special session.
During the first Friday regular session, OT received a critical treatment from two rather different perspectives. Working within the highly restrictive theory of Declarative Phonology (DP), Patrick Bye (CASTL/Tromsø) took on the challenging paradigm of non-possessed absolute and possessed form in Tiberian Hebrew. Surprisingly, these facts yield to a treatment in DP, a unification-based approach that countenances only outputs. The primary goal of the talk was to argue that DP, coupled with an articulated view of naturalness as a guiding principle in phonological systems, can handle opaque alternations.
Bert Vaux (Harvard), in 'Iterativity and optionality', also argued for the superiority of DP ... where DP now stands for 'derivational phonology' and thus exists at the opposite end of the spectrum of current phonological theories from Declarative Phonology. Vaux argued for the superiority of rule-based approaches based on iterative application of rules and optionality phenomena (especially of the 'intralevel' kind). Such patterns are not easily described (if describable at all in a linguistically natural way) within classical OT or many of its enriched versions. Vaux highlighted some essential ingredients for a phonological theory: (i) the means to state the optional/obligatory distinction; (ii) the means for allowing the effects of a single condition to be felt in multiple places; and iii the means to combine (i)--(ii). I think it remains to be seen whether a model-theoretic approach that does not invoke optimality, but relies instead on constraint satisfaction in the usual sense from logic, can meet these demands.
Although Eun-Sook Kim (UBC) employed the concepts of OT, her talk is of a piece with Vaux's in the sense that it exposed a limitation of standard OT. In 'Patterns of reduplication in Nuu-chah-nulth', Kim took on the complex reduplication system of Nuu-chah-nulth, a Wakashan language spoken on the west coast of Vancouver Island. Kim argued that the system does not yield to a treatment based on the usual techniques for reduplication (pre-OT templates or basic OT markedness). However, rather than rejecting the OT paradigm, Kim called upon the relatively new notion of Anti-Faithfulness, which (via negation) requires a global check on candidates rather than the local evaluations (based in pairs of nodes) that are the hallmark of regular faithfulness.
The first syntax--semantics session on Friday opened with Julie Anne Legate's (Harvard) 'The hows of Wh-scope marking in Walpiri'. Legate treated Walpiri scope marking constructions in terms of discontinuous constituents, thereby tying the proposals back to the work of Ken Hale and the nonconfigurationality session earlier in the day. Legate also sought an explanation for Walpiri's cross-linguistically unusual use of a how-type element, here nyarrpa, in scope marking constructions. She linked this with the parallel use of nyarrpa to head attitude reports, and drew analogies with the peculiar use of English how to introduce declarative clauses (They told us how it isn't smart to put your hand into the trash compactor). Since the two main ingredients of her analysis of Walpiri scope marking --- discontinuous constituency, the nature of clausal complements to verbs of saying --- are common in the language, she provided a natural basis for understanding how children acquire this construction despite the apparent poverty of relevant examples.
At an abstract level, discontinuous constituents were central to Shoichi Takahashi's (MIT) 'More than two quantifiers' as well. Takahashi decomposed comparative quantifiers like more than three books into a pair of generalized quantifiers (-er than three and many books; see Hackl 2000). He coupled this with Shortest Move (violable only by scope shifting that has semantic effects, as in Fox's work) and another condition (borrowed from the work of Heim and Kennedy) preventing a degree quantifier -er than three from taking scope over a quantifier that itself scopes over the t1 many books component. These assumptions form the basis for an account of the inability of comparative quantifiers to take wide scope with respect to other operators.
In 'Deducing the future', Tom Werner (Rutgers) developed a lexically-based proposal for the various temporal and nontemporal readings of English modals, working in the framework for describing modalities first set out by Kratzer (1981). Werner's discussion was model-theoretic in every sense: most of his attention was devoted to finding tense models with the right structure to provide an intuitively correct semantics for modal statements with temporal elements; his generalizations about, e.g., the differences between the past and the future were given entirely in terms of the look of the structures involved. The talk was a welcome (re-)connection between the more purely logical work on temporal model theory of Prior and Belnap. Links with the work on hybrid logic by Patrick Blackburn and his colleagues are, I hope, in the offing.
Do we stretch the notion of 'event' too thin when we apply it to predicative clauses? Claudia Maienborn (Humboldt Universität/ZAS Berlin), in 'Against a Davidsonian analysis of copula sentences', proposed to sharpen our semantic ontology by supplementing the usual Davidsonian theory of events with 'Kimian states'. Maienborn showed that, assuming events are perceivable, located in space and time, and subject to variable realization, copular clauses fail to qualify as event-denoting. They cannot serve as the complements to verbs of perception, and they resist modification by locative, manner, and certain quantitative adverbials. These facts pave the way for a treatment of such clauses as denoting Kimian states, which Maienborn located between eventualities and facts on a 'spectrum of world immanence'.
Tania Ionin (MIT) and Ora Matushansky
(CNRS/Université Paris 8) addressed the distinctions between
nominative and instrumental case on Russian predicates. Their
'Encasing the time: Effects on Russian predicative case' located
instrumental case in the head of an Aspect phrase, whose head has a
nontrivial semantic denotation. Their talk was wide-ranging in
its factual support for the claim that the two morphological
cases differ significantly in their semantics,
and a final section of the handout addresses an impressive array of
crosslinguistic data that should inform extensions of their ideas
outside of Russian.
Eric Potsdam (University of Florida) opened Saturday's proceedings with 'Evidence for semantic identity under ellipsis from Malagasy sluicing'. Potsdam sought to support the position of Merchant (2001) that the identity criterion for ellipsis is semantic rather than syntactic. The Malagasy evidence for this position, and against a syntactic identity criterion, is compelling; the Malagasy version of *Someone offered a devastating counterexample, and we can all guess by whom is grammatical! Potsdam's drew much of his evidence from the morphosyntax, and in turn devoted considerable energy to fending off alternatives, in particular an analysis that assumes ellipsis is insensitive to (most) morphology and another that permits ellipsis to rescue otherwise ungrammatical structures.
The topic of Junko Ochi (Osaka) and Masao Ochi's (Osaka/UConn/MIT) 'Reciprocity and null operator movement' was reciprocals in Madurese, in which the preverbal element saleng acts as a distributor. Ochi and Ochi's structures crucially involved a null reciprocator, which moves from a gap position to the location of the distributor (i.e. saleng) in overt syntax, with saleng raised to the position of the plural NP in covert syntax. They used the presence of negative island effects to argue that this null reciprocal operator movement is an instance of A-bar movement.
Broadly speaking, the topic of 'LF-intervention effects in pied-piping', by Uli Sauerland (UMass/Tübingen) and Fabian Heck (Stuttgart), was the question of which aspects of the analysis of whose-type questions fall to the syntax and which to the semantics. Recent literature provides possible analysis in terms of a complex denotation for whose movie; these are generally intended to replace von Stechow's proposal, which raises the Wh-word in the covert syntax. Sauerland and Heck argued for covert raising; their main support came from intervention effects at the hands of quantifiers, adverbs, locatives, and others. These results provide a novel perspective on the subject of island conditions and constraints on movement: although the movement Sauerland and Heck proposed is a (syntactic) left-branch island violation, it is evidently subject to a wide range of (semantic) island-type constraints of its own.
The aim of Kimiko Nakanishi's (UPenn) 'Semantics of measure Functions' was a statement of the monotonicity constraints on Japanese split and non-split quantifiers involving measure or classifier phrases like san-rittoru ('three-liters'). Building on work by Schwarzschild (2002), Nakinishi showed that the arguments to such functions must denote objects with a part--whole structure --- a semi-lattice of individuals in the nominal domain, a semi-lattice of events in the verbal domain. The additional constraints concern the nature of the measure phrases themselves, hinging on whether they denote monotonic functions. In the nominal domain, non-split quantifiers must be monotonic. In the verbal domain, split quantifiers must be monotonic. Ungrammatical cases of apparently split nominal quantifiers (mizu-ga tukue-nouede san-do kobore-ta ('water table-on three-degrees spilled')) reduce to the constraint on the verbal domain: degree', a non-monotone function, applies to the predicate in this example.
After warning that there would be some foul language, Christopher Potts (UC Santa Cruz), in 'Expressive content as conventional implicature', used expressive content modifiers (friggin', the jerk) to motivate a two-dimensional semantics for conventional implicatures, based broadly on the description logic proposed by Karttunen and Peters (1979). Potts emphasized that Grice's original definition of 'conventional implicature' identifies a class of speaker-oriented entailments that are not presupposed.
After Potts's study of the conventional implicature dimension, attention shifted to the focus dimension of natural language meanings, with Luisa Martí's (UConn) 'Only, reconstruction, and informativity'. It is common in studies of the semantics of focus to assign a lot of work to a free variable C, but then to specify simply that its value is "pragmatically determined". Marti took up the challenging task of investigating just where this C gets its meaning from. Her goal was a theory that treats C like any other variable, subject to none of the special restrictions on it that are found in Rooth's work. Her novel proposal is that C finds an antecedent in the same way that pronouns do: in the context. Since C denotes a set of propositions, it can be anaphoric to a (Hamblin--Karttunen) question, also a set of propositions. The theory requires default contexts to be able to contain questions, a premise that receives support from recent work on discourse structuring.
The multi-dimensional theme carried over from the semantics session into the first syntax session of the day (and in turn from the realm of meanings to the realm of syntactic structures). Barbara Citko (UConn), in 'ATB Wh-movement and the nature of Merge', argued for a multi-dimensional view of coordinate structures, calling upon mainly Polish data as factual support. Citko adeptly negotiated the question of whether moving beyond two-dimensions entails an unacceptable loss in computational tractability or insurmountable linearization problems. (I refer also to Rogers 2000 for additional grounds to dismiss these charges as unfounded.) She then showed how a parallel-merge approach to questions involving across-the-board extraction provides a natural basis for stating many familiar and not-so-familiar conditions on such movement.
Barbara Ürögdi (SUNY at Stony Brook) proposed a derivational analysis of 'doubling' verb modifiers in Hungarian. In such constructions, a preverbal particle matches the features of an oblique phrase (e.g., Anna rá írta a vers-et a táblá-ra ('Anna onto-wrote the poem the-board-onto')). The doubling marker enforces an accomplishment reading. Ürögdi argued that the markers are not located in lexical entries, but rather derived by overt 'feature copying' of properties of the oblique into the specifier of VP, an argument position. The copied features then move to the specifier of AspectP.
Alan Yu (UC Berkeley/McGill) opened Saturday's only regular phonology session. 'From Metathesis to Infixation: The story of Pingding' focussed on Pingding, a dialect of Mandarin Chinese in which diminutives are marked with a pre-nuclear retroflex lateral. The result of infixation is often a marked syllable, as infixation can create onset clusters. Such derived markedness effects are, Yu argued, extremely problematic for OT accounts. Yu offered a diachronic explanation appealing mainly to acoustic phenomena.
Jon Nissenbaum (Mass Eye and Ear), John E. Kirsch (Siemens Medical Systems), Morris Halle (MIT), James B. Kobler (Mass Eye and Ear), Hugh D. Curtin (Mass Eye and Ear), and Robert E. Hillman (Mass Eye and Ear), in 'High-speed MRI: A new means for assessing hypotheses concerning phonetic control of voicing and F0', described a method for overcoming the limitation inherent in magnetic resonance imaging as a method for recording and studying the articulatory gestures underlying speech sounds.
Zhiqiang Li's (MIT) 'A perceptual account of asymmetries in
tonal alignment' discussed the interaction of perceptual and acoustic
constraints on rising tone in Mandarin. Relying on experimental results,
Li offered a perception-based account of tonal alignment; the talk was one
of the few at NELS 33 that addressed the nature of the
Orin Percus (San Raffaele/Tübingen) and Uli Sauerland (Tübingen) teamed up for 'Binding restrictions in dream reports'. The talk sought to account for the absence of certain de se--de re combinations in dream reports. Percus presented; as we've come to expect from his work, much depended upon the careful construction of specific scenarios to highlight certain readings. I make no attempt to reconstruct them here. Suffice it to say that Percus and Sauerland located the restriction in the LF meaning language, as a condition based on asymmetric c-command between the relevant pronouns. They drew an explicit connection with superiority effects, but noted that the generalization might also be assimilable to some form of crossover. The important point is that familiar and systematic constraints are at work in this dreamy, unfamiliar domain. For a sense of the way this project fits into the broader research program of Percus, Sauerland, and others, I refer readers to the Tübingen Variables Project webpage: http://www2.sfs.nphil.uni-tuebingen.de/home/variables/www/variables_english.html.
The final semantics session closed with 'Comparative constrictions in Japanese versus English' by Sigrid Beck, Toshiko Oda, and Koji Sugisaki (all of UConn). Their principal claim was that Japanese yori constructions are not based in degree operators. Rather, they more closely resemble (though contrast in notable respects with) the English free adjunct construction compared to, as in Compared to Bill, Sally is tall.
Emi Mukai's (Kyushu) 'On verbless conjunction in Japanese' analyzed some ellipsis constructions in Japanese, with most of the attention on a gapping-like construction (Tom-ga Mary-ni, Mike-ga Susan-ni tyokoreeto-o ageta ('Tom Mary, Mike Susan chocolate gave')). Mukai proposed that these deletions require "identity of phonetic strings at PF, ignoring constituency". She contrasted this with the conditions on stripping in Japanese, which requires only semantic identity. Mukai also offered a useful metaproposal for shielding speakers from aggressive or abusively inarticulate questions: her handout provided a 'Question Menu', from which audience members could choose succinct and fruitful questions. To my mind, Q2 and Q5 were the juiciest on offer, since both touched upon the contrast between her claims about PF identity criteria and those Eric Potsdam reported on Saturday, which centered around a purely semantic account of Malagasy sluicing.
Liina Pylkkänen's (NYU) 'Distributing depictives' began by asking why the first object in an English double object construction cannot take a depictive complement (as in *I gave Mary the meat hungry), whereas comparable structures in Luganda (a Bantu language) are grammatical. Pylkkänen argued that a complex predicate analysis of depictives (rather than assimilation to control) provides the basis for an explanation: by coupling this with a distinction between low applicatives (English) and high applicatives (Luganda), she derived the contrasts from considerations of semantic well-formedness.
Jonathan Bobaljik (McGill) and Susi Wurmbrand (McGill)
explored the variation found cross-linguistically in long distance
agreement systems. In 'Long distance object agreement, restructuring
and anti-reconstruction', Bobaljik and Wurmbrand made sense of the
variation with a phase-based account, in which agreement is always
between 'phase-mates'. Central to the description was the definition of
phases as externally inaccessible save for their highest specifiers. This widely
accepted definition leads directly to an explanation for why matrix
agreement with an embedded object is often contingent upon topicalization
of that object (sometimes covertly). Coupled with assumptions about
restructuring, it also yields an account of certain scope restrictions in German
Following Donca Steriade's lecture on Saturday evening, conference
participants ambled over to the MIT museum for the conference party. Far
and away the coolest thing about this party was that it was held, not in
some drab banquet hall under the museum's roof, but rather right in among
the museum's exhibits. One of the bars was across the room from the first
LISP machine, and the main buffet table stood next to computers doing some
sort of natural language processing. There were probably few
conversational lulls, but any such down time could be filled by staring at
mesmerizing holograms or toying with bizarre gadgets.
Hackl, Martin. 2000. Comparative quantifiers. Doctoral dissertation, MIT, Cambridge, MA.
Hale, Kenneth; LaVerne Masayesva Jeanne, and Paul Platero. 1977. Three cases of overgeneration. In Peter W. Culicover and Thomas Wasow, and Adrian Akmajian, eds., Formal Syntax, 379--416. New York: Academic Press.
Hale, Kenneth. 1982. Preliminary remarks on configurationality. Proceedings of NELS 12, 86--96.
Hale, Kenneth. 1983. Walpiri and the grammar of nonconfigurational languages. Natural Language and Linguistic Theory 1:1 (5--49).
Hale, Ken and Samuel Jay Keyser. 2002. Prolegomenon to a Theory of Argument Structure. Cambridge, MA: MIT Press.
Karttunen, Lauri and Stanley Peters. 1979. Conventional implicature. In Choon-Kyu Oh and David A. Dinneen, Syntax and Semantics. Volume 11: Presupposition, 1--56. New York: Academic Press.
Kratzer, Angelika. 1981. The notional category of modality. In H.-J. Eikmeyer and H. Rieser, eds., Words, Worlds, and Contexts. New Approaches in Word Semantics, 38--74. Berlin: de Gruyter.
Merchant, Jason. 2001. The Syntax of Silence: Sluicing, Islands, and Identity in Ellipsis. Oxford: Oxford University Press.
Rogers, James. 2000. Syntactic structures as multi-dimensional trees. To appear in the Journal of Language and Computation. Special issue including papers from the ESSLLI 2000 Workshop on Trees. Available at http://www.cs.earlham.edu/~jrogers.
Schwarzschild, Roger. 2002. The grammar of measurement. To appear in Proceedings from SALT XII. Ithaca, NY: CLC Publications.