Reminder: Floris Roelofsen (Universiteit van Amsterdam) will be giving a talk for the Construction of Meaning Workshop today in the Greenberg Room on work in Montague Grammar.
Alternatives in Montague grammar
The type theoretic framework for natural language semantics laid out by Montague (1973) forms the cornerstone of formal semantics. Hamblin (1973) proposed an extension of Montague’s basic framework, referred to as alternative semantics. In this framework, the meaning of a sentence is not taken to be a single proposition, but rather a set of propositions—a set of alternatives. While this more fine-grained view on meaning has led to improved analyses of a wide range of linguistic phenomena, it also faces a number of problems. We focus here on two of these, in our view the most fundamental ones.
The first has to do with how meanings are composed, i.e., with the type-theoretic operations of function application and abstrac- tion; the second has to do with how meanings are compared,i.e., the notion of entailment. Our aim is to reconcile what we take to be the essence of Hamblin’s proposal with the solid type-theoretic foundations of Montague grammar, in such a way that the observed problems evaporate. Our proposal partly builds on insights from recent work on inquisitive semantics (Ciardelli et al., 2013), and it also further advances this line of work, specifying how the inquisitive meaning of a sentence, as well as the set of alternatives that it introduces, may be built up compositionally.
This talk presents joint work with Ivano Ciardelli.
Join the P-Interest Workshop as they welcome Jevon Heath (Berkeley), who will present on recent work in phonetic accommodation. The talk will be in the Greenberg Room.
How do we measure phonetic accommodation?
In phonetic accommodation, talkers change the way they talk due to their interlocutors’ speech. This is commonly interpreted as imitation of features found in the received speech signal, and a growing body of literature relies on measuring the degree of imitation talkers evince in order to draw conclusions about attitudes toward their interlocutors. However, the best way of carrying out this measurement is unclear. Some studies have relied on quantitative measurements of particular phonetic features (Shockley et al. 2004, Babel 2010, inter alia); other studies use qualitative judgments from a third party (the AXB paradigm (Goldinger 1998, Pardo et al. 2012, inter alia). I present data from two studies indicating issues with both quantitative and qualitative methods of measuring phonetic accommodation. In the first study, I find that participants converging towards a model talker in one dimension simultaneously diverge in a related dimension, indicating the difficulty of isolating a particular feature or set of features as the locus of imitation. In the second study, listeners participated in an AXB study in which the two model speakers had not interacted. I find that listeners exposed to two recordings of the same speaker at different times report that speaker’s later iteration as sounding more similar to a second speaker, even in a context in which no accommodation is possible. I conclude with suggestions for mitigating these issues in accommodation studies going forward.
The CogLanguage Workshop would like to welcome you to a talk by Gary Lupyan (Wisconsin-Madison) on his work on the realization of language and thought in the mind. All are welcome to this exciting talk, which will take place at 4PM in the Greenberg Room!
How Language Programs the Mind
In contemporary psychology, it is common to view the mind as being split. On one side are processes such as perception, attention, and memory. On the other are linguistic processes such as word and sentence comprehension which translate nonverbal representations into verbal ones. I will argue that this split is untenable. Rather than being a medium into which “thoughts” are translated for communication, language acts as a high level control system for the mind, changing how perceptual and conceptual representations are activated. I will support this position by showing how even subtle linguistic manipulations affect behavior from low-level visual tasks to higher-level categorization and inference. I will then discuss several design features of language that make it an especially powerful tool for programming the mind.
Join the Symbolic Systems Forum next Monday 12:15 in the Greenberg Room, where Thomas Icard (Stanford Philosophy) will speak about his his interdisciplinary work in Philosophy and Cognitive Science.
Familiar formal accounts of rational thought and action — from classical logic, probability, decision theory, etc. — do not take into account computational bounds, and in particular they leave out the costs an agent might incur in figuring out what to think or do. While this fact has long been recognized (famously by H. Simon, but also by the pioneers of these methods, such as L.J. Savage and I.J. Good, and many since), a substantive formal theory of what we might call “algorithmic rationality” has not been forthcoming. I will try to illustrate why such a theory is needed in philosophy and in cognitive science, and offer some speculative remarks about how it might look.
In case you missed it, corpus work from across the department was showcased during this week’s Corpus and VoC Lunch, which offered 5-minute presentations for members of the department to discuss their corpus-related projects. The presentations are listed below:
- Simon Todd – “That’s how kiwis speak, eh: a corpus study of “eh” in New Zealand English”
- Robin Melnick – “Development and application of an Individual Differences corpus”
- Robin Melnick (on behalf of Tom Wasow) – “Optional to after help in COCA”
- Katherine Hilton & Bonnie Krejci – “Agreement Variation Under Existential ‘There’”
- Sam Bowman – “Data for natural language inference”
- Kate Lindsey – “Something’s Afoot in Chuvash”
- Lelia Glass – “Using corpora to test socio-pragmatic predictions: The case of need to, have to and got to”
- Natalia Silveira – “The Universal Dependencies project: syntactic dependencies for all”
- Sara Kessler – “The Stage/Individual Level Distinction as a Predictor of Absoluteness in Gradable Adjectives”
- Daniel Galbraith – “Faroese ballad meter”
- Rob Voigt – “Multimodal Corpora for Understanding Multimodal Prosody”
- Gabe Doyle – “Building ad hoc social media corpora to research pragmatics and dialect geography”
- Janneke Van Hofwegen – “/s/ and gender in Redding, CA”
Wow! Keep up the good work, everyone.
Joke courtesy of fastcompany.com