Recognition of Non-Manual Content in Continuous Japanese Sign Language

The quality of recognition systems for continuous utterances in signed languages could be largely advanced within the last years. However, research efforts often do not address specific linguistic features of signed languages, as e.g., non-manual expressions. In this work, we evaluate the potential of a single video camera-based recognition system with respect to the latter. For this, we introduce a two-stage pipeline based on two-dimensional body joint positions extracted from RGB camera data. The system first separates the data flow of a signed expression into meaningful word segments on the base of a frame-wise binary Random Forest. Next, every segment is transformed into image-like shape and classified with a Convolutional Neural Network. The proposed system is then evaluated on a data set of continuous sentence expressions in Japanese Sign Language with a variation of non-manual expressions. Exploring multiple variations of data representations and network parameters, we are able to distinguish word segments of specific non-manual intonations with 86% accuracy from the underlying body joint movement data. Full sentence predictions achieve a total Word Error Rate of 15.75%. This marks an improvement of 13.22% as compared to ground truth predictions obtained from labeling insensitive towards non-manual content. Consequently, our analysis constitutes an important contribution for a better understanding of mixed manual and non-manual content in signed communication.

Markers of epistemic modality and their origins

Native deaf signers express epistemic modality by different means: mental-state words, clause-internal particles, signs indicating hypothesis, and nonmanually. The data for this study come from two unrelated sign languages, Danish Sign Language and Japanese Sign Language. In dialogues the signers use both calques of majority-language words and signs that appear to have emerged in the sign languages only. Based on the multifunctionality of some word forms, the origin of the epistemic modal particles may be traced back to tags, interjections, and lexical signs, a route motivated by interaction and also found in unrelated spoken languages. Furthermore, in both sign languages, the first-person pronoun can be used, without a verb, as an epistemic “anchor” of a proposition, a construction that seems specific to languages in the gestural-visual modality. Another modality-specific feature is the possibility of transferring the expression of a marker of epistemic uncertainty from one articulator to another.

Variation in the numeral system of Japanese Sign Language and Taiwan Sign Language

Abstract (Japanese Sign Language) The numerals 10, 100 and 1,000 are expressed variably in Japanese Sign Language (JSL) and Taiwan Sign Language (TSL), two languages that also have historic links. JSL was used in deaf schools that were established in Taiwan during the Japanese colonial era, leaving a lasting impression on TSL, but complex sociolinguistic situations have led to different outcomes in each case (Fischer, 2014; Sagara, 2014). This comparative sociolinguistic analysis is based on two datasets comprising a total of 1,100 tokens produced by 72 signers from the Kanto and Kansai regions (for JSL) and the cities of Tainan and Taipei (for TSL). Mixed effects modelling reveals that social factors such as the age and region of the signer have a significant influence on how the variable is realised. This investigation shows how careful cross-linguistic comparison can shed light on variation within and between sign languages that have been in contact, and how regional variation in one language may influence regional variation in another.