Applying Leipzig glossing rules

Matti Miestamo matti.miestamo at
Tue Oct 13 08:56:18 UTC 2009

Dear All,

and thanks, Riho, for initiating this discussion!

Riho raises two points in his message.

The first point is (also) about the technique of glossing. The Leipzig 
glossing rules allow both variants, either the dot or the '>'-sign 
between the subject and the object. Both variants have their positive 
and negative sides, but the innovation of allowing the arrow to be 
turned around solves both problems, so this would indeed be a good 
addition to the LGR in general, not just for Uralic!

As to the second point, it is more about how to analyse certain forms, 
not so much about glossing technique. Whatever analysis one adopts, and 
people will adopt different analyses, the LGR with its different 
optional variants can handle the glossing.

Note also that one and the same researcher may gloss the same form 
differently depending on context. If one is talking about word order, 
one may be satisfied with a less detailed glossing than when talking 
about morphology. For example, depending on context, the Finnish verb 
form "menin" can be glossed as:

menin		men-i-n
go.PST.1SG	go-PST-1SG
'I went'	'I went.'

In sum, there is not just one way of glossing, and no committee can fix 
the correct way of glossing a particular form. The LGR are meant to 
allow for flexibility within a set of constraints.

I would also like to point out that the Leipzig Glossing rules have, so 
to speak, both a syntax and a lexicon. The syntax are the actual rules 
of glossing, and the lexicon is the list of abbreviations suggested in 
the Appendix. The list of abbreviations is not exhaustive and one often 
needs to use abbreviations that are not on the list. In a publication, 
it is always important to mention, e.g in a footnote, that one sticks to 
the syntax of the Leipzig Glossing rules (if one does) and to list the 
abbreviations used.

The syntax of the rules are meant to be universally applicable to 
glossing languages from any family. Amendments can of course be 
suggested to the Leipzig people (e.g., the issue of ">" vs. "<"), but no 
special rules are needed for Uralic. As to the lexicon, i.e. the 
abbreviations, it might indeed be a good idea to make a standard list of 
abbreviations for Uralic languages. Just a minor detail here: for 
connegative, I would use (and have used) CNG instead of CONNEG.

Best wishes,
Matti Miestamo

Matti Miestamo

rgruntha at wrote:
> Dear Uralicists,
> yesterday, we had a discussion on Leipzig glossing rules 
> ( at the PhD 
> seminar of our department. After launching this encoding principle the 
> typologists maintain that, as in linguistic research in general, the 
> glossing rules should be adapted to individual languages only by taking 
> into account language-specific characteristics. This recommendation is 
> very flexible and gives the research of individual languages full 
> freedom to make case-specific applications when and where-ever needed. 
> However, we discussed a couple of more general questions in the light of 
> the Uralic language family and tried to find applied glossing rules that 
> would be acceptable for as many Uralic languages as possible. Given that 
> doing research on Uralic languages would be more consistent, if shared 
> principles were applied according to common rules in the research of the 
> same languages, we tried to find subsequent principles to be followed.
> Firstly (A), OBJECT CONJUGATION is morphologically one of the most 
> complex categories in those languages that have it. The Leipzig glossing 
> rules (Rule 4, examples (11 and 19)) suggest that there are two 
> alternative ways to encode morphosyntacically cumulative forms. 
> Accordingly, the Erzya example (1a-b) can be glossed in two ways:
> (1a)     rivez'    s'ovn-i-z'e        ver'giz'-en'
>     fox    abuse-PRET-3SG.A.3SG.P    wolf-GA
>     ‘The fox abused the wolf.’ (MSFOu 84: 279)
> (1b)     rivez'    s'ovn-i-z'e    ver'giz'-en'
>     fox    abuse-3SG>3SG    wolf-GA
>     ‘The fox abused the wolf.’ (MSFOu 84: 279)
> (GA = genitive-accusative (genitive in the position of object))
> The previous one (1a) follows the linear encoding principle but is 
> robust and implies that the agent-like and patient-like constituent are 
> segmentable at least at a semantic level. In our group the latter one 
> (1b) was considered as the better alternative from the viewpoint of many 
> Uralic languages, because it is more explicit and shows the hierarchy 
> between the agent-like (subject) and patient-like (object) consituent. 
> However, it is often the case that if the object is morphologically 
> distinguishable as it may be in the languages of Siberia, it is located 
> before the agent- (subject-)marking morpheme, on its left side.
> Consequently, from the viewpoint of morphology it would make more 
> justice, if the arrow | > | was turned around | < |. This should work 
> both for Mordvinic (examples 2a-c), Ugric (Mansi in example (3), 
> Hungarian in example (4)) and Samoyedic. There is also the point that 
> the Ob-Ugric and Samoyedic languages distinguish between the number of 
> the object (SG/DU/PL) but not the person.
> Erzya
> (2a)    kunda-si-n'ek
>     catch-3SG-1PL (alternatively catch-3PL-1PL)
>     ‘We catch her/him/them.’
> (2b)    kunda-sin'ek
>     catch-3SG<1PL (alternatively catch-3PL<1PL)
>     ‘We catch her/him/them.’
> (2c)     rivez'    s'ovn-i-z'e            ver'giz'-en'
>     fox    abuse-3SG<3SG    wolf-GA
>      ‘The fox abused the wolf.’ (MSFOu 84: 279)
> Mansi
> (3)    am tuw towləγtasum, nooŋx ti puuγaslum
>     I there wing-FREQ-PRET-1SG, up this catch-PRET-SG<1SG
>     ‘I rowed there fast, I caught it up.’ (Kálmán 1976: 81)
> Hungarian
> (4)    lát-lak
>     see-2SG<1SG
>     ‘I see you.’
> We tested the same applied glossing principle on Nenets, but 
> unfortunately, here I have to leave it to specialists of the Samoyedic 
> languages to make the test publicly. Note that in Erzya the glossing 
> with hyphens in example (2a) is problematic, because -n'ek (1PL) is 
> segmentable, whereas -si actually does not match with 3SG nor 3PL one to 
> one.
> Secondly (B), connegative forms of verbs are seldom marked overtly. The 
> Leipzig glossing rules, for instance, do not include an abbreviation for 
> connegative forms. However, in the Uralic languages the connegative verb 
> form often diverges from the verb stem and could therefore be encoded. 
> Moreover, the connegative usually corresponds to imperative 2SG forms 
> that, in practice, always have to be encoded. So, in our view the 
> connegative should be encoded as well as in the North Saami (5) and Mari 
> (6a-b) examples.
> North Saami
> (5)    Dasgo ii oktage olmmoš dahkkojuvvo vanhurskkisin
>     for NEG.3SG anyone human do-PASS.CONNEG righteous-ESS
>     ‘No one will be made righteous…’
> Mari
> (6a)    tyške tol!
>     this-LAT come.IMP.2SG
>     ‘Come here!’
> (6b)    tyške ot tol mo?
>     this-LAT NEG-2SG come.CONNEG Q
>     ‘Won’t you come here?’
> In sum, we found these conclusions useful and acceptable for the 
> glossing of those languages we are concerned with. We look forward to 
> the feedback of other Uralicists and your comments on both the suggested 
> applying principles and the applicability of Leipzig glossing rules in 
> general.
> With best regards from Helsinki!
> Riho Grünthal
> Department of Finno-Ugrian Studies
> P.O.Box 24 (Unioninkatu 40)
> FI-00014 University of Helsinki
> -- 
> ura-list at - list for Uralic linguistics and related disciplines
> to (un)subscribe, send majordomo at a message:
> (un)subscribe ura-list my.own at email.address
> Mirror archive:
ura-list at - list for Uralic linguistics and related disciplines
to (un)subscribe, send majordomo at a message:
(un)subscribe ura-list my.own at email.address
Mirror archive:

More information about the Ura-list mailing list