Fieldwork, grammar-writing, and archiving

Starting in 2021, I got more involved in doing fieldwork on Armenian lects. I started working on two descriptive grammars. This page keeps track of questions, answers, and advice that I found.

Thanks to Peter Austin, Christine Beier, Bernard Comrie, Martin Haspelmath, Joseph Lovestrand, Laura McPherson, Marianne Mithun, Hannah Sande, Peter Schuelke, Jonathan North Washington.

What to do read before you write

Before you start, read some guides on grammar-writing and fieldwork. The following were the most-often suggested to me. I didn’t read every article, but I went for the titles that resonated the most with me.

Marianne Mithun and Bradley McDonnell have helpful syllabi.

What to read while writing

While writing on a specific topic, it’s always helpful to double-check your descriptions and analyses against cross-linguistic descriptive work.

There are various questionnaires that can help you in guiding what sorts of data points to cover:

It’s also a good idea to check on a grammar of an areally or genetically related language.

Where to publish

There are a small number of series for publishing grammars.

Only one of the above 4 publishers is open-access though. Some of them like Lincom provide open-access options for a price.

For smaller sketches, some nice options are the following:

What to write

The grammar has 3 core parts

Grammar proper

  • Your descriptions and analysis. If possible, provide recordings for every word and example
  • Provide glossing, IPA transcription, translation, and orthography for everything.

Dictionary or glossary

  • A small word list with orthography + meaning + IPA + recordings + nuanced part-of-speech + subcategorization frames.
  • The content can be small like a Swadesh list, or more comprehensive like the top 2k most common words in the language.
  • The glossaries that you can find in language textbooks are helpful in giving lists of words to catalog.

Text

  • A sample of the language that includes orthography + translation + IPA transcription + recording + gloss.
  • You can make comments on subtle syntactic constructions, and then cite the sample in your grammar.
  • Try to have around 10-40 pages of annotated text. The material can come from published books, newspapers, pre-existing archives, YouTube and social media videos (ask for permission), or natural speech that you recorded.
  • If you want to record your own samples, sociolinguistic interviews would be helpful. See Tagliamonte‘s talk for a riveting demo.
  • The content should come from different genres of non-dialogue language use (Christine Beier, p.c.):
    • Historical narrative about non-personal history: in 1492, bla bla.
    • Personal narrative
    • Hortative speech (sermons, speeches): We must bla bla.
    • Advise-giving texts: My child, do bla bla.
    • Wishful-thinking texts: What would you do with 1 million dollars.
    • Procedural text: How do u make a garden?
    • Mythological
  • It’s recommended that you collect texts early in your fieldwork.

How much to write

When writing, your goal is to over as many common constructions in the language. The questionnaires from What to read while writing give a good sense of how much meat is good enough for large grammars.

But you don’t have infinite time so you can’t reasonably write a complete grammar of a language. Writing a complete grammar likewise needs a lot of data, more than one person can collect in a lifetime. So write until you’re comfortable with your depth and breadth.

How do I record

If recording remotely, then the easiest strategy for me was a) my informant is recording themself on audacity while b) I zoom with the informant to guide them through the elicitations.

But unfortunately, you can’t expect your informant to always have the right microphones, laptops, rooms, and so on. So you can’t expect every recording to be high-quality. But any recorded (and transcribed) material is better than nothing. Keep track of how the recordings were made for different sessions because the types of devices and software can drastically change the actual acoustic properties of speech (Sanker et al 2021, slew of work collected by Joseph Lovestrand).

If possible and safe, then recording in a soundproof room is optimal. I’ve had informants go to phonology labs at nearby universities, and I zoom in with them to guide them through the elicitations.

Where to upload your material

It’s always a good idea to provide complete paradigms and transcribed recordings of your data. It can take a while to find a suitable and affordable academic repository for them. Some possible strategies can be found here.

In the meantime, a temporary fix is to upload your material across GitHub, Google Drive, and OSF. You can then make a DOI of your uploaded personal archive, e.g., with Zenodo.