[Lingtyp] How do typologists use examples in grammars?

Thu Jun 17 14:31:02 UTC 2021

Dear Adam, Ilana, et al. — I’d like to add my emphatic support to what Adam said. I was in fact just discussing this very issue with a collaborator over email independently. It goes beyond descriptive linguistics, and in fact beyond linguistics: when the available data about a given phenomenon is compatible with more than one competing analysis/hypothesis, it’s bad practice, but unfortunately nonetheless widespread practice, to reduce the presentation of the data to only the analysis/hypothesis favored by the researcher. This is a very general concern in science in general, but in linguistics in particular, there are a couple of unique twists to it.

Researchers throughout academia of course always face a temptation to push their favorite analysis/account of a given phenomenon, because they want to promote that account and project confidence in it. 

And the specific additional sources of the problem in linguistics are these:

1. An analysis of any given phenomenon (including a construction or sound) of course presupposes analyses of a slew of other, related phenomena (e.g., constituent constructions). This is again a problem not restricted to linguistics. But languages are such darn complex creatures that it is impossible for any linguist to have equally good and comprehensive data for every phenomenon their work draws on. It’s a bit like a video game, where pixel density decreases in the background (I’m told): say your best/favorite example for illustrating a given phenomenon features a morpheme that you’ve only ever encountered in two or three tokens/utterances. You realize that you don’t have enough data to be confident what this thing is, but you’re reluctant to talk about that because it would distract from your discussion of the phenomenon of interest. I have certainly been there.

2. As linguists, we actually know that constructions can be inherently ambiguous between multiple analyses. That may be the result of ongoing change and/or inter-speaker variation. But the inherently non-unique mapping between language-specific categories and comparative concepts is another important source. 

It’s not quite pithy enough for a bumper sticker, but “Linguistic phenomena are under no obligation to fit our preconceived categories” sums up how I’ve come to think about the last-mentioned issue.

Best — Juergen

> On Jun 17, 2021, at 6:22 AM, Ilana Mushin <i.mushin at uq.edu.au> wrote:
> 
> Thank you Adam for raising this fascinating topic. I think your views here are really important to air. Grammatical descriptions are renditions by the describers of a distillation of all of the complexity of linguistic structures and the contexts in which they occur. We do need to think more critically about the ways in which we, as typologists, grasp descriptive nuggets of languages we do not ourselves know. 
> Lots of satisfying food for thought. 
> 
> Ilana
> 
> Associate Professor Ilana Mushin
> Reader in Linguistics
> President, Australian Linguistic Society
>  
> Co-Editor, Interactional Linguistics (https://benjamins.com/catalog/il) 
>  
> School of Languages and Cultures
> University of Queensland
> St Lucia, QLD 4072.
> Ph: (07) 3365 6810
>  
> CRICOS Provider No: 00025B
>  
> I acknowledge the Jagera and Turrbal peoples on whose land I live and work. Their sovereignty was never ceded.
> 
>> On 17 Jun 2021, at 7:57 pm, Adam James Ross Tallman <ajrtallman at utexas.edu> wrote:
>> 
>> 
>> Some of these issues depend somewhat on how each of us do typology and the way we read grammars. I would endorse all of the statements given so far, but would also highlight one aspect of grammar writing that I think we need to work on as a community. 
>> 
>> I often wish descriptive grammarians were more forthcoming about alternative analyses, when data point in different directions, and aspects of the language that are just not understood yet. This is important for reproducibility, but also I think grammars can have an important function in mapping out domains that require future research. In typology, we could potentially code uncertainty in the analysis if that information is present(see here) resulting in more reliable inferences.  
>> 
>> Personally, I think the attitude of typologists (or confessional "theoreticians") has sometimes had a detrimental effect here. There is often an assumption that a lack of certainty on the descriptivist's part regarding how to present or analyze some grammatical fact of a language automatically translates to a lack of knowledge about the language or linguistics in general, when I think, often the opposite is true. It is because the descriptive linguist has a detailed understanding of the phenomena that it cannot be easily fit out with the typologist's terminology or naively constructed comparative concepts ("what!? you don't understand where/what a word is in your language, come back to me when you really understand the language (if not linguistic theory in general *sniff*)!"). 
>> 
>> The upshot of this attitude, and the fact that descriptive and documentary linguistics was demoted in prestige in the 1960s (although things are changing), is  that the descriptivist is impelled to pretend (and then eventually believe) that a much more straightforward analysis is licensed. Analyses for which there are mountains of supporting data are treated the same as those made as best guesses. Glosses or translations are treated as "truth", rather than as heuristics or expositional devices respectively. I think this tendency is changing as a consequence of the development of documentary linguistics and archiving practices, but I think descriptive linguists have special insights about methodological and analytic challenges that could make their ways into grammars more often. Or in short, I want thick translation, and also thick analysis, to be common in grammar writing.
>> 
>> best,
>> 
>> Adam
>> 
>> On Thu, Jun 17, 2021 at 5:56 AM Randy J. LaPolla <randy.lapolla at gmail.com> wrote:
>> Hi Daniel,
>> Chelliah, Shobhana. 1997. A Grammar of Meithei. MGL 17. Berlin & NY: Mouton de Gruyter.
>> is one grammar that does the two line (phrase and clause) translation consistently. It certainly helps with the interpretation.
>> 
>> Randy
>> -----
>> Randy J. LaPolla, PhD FAHA （羅仁地）
>> Professor of Linguistics, with courtesy appointment in Chinese, School of Humanities 
>> Nanyang Technological University
>> HSS-03-45, 48 Nanyang Avenue| Singapore 639818
>> http://randylapolla.net/
>> 
>>> On 17 Jun 2021, at 6:05 AM, Daniel Ross <djross3 at gmail.com> wrote:
>>> 
>>> This has been a useful discussion and I generally agree with the points made by others already.
>>> 
>>> But one thought I'd like to add to this discussion is the importance of the third line, the English translation itself (or whatever the language of description is). Whenever any not-at-issue feature is included in an example, we may rely on that third-line translation for an intuitive understanding of the function of the morphemes in an example beyond what the gloss indicates, especially when it is a relationship between morphemes. A basic illustration of this problem would be the choice to render a verb unmarked for tense as past tense in an example (probably corresponding to the context of the original elicitation, even though that may not be indicated when it is used as an isolated example out of context), or similarly the choice of "he" or "she" for a general 3SG pronoun unmarked for gender.
>>> 
>>> This is a bigger problem, though, when we consider combinations of morphemes. The glosses can tell us something about the individual morphemes, but they don't indicate how to interpret their combination. For example, could there be some interaction between a tense marker and mood? The gloss doesn't indicate that, only the translation (and possibly some other section in the grammar that addresses it directly, if included).
>>> 
>>> The only solution I can suggest for this would be to include two translations, expanding the traditional three-line model to four lines: first a literal translation, and second a more idiomatic English version. For some examples that might seem excessive, but I think often it would be incredibly helpful. English simply doesn't function like many languages it is used to describe, and having an awkward-in-English but more literal translation would help me as a typologist to understand how the language works internally rather than through the lens of English. To some degree the morpheme-by-morpheme gloss is supposed to be used for that purpose, but again it doesn't indicate anything about the semantics of those morphemes when combined. And quite rarely we do have exceptional examples with a literal translation indicated, but almost exclusively when that particular phenomenon is at-issue in that particular section of a grammar, not elsewhere, and often only when there is an obvious lexical mismatch (e.g. in a collocation), not so much for grammatical functions. Personally I would also be comfortable with a less idiomatic English translation, because I think it would be helpful to have the reader pause a moment to try to understand the example in the sense of its usage in the language being described, rather than assuming functional equivalence to an English rendering. Of course that shouldn't go too far as to leave the meaning unintelligible in English.
>>> 
>>> To illustrate this, here is an excerpt from a recent publication where I mentioned this issue (page 53 of: https://doi.org/10.1515/9783110692099-002). I point out that more familiar purposive forms are often used to translate Prior Associated Motion, because that category is not so familiar for the typical European languages used to write grammars. And even when the language does have a corresponding form like English "go and get" (or "go get"), authors may not be so comfortable using it, despite it actually being a better translation.
>>> 
>>> "Glosses in English can sometimes be misleading: 'go (in order) to [verb]' should be expected for purposives, whereas 'go and [verb]'[34] should be expected for Prior [Associated Motion],[35] and rarely do grammars specifically describe the semantics of such constructions beyond the glosses.
>>> 
>>> [34] That is, pseudocoordination..., which problematically is not itself widely recognized for its Prior AM semantics, resulting in potential confusion for both authors and readers of descriptive materials.
>>> [35] Glosses in languages other than English are often similarly unhelpful, especially when these meanings are not typically distinguished (e.g. Spanish ir 'go' + INF, used to gloss either purposive 'go (in order) to' or Prior AM)."
>>> 
>>> This meant that in the survey I was not able to confidently distinguish between Purposive and Prior motion in all of my sample languages, so I had to report them together as a single type.
>>> 
>>> Daniel
>>> 
>>> 
>>> 
>>> 
>>> On Wed, Jun 16, 2021 at 10:30 AM Peter Austin <pa2 at soas.ac.uk> wrote:
>>> Tamsin Donaldson's grammar of Ngiyampaa (CUP, 1980) is exemplary in providing this kind of sociocultural and utterance context. An oldie but a goodie.
>>> 
>>> Best
>>> Peter
>>> 
>>> On Wed, 16 Jun 2021, 17:17 Marianne Mithun, <mithun at linguistics.ucsb.edu> wrote:
>>> Thanks Francoise and Lena! I heartily agree!
>>> 
>>> (And now that we're not buying so much paper and ink, this really doesn't increase the cost so much.)
>>> 
>>> Marianne
>>> 
>>> On Wed, Jun 16, 2021 at 3:15 AM Françoise Rose <francoise.rose at univ-lyon2.fr> wrote:
>>> Dear all,
>>> 
>>>  
>>> 
>>> I will send my answers to Eline in a private message, but I would like to share a recent experience regarding examples in grammars.
>>> 
>>>  
>>> 
>>> I am presently reviewing Lena Terhart’s grammar of Paunaka (a PhD thesis). The author gives this notice regarding how examples are introduced in the grammar:
>>> 
>>>  
>>> 
>>> “One peculiarity of this work is that most examples are introduced by briefly providing
>>> 
>>> the extralinguistic context. This is usually not done in grammatical descriptions. I
>>> 
>>> started with this at some point, when I felt that context was necessary for understanding
>>> 
>>> and then extended it further and further. Thus, the reader will not only learn about
>>> 
>>> Paunaka, but also gain knowledge about the narratives and personal life stories of the
>>> 
>>> speakers throughout this work. Whoever is irritated by this is kindly asked to simply
>>> 
>>> overlook it. »
>>> 
>>>  
>>> 
>>> I generally appreciate a lot when authors add important information on the context of the extract (usually in brackets before the free translation). I was yet planning to be one of the readers who would just overlook these example introductions because my sense was that these introductions would be useful sometimes only (in my view, mainly in sections concerning discourse issues). Actually, it turned out that I appreciate this practice very much: knowing the context very often gives a very clear interpretation to the sentence and strong indications on how the linguistic features under scrutiny works. Also, this is a very strong indication that the linguist perfectly understands what the sentence is about. I can attest that, as a fieldworker on culturally very distinct cultures from mine, I do not always understand the contribution of every sentence to the conversation/narration. Finally, this gives also a sense of familiarity with the culture involved in the grammar, which is extremely pleasant.
>>> 
>>>  
>>> 
>>> Here are two randomly-picked examples from the grammar:
>>> 
>>>  
>>> 
>>> Consider (516), which is about making something fall, just like (510) above. It also comes from a description of the frog story,
>>> 
>>> but this sentence was produced by Miguel and referred to another picture, the one on which the dog has made the beehive (or: wasp nest) fall.
>>> 
>>>  
>>> 
>>> A similar example comes from María S. who had just stated that smoking is bad
>>> 
>>> and now provides the reason:
>>> 
>>>  
>>> 
>>> Best to you all,
>>> 
>>> Françoise
>>> 
>>>  
>>> 
>>>  
>>> 
>>>  
>>> 
>>> De : Lingtyp <lingtyp-bounces at listserv.linguistlist.org> De la part de Eline Visser
>>> Envoyé : lundi 31 mai 2021 15:15
>>> À : lingtyp at listserv.linguistlist.org
>>> Objet : [Lingtyp] How do typologists use examples in grammars?
>>> 
>>>  
>>> 
>>> Dear typologists,
>>> 
>>>  
>>> 
>>> I’d like to learn more about how you use the examples given in grammars. I have just finished a grammar myself, and will continue to do descriptive work in the future, and this is a topic that fascinates me. I'm especially interested in knowing if one can discern the traits of a good example (for typological use). I’d be glad if some of you could take the time to answer the questions below, either briefly or elaborately. You can email me the answers. Also, if there’s is anything published on this topic please do let me know.
>>> 
>>>  
>>> 
>>> 1. In general, do you prefer short (let’s say <1 line) or longer (> 1 line) examples? Elaborate if you wish.
>>> 
>>>  
>>> 
>>> 2. In general, do you have a preference for examples from a certain genre? Which? You can interpret genre broadly or narrowly, in which ever way you like: monologue, dialogue, anecdotes, recipes, hymns, picture-matching tasks…
>>> 
>>>  
>>> 
>>> 3. In general, do you have a dispreference for examples of a certain genre?
>>> 
>>>  
>>> 
>>> 4. Say you have two examples that illustrate your point equally well. What could be a deciding factor for choosing one over another?
>>> 
>>>  
>>> 
>>> 5. Say you can’t find an example that illustrates your point well. On a scale from 1-5, how likely is it that you will go to the language’s corpus or the attached texts in the grammar to find one yourself? (1= very unlikely, 5 = very likely)
>>> 
>>>  
>>> 
>>> 6. Anything else you’d like to share about examples in grammars? Feel free to rant.
>>> 
>>>  
>>> 
>>> Eline
>>> 
>>>  
>>> 
>>> P.s. For those who ordered a Kalamang grammar hard copy - they’re in Sweden, I’m in Norway, traveling isn’t as easy as I thought yet, so this takes a bit longer than I thought, sorry!
>>> 
>>>  
>>> 
>>>  
>>> 
>>>  
>>> 
>>> _______________________________________________
>>> Lingtyp mailing list
>>> Lingtyp at listserv.linguistlist.org
>>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>>> _______________________________________________
>>> Lingtyp mailing list
>>> Lingtyp at listserv.linguistlist.org
>>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>>> _______________________________________________
>>> Lingtyp mailing list
>>> Lingtyp at listserv.linguistlist.org
>>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>>> _______________________________________________
>>> Lingtyp mailing list
>>> Lingtyp at listserv.linguistlist.org
>>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>> 
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
>> 
>> 
>> -- 
>> Adam J.R. Tallman
>> Post-doctoral Researcher 
>> Friedrich Schiller Universität
>> Department of English Studies
>> _______________________________________________
>> Lingtyp mailing list
>> Lingtyp at listserv.linguistlist.org
>> http://listserv.linguistlist.org/mailman/listinfo/lingtyp
> _______________________________________________
> Lingtyp mailing list
> Lingtyp at listserv.linguistlist.org
> http://listserv.linguistlist.org/mailman/listinfo/lingtyp