How a story is performed: traditional storytelling in the hands of computing

General description
The main idea is to find out how to edit a collection of oral transmitted texts of literary interest (often called orature), in the future resulting as a marked-up database, serving as a model to future researchers and projects. The main goal is to analyse both the transcribed texts and the way they are performed orally. Until today, several research areas focus on particular parts of this kind of enunciation – Literary studies concentrate on the transcription, Anthropologic studies concentrate on the artistic performance or on the interactants’ behavior, etc. But nowadays we have the means that allow us to study a more complete contextualization of an oral composition, and interaction including voice characteristics (intonation, prosody), body movements (face, torso, arms, hands) and proxemic behaviour (the relation between an interviewee and the interviewer). My presentation will focus on the main parts of my PhD project: a) the creation of a representative corpus of Portuguese oral tradition with performative and literary annotation, b) digital publication of a more complete edition model, and c) gesture and literary classification.

Previous Studies
From previous incipient work done on oral prayers themed on the Passion of the Christ, it proved to be productive to assume that performance integrates verbal art in its many strands. Performative elements have given some hints on their importance, having this previous study focus on the intonational contrast between the informant ́s enunciation of a prayer and in spontaneous speech.

Fig. 1: Image A – spontaneous speech

Fig. 2: Image B – enunciation of prayer a)

Fig. 3: Image C – enunciation of prayer b)

In image A ascending and descending melodic variations contours can me seen, characteristic of human speech. In images B and C, F0, paradoxically, is almost flat, indicating a type of monotonous speech, showing a lowering of the medium tone of voice of the speaker in question. Moreover, the total duration of each one of the extracts, the informant produces a greater amount of words during prayer enunciation than in conversation mode in which, as indicated in A, disfluencies or pauses occur, which rarely occur during prayer ennunciation. These characteristics are due to the fact that, among other reasons, discourse planning is slower during conversation that during prayer enunciation (being a discourse cited by memory, not involving simultaneous processing to the production).

Concluding, the above-explained study shows these to be performative marks to the production of a prayer orally and it gives us hints to question whether other types of orally transmitted texts have them or not.

Data collection and tools used
The data is being collected through fieldwork interviews and archival research, which result in the primary database in video and audio formats that share a common structure for storage. Each entry is accompanied by metadata related to the context of collection: informant’s data, place and date, technical information about support files. The data is worked to ensure time alignment of the different types of representation: transcription of the text, kinesic and verbal (not literary) markup. To this end, the programs being used until now are 'EXMARaLDA' – for the alignment of text and video annotation –, and 'Praat' for intonation and prosodic analysis of the compositions. Together, these elements constitute the support for the database analysis: literary classification according to the taxonomic reference indexes, annotation of text versions and variants, motifs (using AMICUS network labelling), UNESCO intangible heritage taxonomy and Matriz PCI, formal characterization of the text (composition, data contamination or counterfeiting, domain – religious, artistic, etc. – fictionality degree), linked existing Wikipedia entries for each type of text, and literary and compositional context. This way, the database will gather several interpretations on the same composition, letting the user/researcher work with the information that suits their interests.