# Week of Oct 7 October 07, 2024 ## **1100 Mass Ave Open House** An open house was held from 1:30-3pm on Thursday Oct 3 at the new departmental space for Linguistics at 1100 Mass Ave, co-hosted by the M&M Lab and the ASL Program. It was a great opportunity to learn more about Linguistics, learn some ASL signs, and explore one's own interests. Many also enjoyed the fun of pumpkin decorating! Thank you to everyone who came over!! ![1100massoh1.jpg](/sites/g/files/omnuum5001/files/styles/hwp_1_1__360x360_scale/public/linguistics/files/1100massoh1.jpg?itok=H1m-118h) ![1100massoh3.jpg](/sites/g/files/omnuum5001/files/styles/hwp_1_1__360x360_scale/public/linguistics/files/1100massoh3.jpg?itok=WwhlajTo) ![1100massoh5.jpg](/sites/g/files/omnuum5001/files/styles/hwp_1_1__360x360_scale/public/linguistics/files/1100massoh5.jpg?itok=DO4npRJ7) ![1100massoh6.jpg](/sites/g/files/omnuum5001/files/styles/hwp_1_1__360x360_scale/public/linguistics/files/1100massoh6.jpg?itok=6tEm-pBD) ![1100massoh4.jpg](/sites/g/files/omnuum5001/files/styles/hwp_1_1__360x360_scale/public/linguistics/files/1100massoh4.jpg?itok=lyQbA4HU) ## **Harvard Linguistics Colloquium** The second talk in the Harvard Linguistics Colloquium Series this semester is going to take place on **Friday Oct 11** at **12-1:30pm**. The speaker will be Peter Svenonius from University of Tromsø – The Arctic University of Norway. Details are as follows: **Title**: Cyclic spell-out and biclausal words **Location**: The Fong Auditorium, Boylston Hall (1st floor) **Time**: Friday Oct 11 @ 12-1:30pm **Abstract**: Causative constructions show varying degrees of biclausality, as has been discussed since the 1970s, ranging from “Floyd broke the glass” to “Floyd made it so that the glass broke”. One faultline runs between inner and outer causatives (e.g., North Sámi *buori-d-it* “good-cause-inf” ‘improve’ vs. *buori-d-ahtt-it* “good-cause-outer.cause-inf” ‘cause to improve’), closely related to the distinction between direct and indirect causation. Simply put, inner causatives show little sign of biclausality, while outer causatives pass more tests which point in the direction of biclausality. Since Marantz (1997, 2001), the head introducing the external argument---usually called Voice these days---has been recognized as a significant boundary in variations on the theory of cyclic word spell-out. Outer causatives provide clear morphophonological evidence of this boundary, while inner causatives do not consistently do so. The pattern can be explained if outer causatives involve recursion of Voice (possibly with one or more layers of structure between the two Voice heads) and if recursion of Voice forces cyclic word spell-out of material below Voice. The contrast with inner causatives suggests, however, that cyclic word spell-out is not forced by non-recursive Voice. I discuss the implications for the word and the clause. ## **LangCog** The next LangCog meeting of the semester will be on **Tuesday, October 8th** from **5:30-7:00pm.** It will take place in **William James Hall, Room #1550**. Our speaker is [Hayley Ross](https://urldefense.proofpoint.com/v2/url?u=https-3A__rossh2.github.io&d=DwMFaQ&c=WO-RGvefibhHBZq3fL85hQ&r=LEs7ofJHpoxoL00XPDAAYqergP6rofR-63Zf7h0PIE4&m=SrkLvJzDBVPi2D1-zeaUdtNL1yxdcul0N0m-VY6OiFd_ZqbL3NkNtCHZDXABuy80&s=N6bVbiwzE_egfSbSe9EBc9nUo4E6NlSnUX6xMX6N7ak&e=) (Harvard), and the title and abstract of her talk can be found below. Food will be available at the meeting, and you can find the schedule for the remainder of the semester on our [website](https://sites.harvard.edu/langcog/). **Speaker**: Hayley Ross **Title**: When is artificial intelligence still intelligence? Measuring and modelling adjective-noun inferences **Abstract**: Unlike typical adjectives like yellow, adjectives like fake or counterfeit can yield what is called a privative inference: while e.g. a yellow floweris always a flower, a fake gun is precisely not a gun and a counterfeit dollar is not a dollar. Moreover, recent work shows privativity cannot easily be encoded as a property of specific adjectives like counterfeit, since e.g. a counterfeit watch is still a watch (Martin, 2022). We gather judgments on nearly 800 adjective-noun bigrams (of which 180 are novel, i.e. zero corpus frequency), asking humans to rate "Is an {adjective} {noun} still a {noun}?" We show that privativity depends on the adjective, noun and context, and can be manipulated for the very same adjective-noun bigram by presenting it in different contexts. Moreover, we find no difference in participant behavior between novel adjective-noun bigrams and high frequency ones, suggesting that the process is nonetheless compositional and not the result of convention or memorized idiosyncrasy. In the second part of this project, we explore how humans draw these varied and context-dependent inferences. One likely source of within-bigram variation in human ratings when the bigram is presented out of the blue is that different humans are likely imagining different instances of the adjective-noun bigram, loosely corresponding to different contexts. We attempt to model this component of the variation by having an LLM generate a range of contexts and then rate "Is an {adjective} {noun} still a {noun}?" given that context, having first established that LLMs are sensitive to context and can draw the expected inferences given the contexts supplied to humans in the previous experiment. We find that the distributions of ratings generated in this way correspond well to human distributions for 61% of our nearly 800 bigrams, which is promising but also leaves room for improvement in future experiments. Share on:- [ Facebook ](#) - [ Twitter ](#) - [ Linkedin ](#)