OjibweMorph: A finite-state transducer for Ojibwe (2022–Present)

The goal of this project is to build a computational model of all possible word forms in Ojibwe, with a particular focus on the Southwestern dialects. Our model is based in a technology called finite-state transducers (FSTs), which uses rules to generate precise outputs. The end result is a program that can provide a sophisticated linguistic analysis for almost any word of Ojibwe, and also produce correct forms of a word given a set of inputs.

Anyone can access and use the software for non-commercial purposes, and we are eager to hear about ways you might find it useful in your own work, or ways you might contribute to its continued development.

We are currently using the model to:

Automate linguistic annotation for our under-development Wiigwaas corpus (see below)
Improve search results for morphologically complex words in the Ojibwe People’s Dictionary
Serve as the engine of a verb conjugation reference tool for learners of the language
Provide the foundation for a basic machine-translation model for Ojibwe
Support the first ever spell-checker for Ojibwe

Try it out! Here are some ways to access OjibweMorph:

Use the online text analysis tool made in collaboration with Dustin Bowers
Use the web-based version of the spellchecker powered by our partner organization Divvun
Install the spellchecker on your machine with the Divvun Manager
Go to the OjibweMorph repository or check out the latest release
Go to the lang-ciw repository from GiellaLT for the version compiled into the spellchecker

Relevant Outputs

Hammerly, C., Livesay, N., Arppe, A., Stacey, A., & Silfverberg, M. (2026). OjibweMorph: An approachable finite state transducer for Ojibwe (and beyond). Language Resources and Evaluation, 60(2), 1–44.
Nguyen, M., Hammerly, C., & Silfverberg, M. (2025). A hybrid approach to low-resource machine translation for Ojibwe verbs. In Proceedings of the 5th Workshop on Natural Language Processing for Indigenous Languages of the Americas (AmericasNLP 2025), 18–26.
Hammerly, C., Parkhill, S., Goodyear, L., Silfverberg, S., Quinn, C., & Parkhill, M. (2025). An Automatic Verb Conjugation Tool for Ojibwe. Talk at the 9th International Conference on Language Documentation and Conservation (ICLDC9), University of Hawai’i at Mānoa.

This work was supported by funding from the Social Sciences and Humanities Research Council (SSHRC) of Canada through the Insight Grant (435-2023-0474) and Partnership Grant (895-2019-1012) programs.

Processing grammatical voice in Ojibwe (2022–Present)

The core question of this line of work is: How do Ojibwe speakers make use of information related to obviation, animacy, and voice to put together argument structure relationships in real time?

UBC Linguistics PhD student Marcella Jurotich has been looking at how information structure factors such as givenness and focus impact the comprehension of the non-canonical “inverse” voice in L1 speakers, as recently presented in a poster at the 2026 HSP conference.

Building on some early work by Hammerly et al. (2022), we have also been looking at how new (L2) speakers of Ojibwe make sense of obviation and the direct/inverse alternation through a visual world eye tracking study. This study includes instructional videos created by our collaborator Rochelle Allan at the University of Toronto with additional guidance from John-Paul Chalykoff at Algoma University and support from former RA Hope Trischuk.

Relevant Outputs

Jurotich, M. & Hammerly, C. (2026). Impact of Information Structure on Comprehension of the Inverse Verb Form in Ojibwe. Poster at the 39th Annual Conference on Human Sentence Processing (HSP), MIT, Cambridge, MA.
Hammerly, C., Staub, A., & Dillon, B. (2022). Person-based prominence guides incremental processing: Evidence from obviation in Ojibwe. Cognition, 225, 1–13.

This work was supported by funding from the Social Sciences and Humanities Research Council (SSHRC) of Canada through the Insight Grant (435-2023-0474).

Wiigwaas and Wiigwaasaatig: A corpus and treebank of contemporary Ojibwe (2024–Present)

This project has two closely related goals:

To build a fully annotated text-based corpus of contemporary Ojibwe (Wiigwaas, “birchbark”)
To build a treebank of syntactic dependencies in Ojibwe (Wiigwaasaatig, “birch tree”)

We expect the initial release of these tools to total around 250,000 words, sourced from published contemporary (late 20th and 21st century) texts, many of which have parallel English translations.

We are using the OjibweMorph model as an automatic annotation tool. Our recent set of constraint grammars (Diederichsen & Hammerly 2026) is used to select analyses when a word has multiple possible tags and to automatically parse syntactic dependencies for the treebank.

Corpora are the foundation for many digital tools and a resource for researchers working on the Ojibwe language, but can also be used as pedagogical tools for finding naturally occurring example sentences.

We are aiming to use the corpus in our own work to understand patterns of word order and argument realization in the language, especially as related to the systems of obviation, animacy, and voice.

Stay tuned for more soon!

Relevant Outputs

Diederichsen, M. & Hammerly, C. (2026). Two Ojibwe Constraint Grammars: Morphological Disambiguation and Dependency Parsing. In Proceedings of the Fifteenth International Conference on Language Resources and Evaluation (LREC 2026), 9756–9766.

This work was supported by funding from the Social Sciences and Humanities Research Council (SSHRC) of Canada through the Insight Grant (435-2023-0474) and Partnership Grant (895-2019-1012) programs.

Completed Projects

VisualWorldTools: Code for creating visual world studies with Tobii eye trackers (2022–2024)

The goal of this project was to create a general set of open-source tools for running visual world studies in fieldwork settings using Tobii eye trackers and touch screens (with support for other input sources as well).

The code is written in Python and can be run in PsychoPy. The modular design of the code allows for a wide range of different visual world designs.

You can access the toolset at our VisualWorldTools repository.

This project was funded by the Canada Foundation for Innovation (CFI), John R. Evans Leaders Fund (JELF).

Speech synthesis for Ojibwe, Mi’kmaq, and Maliseet (2021–2025)

The goal of this project was to create and evaluate speech synthesis systems, especially Text-To-Speech systems, to support app-based language learning for three Algonquian languages (Ojibwe, Mi’kmaq, and Maliseet).

This project included not only the technical development of these systems (Hammerly et al. 2023; Wang et al. 2025), but also a set of participatory workshops with Ojibwe language teachers aimed at understanding how instructors and students can use these technologies as part of their language learning journey (Chan & Hammerly 2025).

Relevant Outputs

Wang, S., Yang, C., Parkhill, M., Quinn, C., Hammerly, C., and Zhu, J. (2025). Developing multilingual speech synthesis system for Ojibwe, Mi’kmaq, and Maliseet. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics (NAACL), 817–826.
Chan, V. & Hammerly, C. (2025). Evaluating Indigenous language speech synthesis for education: A participatory design workshop on Ojibwe text-to-speech. In Proceedings of the 8th Workshop on the Use of Computational Methods in the Study of Endangered Languages (ComputEL-8), 47–64.
Hammerly, C., Frazier, S., Sierra, G., Parkhill, S., Porteous, H., & Quinn, C. (2023). A text-to-speech synthesis system for Border Lakes Ojibwe. In Proceedings of the 6th Workshop on the Use of Computational Methods in the Study of Endangered Languages (ComputEL-6), 60–65.

This project was funded by an industry-sponsored grant from LanguageFoundry (formerly CultureFoundry) as well as a Mitacs Accelerate grant.