From oh.531 at buckeyemail.osu.edu Thu Jan 12 12:50:20 2023 From: oh.531 at buckeyemail.osu.edu (Oh, Byung-Doh) Date: Thu, 12 Jan 2023 17:50:20 +0000 Subject: [CaCL] Zoom link for CaCL meetings Message-ID: is this one: https://osu.zoom.us/j/92759547111?pwd=eS90SXVsaTR6VkNyTWdkZm9Dd0NjUT09 ================= Byung-Doh Oh (he/him/his) Ph.D. Student Department of Linguistics The Ohio State University -------------- next part -------------- An HTML attachment was scrubbed... URL: From schuler.77 at osu.edu Fri Jan 13 23:03:32 2023 From: schuler.77 at osu.edu (Schuler, William) Date: Sat, 14 Jan 2023 04:03:32 +0000 Subject: [CaCL] Fwd: CogSci / computational linguistics Google student researcher position In-Reply-To: References:

Message-ID: Of possible interest, wm ________________________________ From: Tal Linzen Sent: Thursday, January 5, 2023 11:38 AM Subject: Re: CogSci / computational linguistics Google student researcher position Hi all, Apologies for multiple emails: it turns out that more advanced PhD students are in fact eligible for this position. They just need to understand that, unlike a research internship, this engagement can't result in a possible "conversion" ZjQcmQRYFpfptBannerStart This Message Is From an External Sender This message came from outside your organization. Report Suspicious ZjQcmQRYFpfptBannerEnd Hi all, Apologies for multiple emails: it turns out that more advanced PhD students are in fact eligible for this position. They just need to understand that, unlike a research internship, this engagement can't result in a possible "conversion" to full-time employment at Google. Thanks again, Tal On Mon, Jan 2, 2023 at 3:56 PM Tal Linzen > wrote: Dear colleagues, I'm currently in a position to recruit one or more part-time student researchers to work with me and others at Google on project(s) in the computational cognitive science / computational linguistics area. Examples of relevant topics include compositionality and logical reasoning in large language models, and the contribution of grounding outside of language to language learning. I'm reaching out to you because the ideal person for this position would have both strong engineering skills and an interest in cognitive science and linguistics - so basically one of your students! Please let me know if there's anyone you can think of that I should get in touch with. Your students can also reach out to me informally, and if they're a good fit, I'll encourage them to apply officially. Some context about the Google student researcher program: this program is currently open to students with a graduation date after December 2024. It's a part-time position for a duration of a few months, and the time commitment is relatively flexible; for example, the student could spend more time working on the project over the summer and less during the semester. The student needs to be enrolled in a US PhD program and located in the US, but other than that they can be fully remote. You can find more details about the position here. Thanks! Tal -------------- next part -------------- An HTML attachment was scrubbed... URL: From schuler.77 at osu.edu Wed Jan 18 13:09:40 2023 From: schuler.77 at osu.edu (Schuler, William) Date: Wed, 18 Jan 2023 18:09:40 +0000 Subject: [CaCL] cacl cancelled this week Message-ID: Hi all, Since Byung-Doh is still working on his draft, and since there is a talk in CSE I believe (check clippers), let?s cancel CACL this week, wm -------------- next part -------------- An HTML attachment was scrubbed... URL: From clark.3664 at buckeyemail.osu.edu Sat Jan 21 18:57:02 2023 From: clark.3664 at buckeyemail.osu.edu (Clark, Christian) Date: Sat, 21 Jan 2023 23:57:02 +0000 Subject: [CaCL] Reading for 1/26 Message-ID: Hi CaCL members, For our meeting on Thursday, we will read "Unsupervised Discontinuous Constituency Parsing with Mildly Context-Sensitive Grammars" by Yang et al. (2022). Link: https://arxiv.org/pdf/2212.09140.pdf Abstract: We study grammar induction with mildly context-sensitive grammars for unsupervised discontinuous parsing. Using the probabilistic linear context-free rewriting system (LCFRS) formalism, our approach fixes the rule structure in advance and focuses on parameter learning with maximum likelihood. To reduce the computational complexity of both parsing and parameter estimation, we restrict the grammar formalism to LCFRS-2 (i.e., binary LCFRS with fan-out two) and further discard rules that require O(n6) time to parse, reducing inference to O(n5). We find that using a large number of nonterminals is beneficial and thus make use of tensor decomposition-based rank-space dynamic programming with an embedding-based parameterization of rule probabilities to scale up the number of nonterminals. Experiments on German and Dutch show that our approach is able to induce linguistically meaningful trees with continuous and discontinuous structures. ---- Christian Clark Ph.D. Student Department of Linguistics The Ohio State University -------------- next part -------------- An HTML attachment was scrubbed... URL: From oh.531 at buckeyemail.osu.edu Thu Jan 26 14:09:55 2023 From: oh.531 at buckeyemail.osu.edu (Oh, Byung-Doh) Date: Thu, 26 Jan 2023 19:09:55 +0000 Subject: [CaCL] CaCL 2/2: Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations Message-ID: Hello everyone, Next week we'll be discussing the following TACL paper: https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00504/113019/Structural-Persistence-in-Language-Models-Priming Structural Persistence in Language Models: Priming as a Window into Abstract Language Representations We investigate the extent to which modern neural language models are susceptible to structural priming, the phenomenon whereby the structure of a sentence makes the same structure more probable in a follow-up sentence. We explore how priming can be used to study the potential of these models to learn abstract structural information, which is a prerequisite for good performance on tasks that require natural language understanding skills. We introduce a novel metric and release Prime-LM, a large corpus where we control for various linguistic factors that interact with priming strength. We find that Transformer models indeed show evidence of structural priming, but also that the generalizations they learned are to some extent modulated by semantic information. Our experiments also show that the representations acquired by the models may not only encode abstract sequential structure but involve certain level of hierarchical syntactic information. More generally, our study shows that the priming paradigm is a useful, additional tool for gaining insights into the capacities of language models and opens the door to future priming-based investigations that probe the model?s internal states. Best, Byung-Doh ================= Byung-Doh Oh (he/him/his) Ph.D. Student Department of Linguistics The Ohio State University -------------- next part -------------- An HTML attachment was scrubbed... URL: From schuler.77 at osu.edu Tue Jan 31 09:14:05 2023 From: schuler.77 at osu.edu (Schuler, William) Date: Tue, 31 Jan 2023 14:14:05 +0000 Subject: [CaCL] Fwd: Undergrad presentation opportunity In-Reply-To: References: Message-ID: Of possible interest to undergraduates on this list: ________________________________ From: Hiersche, Kelly Sent: Thursday, January 26, 2023 12:41:42 PM To: Schuler, William Subject: Undergrad presentation opportunity Hello Dr. Schuler, I just spoke with you in class about a poster presentation opportunity for undergraduate students. Feel free to disseminate the information below to any students you feel may be interested. Thank you! Kelly The Center for Cognitive and Brain Sciences is looking for undergraduate student volunteers to present at a small research conference on March 4, 2023 hosted in the Thompson Library. This conference will be attended by local high school students as part of the Humanities & Cognitive Sciences High School Summer Institute (https://cog.osu.edu/programs/humcog) The goal of this program is to expose students to undergraduate research as they look to pursue similar interests in college. Presenters should come with the goal of participating in outreach ? not with the goal of getting expert feedback on their current research projects. We are looking to recruit 6-8 presenters from various disciplines in the humanities and cognitive sciences. Interested students should send their name, major/department, research advisor, and abstract of their project to ccbs at osu.edu by February 9th, 2023. Students should have a research poster created and experience presenting to an audience ? CCBS will cover the cost of printing new posters if necessary. All presenters will receive a gift card for their participation, as well as food and fun the day of the event! With your help, we?ll inspire the next generation of researchers at OSU! Please send all questions to ccbs at osu.edu. -------------- next part -------------- An HTML attachment was scrubbed... URL: