Corpora, Collections, and Datasets

There has been some discussion of what constitutes a collection among language documentation practitioners. Johnson (Citation: Johnson, 2004, p. 142) Johnson, H. (2004). Language documentation and archiving, or how to build a better corpus. Language Documentation and Description, 2. 140–153. Retrieved from http://www.elpublishing.org/PID/026 equates a collection and a corpus saying:

“A collection, or corpus, is the body of documentary materials created by linguists and native speakers in the course of their research.”

Simons (Citation: Simons, 2008) Simons, G. (2008). Documentary Linguistics and a New Kind of Corpus. Retrieved from https://scholars.sil.org/sites/scholars/files/gary_f_simons/presentation/doc_ling.pdf and Boerger (Citation: Boerger, 2011) Boerger, B. (2011). To BOLDly Go Where No One Has Gone Before. Language Documentation & Conservation, 5. 208–233. Retrieved from http://scholarspace.manoa.hawaii.edu/handle/10125/4499 ¹ advance the idea that the materials generated as part of a language documentation effort are a corpus, not distinguishing between a corpus and a collection². Many other language documentation practitioners (Citation: Lüpke, 2010) Lüpke, F. (2010). Research methods in language documentation. Language Documentation and Description, 7. 55–104. , (Citation: Good, 2011) Good, J. (2011). Data and language documentation. In Austin, P. & Sallabank, J. (Eds.), The Cambridge Handbook of Endangered Languages. (pp. 212–234). Cambridge University Press. https://doi.org/10.1017/CBO9780511975981.011 , (Citation: Austin, 2013) Austin, P. (2013). Language documentation and meta-documentation. In Jones, M. & Ogilvie, S. (Eds.), Keeping Languages Alive: Documentation, Pedagogy and Revitalization. (pp. 3–15). Cambridge University Press. https://doi.org/10.1017/CBO9781139245890.003 extend or modify the meaning of “corpus” when referring to a collection; using a phrase like language documentation corpus. In contrast to the coalescence of corpus and collection, Thieberger (Citation: Thieberger, 2018) Thieberger, N. (2018). Texts and more texts: corpora in the CoEDL [Blog]. Endangered Languages and Cultures. Retrieved from https://www.paradisec.org.au/blog/2018/02/texts-and-more-texts-corpora-in-the-coedl/ , in a blog post, credits Jane Simpson with proposing the following terms: Assemblage – all material collected, working files, early sources, multiple versions and drafts; Collection – the archived material, a subset of the above, but curated with sufficient metadata to allow the user to know what all items are; Corpus – a crafted set of texts in the language that can be used for further analysis. For a recent treatment of these ideas see Sullivant (Citation: Sullivant, 2020) Sullivant, R. (2020). Archival description for language documentation collections. Language Documentation & Conservation, 14. 520–578. Retrieved from http://hdl.handle.net/10125/24949 which expounds on Simpon’s ideas. I am not convinced that the distinction between assemblage and collection is relevant for a variety of reasons. But my interests are specifically how the distinction might impact the crafting of a bibliographic reference. I suggest that reference formulations do not need to be different between an assemblage and a collection.

Lest I leave the definitions distinguishing a corpus and a collection undisputed, let me suggest some pragmatic distinctions. Biber (Citation: Biber, 1993) Biber, D. (1993). Using Register-Diversified Corpora for General Language Studies. Computational Linguistics, 19(2). 219–241. Retrieved from https://www.aclweb.org/anthology/J93-2001 , (Citation: Biber, 1993) Biber, D. (1993). Representativeness in Corpus Design. Literary and Linguistic Computing, 8(4). 243–257. https://doi.org/10.1093/llc/8.4.243 set up diversity-based criteria for corpus design. One interpretation of Biber’s work is that if a body of evidence does not meet the diversity threshold then it does not qualify as a corpus. The diversity dynamic as a goal for a set of materials generated through language documentation efforts has been a long standing tenant of creating an evidence based documentary record for a language (Citation: Himmelmann, 1998) Himmelmann, N. (1998). Documentary and Descriptive Linguistics. Linguistics, 36(1). 161–195. https://doi.org/10.1515/ling.1998.36.1.161 . Biber’s diversity arguments have even been suggested as guidelines for guiding elicitation in language documentation (Citation: Lüpke, 2010) Lüpke, F. (2010). Research methods in language documentation. Language Documentation and Description, 7. 55–104. . In this sense it is clear that selection is a process which impacts both corpus design and what is collected during fieldwork. However, a second useful criterion for defining a corpus (in contrast to a collection) might be that there is an existing tool set which can be used to exploit the collection of artifacts as a single work. This could lead to an interpretation that a corpus is more like a dataset in the Dublin Core sense, while a collection is more like a collection in the Dublin Core sense—and indeed we see that (Citation: Xia et al., 2016) Xia, F., Lewis, W., Goodman, M., Slayden, G., Georgi, R., Crowgey, J. & Bender, E. (2016). Enriching a massively multilingual database of interlinear glossed text. Language Resources and Evaluation, 50(2). 321–349. https://doi.org/10.1007/s10579-015-9325-4 blur the line between database and corpus in their work. Both datasets (of which databases are a type) and collections can be considered examples of aggregate works. I realize that linguists and language materials archivists have applied these labels loosely but I side with Society of American Archivists, rule 2.3.19 (Citation: , 2013) (2013). Describing archives: a content standard (2nd). Society of American Archivists. Retrieved from http://files.archivists.org/pubs/DACS2E-2013_v0315.pdf in its suggested distinguishing terms for the highest level of an archival unit of materials: “Archival materials are frequently described by devised aggregate terms such as papers (for personal materials), records (for organizational materials), or collection (for topical aggregations)” (see (Citation: Wahbeh, 2009) Wahbeh, F.(2009, 8/10). Retrieved from https://blogs.cul.columbia.edu/schapiro/2009/08/10/on-records-papers-and-collection-a-dacs-case-in-point/ for an example of application). There are other terms which still need to be fleshed out among language archive practitioners such as: archival deposit, accession, and what one might call an artifact and its associated metadata file.

Bibliography