Somewhat late but my dissertation is now online! I look at how syntax and semantics relate to each other in Pintupi-Luritja, a Western Desert (Pama-Nyungan) language of Central Australia, particularly through negation, association with focus and modality

In AI, humans should not be in the loop (unless the goal is to have invisible workers or scape goats). Machines should be in the loop!

Thanks Steven Bird for your talk at C3NLP last year.

Link to the complete talk here
youtu.be/iQ5aMQNowAU

Quite excited for this shared task and to see the results of how well, or if even can, models generalize to ~211.2 paradigms in Turkish inflectional morphology. If you are also curious send your submissions by May 31st!

🔥We are excited to announce SIGMORPHON SHARED TASK: DATA-EFFICIENT MORPHOLOGICAL INFLECTION AND SIMULATION OF LINGUISTIC FIELDWORK!
💫What is the minimum number of samples to get 100% accuracy?
💫What is the best strategy to sample selection?

👉 github.com/sigmorphon/202… #NLProc

If your language technology work concerns written language processing for a language endowed with a standardised orthography, please make this explicit, e.g. 'written language processing'. Let's remember that 90% are 'oral languages', or 'languages with primary orality' #LT4All

🧵 LLMs are most useful if they can serve users in their language. We highlight performance in three widely spoken non-Latin script languages:
- 🇰🇷 Korean
- 🇯🇵 Japanese
- 🇨🇳 Chinese

Command R+ outperforms Claude 3 Sonnet and Mistral Large and is competitive with GPT 4-Turbo on…

We invite you to participate in @SIGMORPHON’s 9th instalment of its inflection generation shared task, which will be on Data-efficient Morphological Inflection and Simulation of Linguistic Fieldwork.
More details: github.com/sigmorphon/202…

We're proud to announce that CreoleVal has been accepted to TACL! Huge kudos to Heather Lent for her absolutely stellar work on this project, and to the team of co-authors #NLProc

It's claimed perceptual language reflects a biologically grounded hierarchy of the senses. Using a stratified global sample of 100 languages, a new study led by Elisabeth Norcliffe finds domain-general constraints explain lexicalization patterns instead. muse.jhu.edu/pub/24/article…

💜THANK YOU!💜
Huge thanks to Oksana Dereza Oksana 🐈‍⬛ @EACL 2024 , Frederick Riemenschneider, An Tuan Dao, Ercong Nie, Jessica Nieder for being terrific hosts and Michael Hahn Michael Hahn and Priya Rani for leading the organising team and working on the workshop schedule and proceedings!

While I couldn’t join #EACL2024 , Heidelberg University NLP Group was represented, i.a. by Frederick Riemenschneider Frederick Riemenschneider, who won the SIGTYP Shared Task, jointly with Kevin Krahn. Their character-aware hierarchical LM perfectly addresses low-resource Historical Languages. Well done, Frederick 🥳 👏

💜THANK YOU!💜
Huge thanks to Oksana Dereza Oksana 🐈‍⬛ @EACL 2024 , Frederick Riemenschneider, An Tuan Dao, Ercong Nie, Jessica Nieder for being terrific hosts and Michael Hahn Michael Hahn and Priya Rani for leading the organising team and working on the workshop schedule and proceedings!

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos

📸 #SIGTYP2024 Photos