The Weekly Newsletter of MIT Linguistics

Issue of Monday, November 5th, 2018

LingPhil Reading Group 11/5 – on Cable (2018)

Maša Močnik and Keny Chatain will be presenting a paper by Cable on a new choice-function-based approach to De Re/De Dicto. The meeting will take place on Monday 5th in the 8th floor seminar room.

Title : A Choice Functional Semantics for De Re Attitudes.

Author(s) : Seth Cable

Abstract :

In this paper, I present a relatively simple syntax and semantics for de re attitude reports, one that builds heavily upon the approaches developed by Percus & Sauerland (2003), Charlow & Sharvit (2014), and Sauerland (2014). The principle innovation of this account is that such readings involve quantification over choice-functions, rather than direct quantification over accessibility relations (Cresswell & von Stechow 1982), concept generators (Percus & Sauerland 2003, Charlow & Sharvit 2014), or counterpart functions (Sauerland 2014). I show that the proposed system accounts for the same range of complex data and judgments as prior approaches, particularly (i) the behavior of bound de re pronouns (Charlow & Sharvit 2014), (ii) interactions between de re readings and non-upward monotone operators (Charlow & Sharvit 2014), (iii) constraints on de re readings of complex DPs (Sauerland 2014), and (iv) the inability for DPs to be construed de re when they contain opaquely interpreted NPs (Charlow & Sharvit 2014). Despite this equivalent empirical coverage, the proposed system offers a relatively simple compositional semantics for de re attitude reports, one where (i) the object-language quantification at play is of a type independently observed in natural language, (ii) attitude verbs do not require a ‘type-flexible’ denotation, (iii) the intensional argument of an attitude verb is simply a proposition (or property of entities), (iv) there is no lexical ambiguity between ‘existential’ and ‘universal’ readings of such verbs. I show that the facts which motivated the complexities found in prior accounts – particularly the interactions between de re readings and quantificational DPs – can follow from independently known properties of choice-functional quantification. [[Author’s Note (9/7/2018): The general approach presented in this paper was independently suggested in earlier, unpublished work by Orin Percus and Simon Charlow: Percus (2013) “Res-Marking in Belief Reports”, Charlow (2014) “Locality in the Syntax and Semantics of Attitude Ascriptions.”]]

MorPhun 11/5: Ray Jackendoff on Relational Morphology

Speaker: Ray Jackendoff (Tufts/MIT) and Jenny Audring (Leiden)
Title: The Texture of the Lexicon: Relational Morphology in the Parallel Architecture
Date and time: Monday 11/5, 5-6pm
Location: 32-D831

Linguistic theory has emphasized Humboldt’s “infinite use of finite means,” to the relative neglect of the “finite means,” i.e. the lexicon. What does a language user store in the lexicon,and in what form? We explore this question in the context of the Parallel Architecture (Jackendoff 1997, 2002; Culicover and Jackendoff 2005). Within this outlook, lexical items are pieces of phonological, (morpho)syntactic, and semantic structures, and morphology is the grammar of word-sized pieces of structure.

Unlike classical syntactic patterns, the majority of morphological patterns are not productive, and their instances must be listed in the lexicon. They therefore present a number of important difficulties for a grammar formulated in terms of traditional rules. More adequate is a formulation in terms of schemas, along the lines of Construction Grammar and especially Construction Morphology. Nonproductive schemas do not build new structures; rather, they motivate relations among items stored in the lexicon.

In addition to building novel structures, productive schemas can also motivate relations within the lexicon. This leads to a new perspective on productive patterns: the principles used to build novel structures are simply a subset of the schemas in the lexicon, and productive schemas canbe thought of as schemas “gone viral.” We conclude that the focus in linguistic theory on the“infinite use of finite means” has deflected attention from a more basic issue: the form of the lexicon and the relationships within it.

This outlook on morphological relations in the lexicon has consequences for syntactic theory as well. First, it offers an account of nonproductive syntactic constructions – Culicover’s (1999)“syntactic nuts.” Second, it suggests a nontraditional approach to syntactic alternations, in which movement operations are replaced by relations between independent constructions. We will briefly mention some possible cases.

Syntax Square 11/6 - Itai Bassi on Bruening (2014)

Speaker: Itai Bassi on Bruening (2014) “Precede and command revisited
Date and Time: Tuesday, November 6, 1-2pm
Location: 32-D461

I will present Benjamin Bruening’s (2014) paper ”Precede and command revisited” (click here for a link). In this paper Bruening argues that the relation of c-command is irrelevant to binding phenomena, and proposes to replace it with a coarser notion of command, “phase command”.  To give a taste of the argumentation, consider the known fact that (1) feels like a principle C violation, although “her” doesn’t c-command “Rosa”; no constituency test tells us that “her” in (1) forms a constituent with “about Rosa’s son” to the exclusion of everything else.

(1) *We talked with her_1 about Rosa_1’s son. (Reinhart 1976)
Furthermore, Bruening argues that precedence also plays a role in accounting for binding phenomena. Hence “precede and (phase-)command” is the correct relationship underlying binding effects.
We will discuss and evaluate Bruening’s attempt to derive “precede and command” from a particular view of the grammar where sentences are built left-to-right, phase by phase.

LF Reading Group 11/7 - Kai von Fintel (MIT)

Speaker: Kai von Fintel (MIT)
Title: Conditional desires
Date and time: Wednesday, November 7th, 1-2 pm
Location: 32-D461


Sentences with an *if*-clause and a desire predicate have a “restricted” reading that claims that among the worlds described by the *if*-clause, the attitude holder prefers certain worlds. Deriving this reading compositionally reveals hidden properties of conditionals and attitude predicates. There are intriguing connections to puzzles in the analysis of deontic conditionals (*if p, ought q*) and conditional imperatives (*if p, do q!*).


Phonology Circle 11/7 - Gaja Jarosz & Aleksei Nazarov

Presenter: Gaja Jarosz (UMass Amherst) & Aleksei Nazarov (University of Toronto) 
Title: Domain-general learning of stress parameters: A typological investigation
Date/Time: Wednesday, November 7, 5:00-6:30pm
Location: 32-D831

Existing learners for stress in OT (e.g., Tesar and Smolensky 2000, Boersma and Pater 2013/2016, Jarosz 2013, 2015) are all so-called domain-general learners (need not be specified in UG). However, Pearl (2007, 2011) argues that Principles & Parameters models of stress (e.g., Dresher and Kaye 1990) do necessitate domain-specific learning mechanisms (specified in UG). Nazarov and Jarosz (2017) proposed a domain-general learning with more inference power than Yang’s (2002) learner (which Pearl 2011 showed didn’t work for stress parameters), arguing that domain-general learners may be sufficient for both OT and parametric models.

However, Nazarov and Jarosz (2017) only tested their learner on a small set of stress systems with a subset of Dresher and Kaye’s parameters. To validate these initial results, we have now tested our learner, in addition to 3 versions of Yang’s (2002) learner, on all 280 stress systems that are possible in Dresher and Kaye’s (1990) framework. Similarly to Nazarov and Jarosz (2017), it is found that the proposed learner converges on the correct stress system about 95% of the time, while Yang’s learner does so no more than 10% of the time. To further compare the learners, we have extracted a few statistics on the stress systems themselves that predict their performance under each learner. Finally, we found that the stress systems never learned by Nazarov and Jarosz’s learner are unattested (save for two cases that warrant further investigation), suggesting that the only function of domain-specific learning mechanisms might be to enable the learning of unattested languages.


Ling-Lunch 11/8 - Giuseppe Ricciardi (Harvard)

Speaker: Giuseppe Ricciardi (Harvard)

Title: “She must be mad” Really? Must? Epistemic must statements are felt “weak” because they are usually false [work with Edward Gibson and Rachel A. Ryskin]
Date and time: Thursday, 11/8, 12:30-1:50pm
Location: 32-D461
Two main hypotheses have been proposed in the literature about the semantics of epistemic ‘must’. The Strong Must Hypothesis (von Fintel & Gillies, 2010) and the Weak Must Hypothesis (Karttunen, 1972; Kratzer, 1981; Lassiter, 2017). Lassiter (2016) offers an experiment whose results support the Weak Must Hypothesis. I argue that Lassiter (2016)’s task is invalid for testing participants’ competence about truth-conditions of statements and I show that if one modifies Lassiter’s task so that participants understand it as a truth-value task then the results support the Strong Must Hypothesis over the Weak Must Hypothesis. Moreover, Lassiter (2016) offers some weak uses of epistemic ‘must’ derived from a corpus. I discuss these examples as well and challenge their status as piece of evidence supporting the Weak Must Hypothesis.
Overall, I argue for the following claim: the weakness observed since Karttunen (1972) with respect to epistemic ‘must’ statements is not due to their being true in weak contexts but to their being systematically falsely uttered in weak contexts. To account for such systematicity, I propose that epistemic ‘must’ statements are often used hyperbolically.