@[youtube](https://youtu.be/rM1P6zrvO88) *Thank you for visiting our page. While we've uploaded the filmed presentation here along with a SRT and text file describing images in the video, you may also view the captioned video via YouTube for easier playback.* **Poster Presentation abstract:** Human languages allow for incredible creativity through re-use of familiar parts. This flexibility is difficult to capture in discrete documentation and coding systems. Here, we explore how the ASL Signbank (Hochgesang et al. 2020) represents relationships between words. Usage-based lexical databases have been created for several signed languages (e.g., Auslan, BSL, NGT) to support corpora. The ASL Signbank is an open-access database of over 3500 ASL signs used to facilitate annotation across data sets. Each entry contains a movie, image, ID gloss (Johnston 2001), and information like formational components, grammatical categories, and relationships to other signs. Corpora inform basic questions (which signs to include) to refinement of postulated linguistic features. Categories of entry information follow prior signbanks and annotation conventions (Chen Pichler et al. 2015). Categorization follows lemmatization principles (Fenlon et al. 2015) and field descriptions, like sign relations, in the Global Signbank (Crasborn et al. 2018). ![In a white box, black text is arranged in three rows/three columns. The top row begins with the text “Lemma ID Gloss” in a box on the left. Two black arrows move rightward from this box. Each arrow points to a separate box. Each box contains the text “Annotation ID Gloss.” To the right of these boxes is the text “Different phonological forms, but (morphological) instantiations of the same base form; Meaning strongly similar.” The middle row begins on the left with text “Annotation ID Gloss” in a box. A double-sided black arrow joins this box with a box in the middle column that also says “Annotation ID Gloss”. The word “homonym” is printed above this arrow. To the right of these boxes is the text “Exactly the same phonological form, sufficient semantic difference to warrant two glosses.” The bottom row also has two boxes, each with text “Annotation ID Gloss” and connected with a double-sided arrow. Above this arrow is the word “synonym”. To the right is the text “Different phonological forms, highly similar meaning, different lemmas”.][1] *Figure 1. Some types of relationships among signs in Global Signbank such as lemma, homonym, synonym (Crasborn et al. 2018).* Signs with dissimilar meanings and forms receive unique lemmas and ID glosses. ![This image shows a rectangular screenshot from ASL Signbank. In the top left is a dropdown menu with gray background and black text. To the right of the dropdown menu is an entry field with grayed out text “Annotation ID Gloss”. To the right of this field is a blue button with white text “Add Relation”. In the dropdown menu on the left, the top option is selected by a blue highlight and white checkmark. This option contains “- - - - - - - - - -“. The remaining seven options are: “Homonym”, “Synonym”, “Variant”, “Antonym”, “Hyponym”, “Hypernym”, and “See Also”.][2] *Figure 2. Sign Relations List in ASL Signbank (same as Global Signbank)* Signs with similar meanings and one or two phonological differences share a lemma, receive unique annotation ID glosses, and are marked as variants. When meanings overlap but forms are different (enough), signs are considered synonyms under separate lemmas. When only forms overlap, signs are assigned different lemmas and considered homonyms. ![This image shows boxes arranged in four rows and three columns. The leftmost column contains white boxes with one word in bold black text in each box. The middle and rightmost columns show images from ASL Signbank. In each image, a signer standing in from of a teal backdrop produces an ASL sign, and the sign’s Annotation ID Gloss is printed in bold, white, capital letters in the bottom left corner of the image. On the top and bottom rows, the ASL Signbank logo is visible in the top left corner of each image. In the top row, the box on the left contains the word “Variant”. The image in the middle column shows a Latina woman in a black t-shirt signing the ASL sign labeled as DOLPHINd. The image in the rightmost column shows a white woman in a gray t-shirt producing the ASL sign labeled DOLPHINr. These signs have similar form except for handshape and use of the nondominant hand in DOLPHINr. The second row from the top contains the word “Synonym” in the leftmost column. The image in the middle column shows a Latino man in a gray collared shirt producing the ASL sign labeled DON’T-FEEL-LIKE. The image to the right shows an Afghani woman in a long-sleeved black t-shirt producing the ASL sign labeled NOT-INTERESTEDfh. These signs have different forms. In the third row down, the box on the left contains the word “Homonym”. The image in the middle column shows a black woman in a black t-shirt signing the ASL sign labeled ACCEPT. The image in the righthand column shows a white man in a long-sleeved collared black shirt signing the ASL sign labeled OWN. These signs have the same form. In the bottom row, the box on the left contains the text “See Also”. The image in the middle column shows a white man wearing a gray collared shirt signing the ASL sign labeled GROUPg. The image in the righthand column shows a white woman in a gray t-shirt producing the ASL sign labelled FAMILY. These signs have the same form except for handshape.][3] *Figure 3. Some examples of sign relations in ASL Signbank* Some relationships are not reflected through available means. For example, groups of initialized signs (e.g., [GROUPg][4] and [FAMILY][5] are not variants, homonyms, etc. Our current system applies the catchall “see also” relation in these cases. Because this kind of categorization of ASL signs has never been done on a scale like this before - one that has been continually informed and refreshed by actual usage data - as our data set grows, our ability to recognize and define these kinds of relationships will improve. By [Julie A. Hochgesang][6], Amelia Becker, and Donovan Catt **Acknowledgments** - Thank you to the participants and families, as well as all of our research assistants. - We also appreciate Ryan Lepic’s input on our HDLS submission. - The research reported here was supported in part by the National Institute on Deafness and Other Communication Disorders of the National Institutes of Health under award number R01DC013578 and award number R01DC000183. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. **References** Chen Pichler, D., J.A. Hochgesang, D. Lillo-Martin. (2015, March). BiBiBi Project ASL Annotation Conventions. Poster presented at “Digging into Signs Workshop: Developing Annotation Standards for Signed Language Corpora”. University College London, London, United Kingdom (March 30-31, 2015). Crasborn, O., Zwitserlood, I., Van der kooij, E., & Schüller, A. (2018). Global SignBank manual (Version 1). Radboud University, Centre for Language Studies. https://doi.org/10.13140/RG.2.2.36603.82729 Hochgesang, J.A, O. Crasborn, & D. Lillo-Martin. (2017-2020) ASL Signbank. New Haven, CT: Haskins Lab, Yale University. https://aslsignbank.haskins.yale.edu/ Fenlon, J. K. Cormier, A. Schembri. (2015). Building BSL Signbank: The Lemma Dilemma Revisited. International Journal of Lexicography, 28(2), 169-206. Johnston, T. (2001). The lexical database of Auslan (Australian Sign Language). Sign Language & Linguistics, 4(1/2), 145-169. [1]: https://files.osf.io/v1/resources/4a29z/providers/osfstorage/5faadbce769b33000e157dca?mode=render [2]: https://files.osf.io/v1/resources/4a29z/providers/osfstorage/5faadb5b769b3300171587f4?mode=render [3]: https://files.osf.io/v1/resources/4a29z/providers/osfstorage/5faada8e6a1ea7000ef7fa06?mode=render [4]: https://aslsignbank.haskins.yale.edu/dictionary/gloss/643.html [5]: https://aslsignbank.haskins.yale.edu/dictionary/gloss/1379.html [6]: https://my.gallaudet.edu/julie-hochgesang