Ruth Singer recaps last week’s Linguistics in the Pub, a monthly informal gathering of linguists in Melbourne to discuss topical areas in our field.
Linguistics in the Pub on Tuesday the 24th of February, 2015 centred around the theme: grammar writing. Harriet Sheppard (Monash University) led the discussion. The announcement and short background reading are here.
The descriptive grammar although often reported to be dead is a form of scholarship that is still very much alive. And although e-grammars are said to be the way of the future, most grammars still take the form of a hard copy, whether it is a PhD thesis or published book. The discussion in this session of linguistics in the pub was kicked off with a discussion of the article by Ulrike Mosel cited below, part of a special publication of LDC on grammar writing.
Mosel, U. 2014. ‘Corpus linguistic and documentary approaches in writing a grammar of a previously undescribed language‘. LD&C Special Publication 8: The Art and Practice of Grammar Writing. (open access)
Among the LIP audience last night were PhD students in their first year who are planning to write either a full grammar for their PhD or a sketch grammar with a focus on specific topics. A few of the linguists present are also editing their descriptive PhD theses for print publication. Thus we all had a keen interest in the methodology of grammar writing, as well as how grammars should be presented to their audience and what the status of descriptive grammars are within our field.
We began the discussion by debating the role of a searchable corpus in developing robust grammatical analyses. Searching a corpus enables us to develop much more rigorous analyses. Corpus searches are also a way to look for variation in grammar through comparisons of speakers, locations or genres. We compared the various tools that we commonly used and discussed how Elan is more powerful than Flex or Toolbox because you can search for features across tiers and output lists of examples that satisfy particular conditions. Although we also noted that the search function can be difficult to use so for some searches specialist help is needed. As usual we discussed the various pros and cons of our favourite tools. It was noted that Flex saves formatting time as examples can be exported as tables and many publishers currently insist that examples are formatted as tables
When we came back to discussing Mosel (2014), one participant wondered why she felt the need to point out the importance of citing the source of each example when presenting it in an academic publication. This seemed obvious to those present yet it is still not a requirement of any publisher and there are still grammars being published for which it is not possible to trace example sentences to a recording in a publicly citable corpus. In addition to this need to cite the source of each example, it was discussed that we need to know whether each example was elicited, recorded in spontaneous narrative or conversation, to really be able to weigh up the claims the author is making. One of the perils of writing a descriptive grammar of an endangered languages is that theoretically the author could say whatever they liked if nobody else has worked on the language. In fact, though descriptive linguists have many ways of evaluating each others’ work, whether it be through their argumentation or through comparison with similar languages that have already been described. Making the source of examples explicit is another way we can make our descriptions convincing.
The lack of a methodology section in most grammars was pointed out as a failing, which makes it difficult to evaluate the data. Typically authors mention how much time they spent in the field. The size of the corpus collected is sometimes mentioned, although there is no standardised measure for corpora (hours recorded, words, morphemes, intonation units etc.). Linguists rarely describe their data collection methods in detail, possibly because none feel they have met the high standards of the ideal (monolingual data elicitation, covering all speech events and registers etc.). Or perhaps linguists assume that everyone is using the same methods as they are when in fact there are quite diverse traditions among documentary linguists. A related tangent was the perennial question: what size corpus do you need to create a good grammar? (see Canberra LIP coming up).
Supervisors discussed that many of the aspirational ideals of documentary linguistics are quite a burden on students who need to balance these ideas with a practical approach. Andy Pawley’s article in the LDC volume in which Mosel (2014) appears was cited as taking this tack. In addition, participants emphasised the need to be opportunistic in terms of the kind of data you collect in the field. Procedural texts for example, may not be a useful genre if they are not something people usually do. And if you force people to create the genre, it may come out unnatural and forced.
We discussed how the history of linguistics and the scholarly traditions that we are part of influence our approach. There are only a few places in the world where the descriptive grammar is an acceptable form for PhD thesis. These include most of the universities in Australia and New Zealand, many in the Netherlands, Brazil, Leipzig MPI and some US universities (see Evans (2010) Dying Words for a similar discussion). We debated whether the presence of many undescribed languages nearby in the Pacific has influenced the value placed on descriptive grammars in Australia and New Zealand, or whether this is due to particular influential linguists such as Bob Dixon and their legacies.
Given that some participants were writing theses, others editing theses into books, we discussed what the differences are. One idea was that in a book, you do not need to discuss alternative analyses of particular phenomena as much as in a thesis, where you need to trace the steps you have taken more carefully. In addition, there is less place for historical speculation in a book than in a thesis, contemplating how phenomena developed etc. We also reflected on the fact that a descriptive grammar is unusual in that it is a reference work, intended to be of use for 50-100 years, possibly more if the language will not be spoken for much longer. This puts a lot of pressure on the linguist to do the language justice.
Next month we will have Miriam Meyerhoff (University of Wellington) to get us all going on the topic of sociolinguistics in the field (probably 25th March TBC)