Sign language corpus data are useful to study phonological and lexical variation. The focus in studies of lexical variation has often been on the regional distribution and age variation of competing variants (e.g. McKee & McKee 2011, Stamp et al. 2014). When studying variation in corpus data, it is helpful not only to look at the distribution of each single variant candidate but to contrast all synonymous candidates of a variant cluster as well. Hanke et al. 2017 explored the possibility to detect regional and age variations in lexeme clusters that share a meaning as indicated by the same gloss name but differ in form. Clusterings on the basis of gloss names can only be a first step of analysis. The findings suggest two further directions of research that will be explored in this paper. 

First, it stands out that in the same region often several variants of a seemingly synonymous lexeme cluster are used. Thus, one can take a closer look at individual regions in order to identify what conditions or reasons might determine the choice of variants within that region. As long as it is different persons using different variants, the reasons could be personal preference (including conscious choices for reasons of political correctness) and different backgrounds of signers (e.g. school, age, influence of signing parents or partners). However, cases of individual signers using several of the signs of a variant cluster require further explanation. Possible determining factors might be different syntactic behaviours, phonotactic, or iconic reasons, or pragmatic reasons such as adapting to the conversation partner’s lexical choice. Also, it can be expected that seemingly synonymous signs actually differ with regard to their contextual meaning and usage and thus do not cover the same range of senses.

Corpus data can help to investigate possible conditions of variant choice more thoroughly. We identified sign clusters that seemed to be synonyms according to the rough meaning indication provided by the glosses and narrowed the search down to clusters that had several informants using several synonym candidates of the same cluster. These were the best starting point for a contrastive usage analysis. We investigated the conditions that determine the choice of variants for these individual signers. For this step further detailed annotation such as sense tagging was necessary.

The cluster ‘talk, speak’ consists of five basic sign types, all located at or near the mouth in their citation forms. Signers tended to use different signs when the modality of using spoken language mattered as opposed to when they referred just to the content of a statement regardless whether it was signed or spoken.

The lexeme cluster ‘together’ invites a further sense discrimination. There are two forms with the sense ‘together, as a group’ as well as two forms with the sense ‘together, as two persons’. For the few informants who vary signs of the same sense, we could identify phonotactic reasons (primed by handshape of subsequent sign), stylistic reasons (two-handed form allows for en­larging, one-handed variant does not) and syntactic reasons (different semantic roles indicated).

The variant candidates investigated differed with regard to usage, i.e. the semantic and/or syntactic conditions of use. These findings reflect the fact that the gloss name as a rather rough indication of meaning can be used as a starting point but requires finer-grained analysis – as it is done for lexicographic purposes within our project. 

Second, the analysis along the lines of “same concept – different form” can be complemented by an analysis of the distributional patterns of same form and different concepts (sampling polysemous signs as well as homonyms). This change of perspective is motivated by the incidental observation that two rather different meanings of a sign form (hence homonyms) greatly differed in regional distribution. It has been suggested for spoken (Gilliéron & Roques 1912:17) and also for signed languages (Boyes Braem 1981:44, Cuxac 2000:154) that the principle of avoiding homonyms may lead to a differentiation of form for differing meanings although this point of view has also been challenged (Lass 1980:75-80). So can corpus data support the theory of homonymy avoidance? 

The sign form at hand means ‘girl’ in Northern Germany and ‘Friday’ in the Southern part of Germany. The existence of another sign for ‘girl’ in Southern Germany might be explained as a result of homonymy avoidance. However, the analysis with regard to age and regional distribution shows that the use of the described sign forms is relatively stable over age groups and regions. That is, there is no direct diachronic evidence for homonymy avoidance (which might be due to the limited time span of approx. 50 years covered by apparent time in the DGS corpus). Another sign with the meaning ‘woman’ seems to have its origin in the region of Berlin. This sign spreads over Germany the younger the age group, but is rarely found in the southern parts of Bavaria where that very form is used with the meaning ‘bread’. This case might be speculatively explained as homonymy avoidance being active in the apparent time span covered.

From the data, it is clear that homonymy avoidance is at least not a strong factor determining lexical selection in a region: There are also many examples of lemma pairs or clusters where homonymous signs are used in one region (e.g.  ‘class’ – ‘wood’ – ‘why’ with alternatives available, or  ‘Monday’– ‘father’ with ‘father’ used throughout Germany, but only in Bavaria the sign form also means ‘Monday’).

In the examples presented in this paper, we looked at cases of potential avoidance resulting in the selection of different lexemes, not phonological variants. In order to analyse cases of homophony restricted to specific inflected forms (cf. Baerman 2011), we expect progress once a larger part of the DGS corpus has undergone more detailed annotation.

In conclusion, when analysing the distribution and especially the regionality of polysemous signs in the context of competing variants, one should primarily take potential differences in meaning into account and ideally consider different uses or senses of a sign. However, such studies require at least some degree of sense tagging which is not part of the general annotation scope of any larger sign language corpus up to date. 

References 

Baerman, Matthew (2011): Defectiveness and homophony avoidance. Journal of Linguistics, vol. 47, no. 1, pp. 1-29. https://doi.org/10.1017/S0022226710000022.

Boyes Braem, Penny (1981): Features of the Handshape in American Sign Language. Ph.D. thesis, University of California, Berkeley.

Cuxac, Christian (2000): La langue des Signes Française (LSF). Les voies de l’iconicité. Paris: Ophrys.

Gilliéron, Jules / Roques, Mario (1912): Études de géographie linguistique d'après l'Atlas linguistique de la France. Paris: Champion.

Hanke, Thomas / Konrad, Reiner / Langer, Gabriele / Müller, Anke / Wähl, Sabrina (2017): “Detecting Regional and Age Variation in a Growing Corpus of DGS”. Poster presented at the workshop “Corpus-based approaches to sign language linguistics: Into the second decade”, Birmingham UK, July 2017. http://dgs-korpus.de/files/inhalt_pdf/DGS-Korpus_Poster_Birmingham2017_Variation.pdf.

Lass, Roger (1980): On explaining language change. Cambridge: Cambridge University Press.

McKee, Rachel / McKee, David (2011): “Old Signs, New Signs, Whose Signs?: Sociolinguistic Variation in the NZSL Lexicon.” Sign Language Studies, vol. 11 no. 4, pp. 485-527. https://doi.org/10.1353/sls.2011.0012.

Stamp, Rose / Schembri, Adam / Fenlon, Jordan / Rentelis, Ramas / Woll, Bencie / Cormier, Kearsy (2014): “Lexical Variation and Change in British Sign Language.” PLoS ONE 9(4): e94053. https://doi.org/10.1371/journal.pone.0094053.