Sign language annotation

Currently, annotating sign language generally consists of

  1. defining the set of annotation tracks for several or the entirety of articulators used in SL (chest, shoulders, arms, hands, head, look, cheeks, eyebrows, mouth)
  2. segmenting into temporal segments the production of units
  3. identifying the units by using categories that depend on the selected approach (annotation of form or of function) or the help of glosses. Generally, a supplementary track has a near translation in written French.

Issues with sign-language transcription: There are various specific systems for SL notation, of which the two main ones are HamNoSys and SignWriting. HamNoSys is a system that is intended to be phonetic and which allows for the description of the form of conventionalized signs (see gestural unity) according to five parameters (configuration, placement, movement, orientation, mimicry) It is unilinear and exists as a typographic font that can be used in the ELAN and iLex annotation software It is relatively efficient for the representation of isolated signs but does not permit the representation of speech in SL, which is characterized by important uses of space and simultaneous action in building meaning from bodily composition. SignWriting was conceived of as a way of writing in SL. It offers a multilinear representation and an analogic element. Importantly, for a decade it has been the tool of promising experimentation for the transcription of speech in Italian SL. Nevertheless, the non standardization of writing rules (spelling) on the one hand, and the difficulties in implementing it as software on the other, still limit its use for transcribing Sl corpora within annotation software. Thus, there is currently no way of annotating sign language.

Gloss-based notation: A term used in the field of linguistic research on sign language (SL) to designate practices consistent with the representation of speech in SL by written oral language (OL) (the oral language of a country with a SL and/or an international oral language).

The basis of the majority of corpus annotations up till now, this mode of representation is most frequently used right from the start of a track, aligned with the signal. Less standardized, variable from scholar to scholar or between countries, ways of using written OL are heterogeneous, and include the labeling of SL units by a single (sometimes two or three) words in the OL – which is referred to as a “gloss” – (ex. HORSE for the SL sign representing “horse”) to the use of numerous kinds of abbreviations carrying formal, categorical or morpho-syntactic information (ex: IX1, for a 1st person pronoun).

ID-Gloss: A form of annotation with glosses that is becoming standard. In the context of sign-language (SL) corpus annotation, ID-Gloss (Identifying Gloss, Cf. Johnston 2001) is a word from the annotating vocal language (selected by the annotator) used in a systematic and consistent way to label a sign (see gestural unit) in a corpus, an abstraction of semantic and contextual variations as well as morphological and morpho-syntactic variation of the sign. ID-Gloss is intended to represent the lemmatized form (lemma, or lexeme, according to the terminology used, or citation form) of the sign (type vs. token). Even if the choice of the VL word is motivated by its semantic proximity to the sign, ID-Gloss is not intended, in any way, to be a translation of that sign.
The establishment and designation of ID-Gloss requires the SL to be annotated according to lemmas (on the debates surrounding the procedure of lemmatisation and assignment of ID-Glosses, see Johnston 2008, 2011 and Konrad 2011).
The use of ID-Gloss in applied annotation on multimedia annotation software such as ELAN or iLex, which requires the association of a constant and evolving lexical data base, allows a regular and homogeneous annotation, permitting significant searches to be performed.
Contents validated by the Groupe de Travail 4 (multimodality and visual-gestural modality)