Search

This study investigates word order in DGS. DGS has been claimed to be an SOV language. This study involves an online judgment task administered online to ~100 DGS signers. This task will test whether event semantics (i.e., reversible and non-reversible, extensional and intensional) influence preferred word orders in DGS. This study is part of a larger project investigating the extent to which the emergence of naturalness at the syntactic level is dependent on the availability of iconicity at the level of the lexicon. This involves investigating what individual biases affect word order in improvised communication (silent gesture) and natural languages (sign languages). This larger study includes examining natural word order in two established sign languages: British Sign Language (BSL) and German Sign Language (DGS). Previous work on improvised silent gestures suggests that word order depends on semantics of action involved (Schouwstra & de Swart 2014). Extensional semantics elicit SOV, while intensional semantics elicit SVO. Previous work comparing an established sign language, Libras, and an emerging sign language, Nicaraguan Sign Language, suggests identical results (Napoli, et al., 2017; Flaherty, et al., under review). However, there were some problems with these studies, such as the inclusion of complex multiverb constructions, variable argument presence, and spatial modifications. Contexts were also all non-reversible, and there was no measure of spoken or written Portuguese or Spanish repertoires. BSL and DGS are both established sign languages with conventions for word order. BSL has been claimed to be an SVO language. DGS has been claimed to be an SOV language. However, both languages exhibit considerable flexibility in word orders. Investigating the preferred basic and flexible word orders in BSL and DGS will allow us to test what drives word order preferences. It will also provide us with information about how principles that were dominant in the emergence of sign languages can potentially still affect a language after it has been used for many generations. This method of testing word order preferences in sign languages is novel. As such, we are pre-registering here only the methods in advance of testing, to ensure comparability with the BSL study, which is ongoing. We will perform statistical analysis modelled on the BSL study, ensuring both data sets are analysed in the same way. Findings from these two studies will be compared with experimental data on English speakers' comprehension of silent gesture word orders. Please note that this registration focuses largely on study design and data collection methods, and not on the analysis plan.

Although there have been some recent advances in sign language recognition, part of the problem is that most computer scientists in this research area do not have the required in-depth knowledge of sign language, and often have no connection with the Deaf community or sign linguists. For example, one project described as translation into sign language aimed to take subtitles and turn them into fingerspelling. This is one of many reasons why much of this technology, including sign-language gloves, simply doesn’t address the challenges. However there are benefits to achieving automatic sign language recognition. The process of annotating and analysing sign language data on video is extremely labour-intensive. Sign language recognition technology could help speed this up. Until recently we have lacked large signed video datasets that have been precisely and consistently transcribed and translated – these are needed to train computers for automation. But sign language corpora – large datasets like the British Sign Language Corpus (Schembri et al., 2014) - bring new possibilities for this technology. Here we describe the project “ExTOL: End to End Translation of British Sign Language” – which has one aim of building the world's first British Sign Language to English translation system and the first practically functional machine translation system for any sign language. Annotation work previously done on the BSL Corpus is providing essential data to be used by computer vision tools to assist with automatic recognition. To achieve this the computer must be able to recognise not only the shape, motion and location of the hands but also nonmanual features – including facial expression, mouth movements, and body posture of the signer. It must also understand how all of this activity in connected signing can be translated into written/spoken language. The technology for recognising hand, face and body positions and movements is improving all the time, which will allow significant progress in speeding up automatic recognition and identification of these elements (e.g. recognising specific facial expressions or mouth movements or head movements). Full translation from BSL to English is of course more complex but the automatic recognition of basic position and movements will help pave the way towards automatic translation. In addition, a secondary aim of the project is to create automatic annotation tools to be integrated into the software annotation package ELAN. We will additionally make the software tools available as independent packages, thus potentially allowing their inclusion into other annotation software such as iLex. In this poster we report on some initial progress on the ExTOL project. This includes (1) automatic recognition of English mouthings, which is being trained using 600+ hours of audiovisual spoken English from British television and TED videos, and developed by testing on English mouthing annotations from the BSL Corpus. It also includes (2) translation of IDGloss to Free Translation, where the aim is to produce English-like sentences given sign glosses in BSL word order. We report baseline results on a subset of the BSL Corpus, which contains 10,000+ sequences and over 5,000 unique tokens, using the state-of-the-art attention based Neural Machine Translation approaches (Camgoz et al., 2018; Vaswani et al., 2017). Although it is clear that free translation (i.e. full English translation) cannot be achieved via ID glosses alone, this baseline translation task will help contribute to the overall BSL to English translation process - at least at the level of manual signs.

This paper presents results from a corpus-based study investigating lexical variation in BSL. An earlier study investigating variation in BSL numeral signs found that younger signers were using a decreasing variety of regionally distinct variants, suggesting that levelling may be taking place. Here, we report findings from a larger investigation looking at regional lexical variants for colours, countries, numbers and UK placenames elicited as part of the BSL Corpus Project. Age, school location and language background were significant predictors of lexical variation, with younger signers using a more levelled variety. This change appears to be happening faster in particular sub-groups of the deaf community (e.g., signers from hearing families). Also, we find that for the names of some UK cities, signers from outside the region use a different sign than those who live in the region.

Digital video archives of Auslan (Australian sign language) and BSL (British Sign Language) are slowly being transformed into machine-readable linguistic corpora. Each archive (Auslan 2004-2008, BSL 2008-2001) consists of data collected from deaf native and near-native signers. The datasets are being annotated using ELAN software. The majority of the video data will be made accessible online (with some limits to access for sensitive data). In this presentation, we report on the on-going studies of lexical frequency in these two signed languages—63,436 sign tokens produced in 360 clips by 109 participants in the currently annotated Auslan dataset, and 25,000 sign tokens from the corpus conversation data in the BSL dataset (500 signs each from 50 participants). Preliminary results signs indicate that between 65% and 60% of the Auslan and BSL data respectively consist of signs from the core lexicon (i.e. those signs which are highly conventionalised in form and meaning across contexts, (see Johnston, 2011, Johnston &Schembri, 1999, 2010). The next two largest categories are pointing signs (12% and 23% respectively) and signs from outside the core lexicon (i.e., gestures and sequences of enactment or 'constructed action') (6.5% and 9% respectively). The remaining number of tokens consists of fingerspelled signs (5% in both datasets), depicting constructions (i.e., depicting verbs of location, motion and/or handling, 11% and 3% respectively), and sign names (0.2 and 0.3% respectively). We discuss some of the challenges creating a lemmatised corpus of a sign language, including difficulties in differentiating core from non-core signs and sign from gesture, as well as how our work informs both sign language documentation and description specifically and linguistic theory more generally. Johnston, T. (2011). Lexical frequency in sign languages. Journal of Deaf Studies and Deaf Education. Johnston, T., &Schembri, A. (1999). On defining lexeme in a signed language. Sign Language and Linguistics, 2(2), 115-185. Johnston, T., &Schembri, A. (2010). Variation, lexicalization and grammaticalization in signed languages. Langage et Société, 131, 19-35. PARADISEC (Pacific And Regional Archive for Digital Sources in Endangered Cultures), Australian Partnership for Sustainable Repositories, Ethnographic E-Research Project and Sydney Object Repositories for Research and Teaching.