Whamit!

The Weekly Newsletter of MIT Linguistics

Issue of Monday, March 17th, 2025

Phonology Circle 3/17 - Runqi Tan (MIT)

Speaker: Runqi Tan (MIT)
Title: The role of perceptual contrast in tone inventory
Time: Monday, March 17th, 5pm - 6:30pm
Location: 32-D831

Abstract:

Theoretically, there are 2366 possible types of two- to five-tone inventories. However, a survey of 63 two- to five-tone languages reveals only 12 attested tone inventories. This discrepancy suggests that tone inventories are not formed through random selection of tones but are shaped by universal constraints. Yip (2001), based on a large-scale survey of tone languages, proposed universal rankings of markedness constraints: *RISE ≫ *FALL and *HIGH ≫ *LOW. Similarly, Chen (2000) proposed a markedness hierarchy: *RISE ≫ *HIGH ≫ *FALL ≫ *LOW. However, the markedness hierarchy alone does not account for the observed tone inventories.

This study examines the role of perceptual contrast in shaping tone inventories within the framework of Dispersion Theory. By adding a set of MINDIST constraints, which evaluate the perceptual distinctiveness of tonal contrasts, to a set of markedness constraints that evaluate the articulatory efforts of tones, the new model successfully predicted 8 of the 12 attested tone inventories while avoiding over-generation. In addition, the model captures the following generalizations: (1) the mid-level tone does not co-occur with small contour tones and only co-occurs with large contour tones when both L and H are present in the inventory; (2) a tone system that only relies on pitch and slope for tonal contrasts can have no more than three contrastive levels per dimension; (3) it is possible for a language to have only contour tones without level tones. These findings suggest that perceptual contrast plays a crucial role in shaping tone inventories.

LingLunch 3/20 - Nina Haslinger (ZAS Berlin, MIT)

Speaker: Nina Haslinger (ZAS Berlin, MIT)
Title: Pragmatic constraints on morphosyntactic organization: A case study on homogeneity and imprecision
Time: Thursday, March 20th, 12:30pm - 2pm
Location: 32-D461

Abstract:

Across unrelated languages, there is a containment asymmetry between definite plurals and all-type plural universal quantifiers (UQs) — for instance, the surface form all the books contains the plural definite the books, while the reverse asymmetric pattern—a definite-plural structure properly containing a UQ structure—is (to my knowledge) unattested. This cross-linguistic asymmetry led Matthewson (2001), Winter (2001) and others to hypothesize that UQs are associated with “bigger” structures than definite plurals. At first sight, we might try to capture this at the descriptive level by cartographic means, or alternatively by constraining the syntactic category/semantic type correspondence such that nominal quantifiers must take a type e argument (cf. Matthewson 2001).

This talk makes two main points. The first point is to argue that both cartographic and lexical-semantic approaches miss a broader generalization of which the definite/UQ asymmetry is a special case. Recent work in plural semantics (Malamud 2012, Križ 2015, Križ & Spector 2021, Bar-Lev 2020, Feinmann 2020, Guerrini & Wehbe to appear a.o.) has focused on imprecision, a form of semantic underspecification driven by implicit QUDs (cf. Lasersohn’s (1999) “pragmatic slack”), and on truth-value gap phenomena that systematically correlate with imprecision (“homogeneity effects“). I will argue that across seemingly unrelated syntactic categories and semantic types, imprecise expressions correspond to less complex structures than their precise counterparts. Thus, the phenomenon is not specific to the extended NP or to quantifiers with type e arguments.

Given the generalization that imprecision correlates with smaller structures, my second point is to suggest an account of this generalization in pragmatic terms. In the special case of contexts in which the precise and imprecise alternatives make the same truth-conditional contribution, this generalization can be viewed as the result of a trade-off between pragmatic maxims falling under the Gricean category of Manner — one expressing a preference for simpler structures (“Be brief!”) and one expressing a preference for expressions that do not depend on the QUD for their truth conditions (“Be precise!”). I formalize this idea using Katzir’s (2007) notion of structural complexity and Križ & Spector’s (2021) semantics for plural imprecision. Time permitting, I will also try to provide independent support for the proposed constraint interaction, by looking at exceptional cases in which one of the two pragmatic maxims is trivialized.

On the face of it, this kind of Manner-based neo-Gricean reasoning falls short of fully deriving the generalization, since there are many contexts in which imprecise expressions and their precise counterparts are *not* truth-conditionally equivalent, and therefore arguably not expected to be competitors for the purposes of Manner. I propose that, instead of giving up on the pragmatic approach, we should bite the bullet and consider a competition mechanism that applies as soon as two expressions become equivalent under *some values* of contextual parameters such as the QUD, even if these are not the actual values in the utterance context at hand. This would amount to a theory in which Manner-based pragmatic reasoning can sometimes apply “automatically” even if its Gricean preconditions are not met, in the same loose sense in which Quantity-based reasoning applies “automatically” on grammatical theories of implicature.

Miyagawa’s publication on early human language featured in MIT News

Our faculty colleague Shigeru Miyagawa recently published an article titled “Linguistic capacity was present in the Homo sapiens population 135 thousand years ago” in Frontiers in Psychology on March 11. You can read the abstract below and access the article here:

Recent genome-level studies on the divergence of early Homo sapiens, based on single nucleotide polymorphisms, suggest that the initial population division within H. sapiens from the original stem occurred approximately 135 thousand years ago. Given that this and all subsequent divisions led to populations with full linguistic capacity, it is reasonable to assume that the potential for language must have been present at the latest by around 135 thousand years ago, before the first division occurred. Had linguistic capacity developed later, we would expect to find some modern human populations without language, or with some fundamentally different mode of communication. Neither is the case. While current evidence does not tell us exactly when language itself appeared, the genomic studies do allow a fairly accurate estimate of the time by which linguistic capacity must have been present in the modern human lineage. Based on the lower boundary of 135 thousand years ago for language, we propose that language may have triggered the widespread appearance of modern human behavior approximately 100 thousand years ago.

MIT News featured an article on this publication on March 14, which you can read here.

 

Jou @ Nanzan workshop

Postdoctoral associate and alum Eunsun Jou (PhD 2024) gave an invited talk entitled “Case-marked adverbials and the timing of case evaluation” at the Comparative Syntax, Semantics and Language Acquisition Workshop held on March 8-9 at the Nanzan University Center for Linguistics. A summary of the talk is provided below:

While structural cases such as the nominative and accusative typically appear on arguments, some languages including Russian, Finnish, and Korean show structural case marking on certain adverbials. To explain this phenomenon, some researchers have proposed a distinct case system for adverbials that is separate from the case system of arguments. Focusing on Korean, I argue in my talk that we can avoid this redundancy. With the right configurations, we can explain case on arguments and case on adverbials under the same case mechanism. The crucial observation in achieving this goal is the discovery that case marking on Korean adverbials correlates with something unexpected: the position of the grammatical subject. After providing arguments supporting this correlation from three phenomena (predicate fronting, intervention effects, and the relative scope of the subject and the event), I present a Dependent Case model which derives the correlation without reference to the argument/adjunct status of nominals. Lastly, I entertain two theories about the timing of case evaluation: an Earliness-based theory where case assignment is interspersed with structure-building operations, and a phase-based theory where case assignment happens at the timing of phasal spell-out. I discuss how the latter is better situated to explain the facts discussed in the talk.