Whamit!

The Weekly Newsletter of MIT Linguistics

Archive for October, 2014

NELS at MIT this week

The 45th annual meeting of the North East Linguistic Society will be held at the MIT on October 31 - November 2nd, 2014. Invited speakers are:

Please visit the conference website for more information.

Phonology Circle 10/27 - no meeting this week

There will be no Phonology Circle meeting this week.

ESSL/Lacqlab 10/29 - TBA

Speaker and title: to be announced
Time: Wednesday 10/29, 5:00 pm
Room: 32-D831

Ling-Lunch 10/30 - Andreea Nicolae

Speaker: Andreea Nicolae (ZAS Berlin)
Title: Positive polarity and strength of scalar implicatures
Date/Time:Thursday, October 30, 12:30-1:45pm
Location: 32-D461

The goal of this talk is to offer a new analysis of positive polarity elements in light of the differences between weak and strong disjunction (“or” versus “either or”). I will be addressing the following three properties of disjunctive elements in this talk:
◦ the ease of cancelability (strength) of their scalar inference
◦ the PPI status of a disjunctive element
◦ the ignorance inference that accompanies both weak and strong disjunction

I will argue the following:
◦ the strength of the SI relates to the nature of the alternatives activated by the element at play, namely whether or not the scalar alternatives are obligatory (non-prunable).
◦ the PPI status is the result of an element’s appeal to non-vacuous exhaustification, i.e. proper strengthening.
◦ the epistemic ignorance inference is the result of a last resort way of avoiding a contradiction, namely via the insertion of a covert modal.

If time permits, we will also see how this account can be carried over to account for the PPI status of existential quantifiers.

Phonology Circle 10/20 - Laura McPherson

Speaker: Laura McPherson (Dartmouth University)
Title: Problems in Seeku plural formation
Date: Oct. 20 (M)
Time: 5:00 - 6:30
Place: 32D-461

On the surface, the plural in Seeku (Mande, Burkina Faso) is marked by some combination of tone raising, diphthong formation, vowel fronting, or nasalization. For example, bi21 ‘goat’ has the plural bi3 ‘goats’ with tone raising, ko2koː21 ‘rooster’ has the plural ko3koeː3 ‘roosters’ with diphthong formation and tone raising, dyo1ŋma3 ‘cat’ has the plural dyo1ŋmɛ3 ‘cats’ with vowel fronting, and sa21 ‘rabbit’ has the plural sɛ̃3 with nasalization, vowel fronting, and tone raising. I argue that the segmental changes are best understood as suffixation of a front vowel /-ɛ/, accompanied by vowel harmony ([ATR] and [high]) and vowel elision as a hiatus repair strategy. Thus, a form like ‘goat’ has the following derivation:

UR                   /bi-ɛ/
Harmony       |bi-i|
Elision             |b-i|
SR                     [bi]

While this analysis works in a derivational framework, it runs into trouble in a parallel model of phonology due to counterbleeding opacity between elision and vowel harmony: due to the largely monosyllabic nature of Seeku, most often the only vowel that remains in the plural (under this analysis) is the vowel of the plural suffix, yet it displays harmony with the elided vowel of the stem. In this talk, I show how the rule-based analysis accounts for the data then briefly discuss the varying levels of success of different constraint-based analyses, including standard I-O OT, output-oriented constraints, Harmonic Serialism, and contrast preservation.

Syntax Square 10/21- Kenyon Branan

Speaker: Kenyon Branan (MIT)
Title: A long distance subject/object extraction
Date/Time:Thursday, October 21, 12:30-1:45pm
Location: 32-D461

There are a number of ditransitive verbs that are able to take both a DP and a CP complement. A subset of these verbs exhibit an interesting asymmetry: long distance object extraction of a DP is grammatical, whereas long distance subject extraction of a DP is ungrammatical, even when licensing conditions for long distance subject extraction are fulfilled. Examples of ungrammatical subject extraction are given below.

(1) a.* Who did we convince them [ __ sighted Bigfoot]?
b.* Who did they persuade themselves [ __ should move to Canada]?
c.* What did they assure each other [ __ has sunk]?

Previous accounts of this [Stowell (1981), Bošković and Lasnik (2003)] attribute this ungrammaticality to licensing conditions for elements moved out of subject position. We take a different approach. We show that this ungrammaticality obtains only in cases where the extracted subject is a DP. We give evidence from two tests which suggest that the matrix subject of these verbs originates below [spec,vP]. Putting these two together, we argue that the ungrammaticality of sentences like (1) is the result of an intervention effect. The movement of a DP containing a wh-word to [spec,vP] creates a structure where T is unable to Agree with the low subject, the moved DP acting as an intervener.

We propose that there is a structural difference between long distance subject movement and long distance object movement. Long distance subject movement involves movement of a subject from the CP to matrix [spec,vP]. Long distance object movement involves two steps: movement of the CP to [spec,vP], and subextraction of the object DP from the CP. Crucially, long distance object movement does not create the asymmetric c-command relationship between two syntactic objects of the same type which characterizes intervention effects.

Ling-Lunch 10/23 - Aron Hirsch

Speaker: Aron Hirsch (MIT)
Title: Deconstructing exceptives
Date/Time:Thursday, October 23, 12:30-1:45pm
Location: 32-D461

This talk looks at the semantics of exceptive expressions like but and other than. Building on insights in Gajewski (2008, 2013), I pursue an analysis of exceptives as sharing a common semantic core: a form of subtraction. But in (1) takes John as its argument and returns the set of all entities (atomic or plural) which do not include John. The resultant meaning composes with students by Predicate Modification, yielding the set of students not including John. This set is the restrictor of every.

(1) Every student but John came.

I will argue for an analysis of but as obligatorily strengthened by the Exh operator of Fox (2007). Exh is responsible for deriving the entailment in (1) that John did not come. The literature (in particular, Gajewski 2013) has pursued this approach, but with additional complications, which I will argue are avoidable.

I will show how the analysis extends to account for further empirical puzzles, in particular the incompatibility of exceptives with both, all when there is a numeral present (Moltmann 1993), and singular definites. Each expression in (2) introduces a presupposition about the size of its restrictor: the presupposes uniqueness, both presupposes duality, and all six presupposes a cardinality of six. I will argue that presuppositions project universally out of alternatives over which Exh quantifies, and that the result is presupposition conflict in each of (2a-c).

(2) a. *Both students but John came.
b. *All six students but John came.
c. *The student but John came.

Finally, I will show that the analysis sheds light on the typology of exceptives. But and other than are both a spell-out of the subtraction operator. The dimension on which they differ is that the but allomorph can only occur with Exh, while other than can occur with or without Exh. The availability of a parse without Exh will account for the freer distribution of other than than but and its fewer entailments:

(3) Some student other than/*but John came.
(John could have come also, or not.)
Three students other than/*but John came.
(John could have come also, or not.)

I will motivate the claim that other than is nonetheless optionally strengthened by testing for a parse with Exh using Hurford’s disjunctions.

LFRG 10/23 - Yimei Xiang

Speaker: Yimei Xiang (Harvard)
Title: Mention-Some Readings of Indirect Questions: from Experiments to Formalizations
Time: Thursday, October 23, 5:30-7pm
Location: 32-D461

In this talk, I look for experimental clues and propose a schematized analysis for the following three problems about mention-some (MS) readings of indirect questions. First, which type(s) of indirect questions admit MS readings? Second, is there any MS reading sensitive to false answers (FAs)? Third, are FAs equally bad? Based on the results of five TVJTs on ATurk and the reanalysis of Klinedinst & Rothschild’s (2011) raw data, I find that (i) MS readings are also supported by indirect MA-questions under predicates like tell; (ii) there is an MS reading sensitive to FAs, in parallel to the intermediately exhaustive reading; and (iii) FAs are not equally bad, in particular, over affirmation is relatively more acceptable than over deny in MA-questions, while over deny is relatively more acceptable than over affirmation in MS-questions.

Syntax Square 10/14 - Heidi Klockmann

Speaker: Heidi Klockmann (Utrecht University)
Title: Case alternations and case hierarchies: A view from numerals and negation
Date/Time:Thursday, October 14, 12:30-1:45pm
Location: 32-D461

A number of Slavic and Uralic languages share the property of blocking structural case assignment in the presence of an oblique case assigner. This alternation has attracted the most attention in the context of numerals, whereby the case triggered by the numeral, genitive in Polish (1) and partitive in Finnish (2), fails to appear in oblique contexts.

(1) a. Iwan kupił pięć samochodów
Ivan   bought  five   cars.GEN
Ivan bought five cars
b. z pięcioma samochodami
with five.INST    cars.INST
…with five cars

(2) a. Ivan osti viisi auto-a
Ivan bought five-0 car-PART.SG
Ivan bought five cars (Brattico 2011: 1045)
b. Minä asuin kolmessa talossa
I lived three.INE.SG house.INE.SG
I lived in three houses (Brattico 2011: 1051)

Previous accounts have described the data in terms of case hierarchies, whereby inherent case outranks structural case, leading to the patterns above (cf. Babby 1987). However, such accounts suffer in the face of Finnish, which does not appear to respect the structural-inherent case distinction for case alternations (Brattico 2010, 2011).

In this talk, I discuss various case alternations, focusing specifically on numerals and negation in Polish and to a lesser degree, Finnish. I show that we can model these facts using a case stacking mechanism, which necessitates the use of a case hierarchy in terms of specific cases (nominative, accusative, genitive, etc), rather than case types (inherent, structural). I further show that certain cases appear to have a lexical requirement, leading to case percolation in the context of semi-lexical elements. Finally, I consider the possible underpinnings of the case hierarchy, and suggest that it actually reflects a structural difference between certain cases.

ESSL/LacqLab 10/15 - Amanda Swenson

Speaker: Amanda Swenson (MIT)
Time: Wednesday, October 15, 3:00-4:30pm (note exceptional time!)
Location: 32-D831
Title: The Morphosemantics of the Perfect in Malayalam

In this talk, I will examine the the semantics of the two constructions identified by Asher & Kumari (1997) as expressing the perfect in Malayalam. I consider whether or not the Right Boundary of the Perfect Time Span is set by tense, as it is in perfect constructions in Greek, English and Bulgarian (Iatridou et al. 2001). This question is particularly interesting in light of work by Amritavalli & Jayaseelan (2005) and Amritavalli (2014) which has argued that Malayalam does not have a TP and that temporal interpretation is read off of aspect. I will show, based on evidence from the construction that expresses the Existential perfect, that their system makes incorrect predictions for the perfect. I will provide a compositional analysis for the Malayalam Existential perfect. I also consider the other construction used for the Resultative and Universal perfects and show the ways in which it does and does not match the semantics of the parallel constructions in other languages.

Ling-Lunch 10/16 - Ayaka Sugawara

Speaker: Ayaka Sugawara (MIT) (joint work with Martin Hackl and Ken Wexler)
Title: On acquisition of “only”: Question-Answer congruence and scalar presuppositions
Date/Time:Thursday, October 16, 12:30-1:45pm
Location: 32-D461

There is a long-standing puzzle in acquisition of only since Crain et al. (1994): children up to age 6 display difficulties understanding sentences with pre-subject only (“subject-only”, e.g. Only the cat is holding a flag.) while having no difficulty understanding sentences with pre-VP only (“VP-only”, e.g. The cat is only holding a flag.). We note that neither “subject-only” nor “VP-only” are congruent with a broad question (e.g. What happened?), which is typically used to prompt puppet’s answers in experiments in the literature. Instead, they are congruent with different sub-questions, which we hypothesize that listeners must accommodate during comprehension. Our experiments compare children’s adult-like responses when we use broad questions and their responses when we use sub-questions. The results show that children are sensitive to Question-Answer Congruence (QAC) and support the idea that accommodation of sub-questions of What happened? plays a role in Crain’s puzzle.

LFRG 10/17 - David Nicolas

Speaker: David Nicolas (ENS)
Day/time: Friday, 17 October, 3:30pm
Location: 32-D461
Title: Two and a half apples

With some count nouns, we understand expressions of the form “a half N” and “half of an N” and sentences like this one:

(1) Two and a half apples are on the table.

This is true, for instance, if on the table there are two apples and one half apple (half of an apple).

If instead of “two and a half” we use a simple cardinal like “two”, the truth conditions of a similar sentence can be stated like this:

(2) Two apples are on the table
is true iff exists x (apple(x) & card(x) = 2 & on_the_table(x)) {“at least” semantics}

This “at least” semantics of cardinals just asserts the existence of two things. An “exact semantics” would assert the existence of exactly two things and no more.

Whether one adopts an “at least” semantics or an “exact” semantics, these kind of truth conditions are inadequate for (1) for two reasons (Salmon 1997, Liebesman 2014):

  • Half an apple is not in the denotation of “apples”, so it cannot be in the denotation of two and a half apples if one just “intersects” the meaning of “apples” with that of “two and a half”.
  • The function card() returns the cardinality of a plurality or set, which can never be a fractional number.

So what are the truth conditions of the sentence and how do we get them?

LFRG 10/8 - Brian Buccola

Speaker: Brian Buccola (McGill University)
Time: Wednesday, October 8, 3:30-5pm (note exceptional time!)
Location: 32-D461
Title: Global semantic constraints: the case of van Benthem’s problem

Any adjectival theory of numeral modifiers faces a challenge known as van Benthem’s problem (van Benthem, 1986), whereby non-upward-monotone quantifiers like “fewer than three” give rise to inadequate truth conditions. I propose a novel solution based on general economy principles for LF availability: certain LFs are generated by the grammar but unavailable (blocked) by virtue of (i) their semantic equivalence to LFs of syntactically simpler sentences, and (ii) the simultaneous availability of other, non-trivial LFs. The equivalence check is shown to rely crucially on the distributivity-collectivity properties of the predicates, in particular on whether the predicates distribute to at least some (not necessarily every) subpart (not necessarily atomic). The proposal therefore makes strong predictions regarding the interpretations of sentences with negative (and other) quantifiers and various predicates along the distributive-collective spectrum, which I show are borne out.

ESSL/LacqLab - No meeting this week

We will have no lab meeting this week. Our next lab meeting will be Wednesday, 10/15.

Ling-Lunch 10/9 - Ethan Poole

Speaker: Ethan Poole (UMass Amherst)
Title: Deconstructing quirky subjects
Date/Time:Thursday, October 09, 12:30-1:45pm
Location: 32-D461

Quirky (nonnominative) subjects differ across languages in whether they display the full range of properties exhibited by canonical nominative subjects. Based on data from Icelandic, German, Hindi-Urdu, Basque, and Laz, I show that the subjecthood properties exhibited by quirky subjects crosslinguistically obey an implicational hierarchy. I argue that this hierarchy is the result of DPs exhibiting a set of subjecthood properties as a function of how high they raise in the functional sequence.