This presentation aims to: (1) provide an overview of the text corpora of Italian Sign Language (LIS) collected at our laboratory, pointing out similarities/differences with existing corpora of spoken Italian; (2) illustrate the transcription and notation devices used to describe and analyze some of these corpora, and the theoretical concerns that have guided our notation choices.
The corpora collected at our laboratory include different types of monological and dialogic texts produced by LIS signers (mostly native, but also non-native), in a variety of formal and informal conditions. The corpora represent both adult and child language.
The adult corpora comprise:
The child language corpora comprise:
All the corpora were collected by videorecordings and are currently stored in videotapes (mostly: VHS format ). None of the corpora has been as yet transfered in a digitized format. We are at present evaluating different possibilities for transfering at least part of these data in a multimedia computerized format.
These text corpora are complemented by two distinct lexical corpora created for specific studies: one is stored only on videotape and lists 150 LIS signs corresponding to the items of the Peabody Picture Vocabulary Text (form B). The other comprises LIS and Spanish Sign Language (LSE) signs for food collected within the context of a crosslinguistic study recently undertaken by Mari Luz Perea Costa. This corpus includes LIS/LSE signs for a set of 144 food items, elicited (by means of picture stimuli) from a total of 48 Spanish signers and 30 Italian signers of different Spanish and Italian cities. This corpus, stored on videotapes, includes both the target signs requested from each subject, and additional data drawn from short interviews. The target signs have been transcribed in computerized format, using the sofware EXCEL: each sign is described in a Stokoe-based type of notation transposed in alphabetic text format.
The transcription and notation systems we have used until the present consist of annotated sign glosses with two recent, partial exceptions. These regard: a) the use of Ekman and Friesen’s (1978) Facial Action Coding System (FACS) for coding facial expressions; b) the use of SignFont for the transcription of manual signs in poetic and dramatized texts.
All the transcriptions are in handwritten form, and only a small portion of the transcribed data have been computerized (in a gloss format) for partial statistical analyses. In this presentation we conduct a critical examination of the notation devices we are using, trying to clarify to what extent they succeed or fail to show the relevant morphosyntactic regularities we have identified in LIS utterances and connected discourse.
First, we briefly examine the (im)possibility of using Stokoe-derived notations for the transcription of sign discourse, and point out both the main reasons why such notations are evidently inadequate, and the theoretical/methodological implications of this fact. At the same time, data drawn from the analysis of poetic and dramatized texts, to be also discussed in this presentation, indicate that some type of Stokoe-derived notation does appear to be useful for capturing relevant aspects of sign discourse.
We then consider two types of gloss-based transcriptions used for the analysis of narrative/descriptive texts and isolated utterances, and we compare them to those adopted by researchers working on other sign languages (e.g. Padden, 1981; Engberg Pedersen, 1993). The first type of notation focuses on the morphology of manuals signs, and its relationship to the use of non-manual markers for person and grammatical roles: the annotated glosses signal the manual signs’ place of articulation, but do not specify whether these are articulated with one or two hands, or what is the temporal relationship between the two hands when they are used as two distinct articulators. In the second type of notation, glosses are given in separate lines showing the role of each hand in sign production. A common feature of these two types of notation is that they signal the place of articulation of all manual signs (e.g. of nouns as well as of verb and pronominal signs), exlicitely distinguishing the signs articulated on the body from those articulated in neutral space or in marked positions in space. In our view, this notation permits to highlight relevant uninflective features of LIS morphology, and how these features coexhist with inflective patterns. We also describe our notation for the oral components of sign production. We then consider more recent and detailed analyses of the role and function of non manual markers, and of their relationship with manual signs, as performed by means of the FACS coding system.
Finally, we focus on some of the most relevant features of signs production within the context of poetic and dramatized texts, and on the transcription and notation problems they pose. The analysis of this type of texts in fact most clearly demonstrates the need for overcoming the constraints imposed by the use of sign glosses, due to: - a complex interaction between the "frozen" and the "productive" lexicon (and morphology); - the meaningful, seemingly morphemic role that the signs’ internal components (handshape, movement, place of articulation) appear to play; - the likewise meaningful use of planes (in addition to points or loci) for the articulation of signs in space. We examine to what extent the most significant features of this type of texts can be captured by sign writing systems such as SignFont.