General Description

The Occitan Map Task dialogue corpus began in 2007 with an objective to collect the diverse interrogative intonational data in Occitan. The corpus is made up of 6 dialogues (aprox. 1 hour of recordings) orthographically transcribed. The dialogues have been collected from a number of different towns within the Occitan linguistic territory (dialectal map) and cover different regional dialects (for tehe moment, languedocian, limousine and gascon).

This project is part of a larger international research initiative that aims to collect dialogue data from different languages following the Map Task standard methodology (HCRC Map Task Corpus Project). Other languages that are extensively documented in the Map Task Corpus Project include English, Swedish, Dutch, Italian, Japanese, and Portuguese.

language	dialogues/hours of recordings	year	participants	age	sex/others
american english	16 dialogues	1999	8	adults	women
australian english	19 hours		204	adults and young people	men and women
brittish english	36 hours		12-14 for each variety	16	men and women
dutch	8 conversations	1999	4	students
italian	44 dialogues, 350 minuts			young adults, children
japanese	128 dialogues,: 22 hours	1994		students
portuguese	64 dialogues		32 / 8 groups of four	students	men and women
swedish	50 minuts, 8000 words		4	30-40	1 man, 3 women
catalan	6 hours aprox.	2005	30	20-45	women; friends
occitan	1 hour aprox.	2007	30	20-90	men and women; friends or unknown

Map Task methodology

The Map Task is a collaborative task between subject pairs. It is designed to elicit the production of certain linguistic phenomena and at the same time permits control of context effects (sonorant segments, information structure, prosodic word structure). Each participant in the pair has a map of an imaginary place defined with drawings. One of the speakers has a map with a designated path from a beginning point to an end (map_giver) and has the role of giving instructions of how to arrive at the proper destination, the instruction-giver. In return, the other speaker has a different version of the same map but with certain differences with respect to the other version, also, this person does not have the path to the final destination laid out (map_follower). The point is to obtain the correct path from the other speaker through a series of questions that, with the help of the responses that they receive, they can reproduce the same path on their map.

The names of the places on the map vary with the intension to produce sonorant consonants and to have the rightmost word have penultimate stress. Unlike the maps that were used for the Catalan corpus, it has not been possible to use written place names, because of the problem caused by the poor socialization of written Occitan and the following illiteracy generalized within the native speakers' community. That's why drawing were used.

This method of elicitation permits: a) an ample collection of intonational contours in controlled conditions and, at the same time, with a natural production; b) a good interaction between participants; c) the appearance of dialectal variation that in a more controlled context would not surface. 

From each location it is possible to hear the Map Task recording in audio format and see the corresponding orthographic transcription (transcription criteria).

 

Speakers

As often as possible, speakers were pairs of native speakers that already knew each other, therefore allowing a level of confidence between the two. At least one of the speakers had to be from the location in question and, in that case, was given the role of instruction-follower.

With the intention to keep the contexts as similar as possible across locations, the speakers were given the following instructions:

1. The task consists of trying to reproduce the exact route that the instruction-giver has on her map.  In this way, if you would like to compare the two routes at the end (the original and the reproduction), they will be the same, from the starting pointing point to the end point, with the same inclinations, etc.
2. The instruction-giver does not need to give more information than for what the instruction-follower has asked. 
3. The two maps may not be exactly the same; it is possible that there exists some type of error.  Even if this is the case, the route must continue to be the same between the two, passing the same locations.
4. The two participants must have the same start and finish points.
5. At the end of the task the participants may quickly confirm that the route is correct (through concrete questions, which points the route passes, etc.). 

 

Recordings

The recordings vary in length according to each set of participants and were always conducted at one speaker's home in a quiet setting using an audio format. The speakers sat facing each other with an object between them impeding their view of the other participant’s map, but still being able to have visual contact. For the recordings, a Marantz professional recorder was used (model PMD660), and two microphones, AKG C 1000 and Rode NTG 2.

All of the data, in different formats, is available online for free through the web for academic and research use. It is possible to listen to the archive in mp3 format while following the orthographic transcription, or also, it is possible to download the archive in .wav format.

 

Transcription

The complete corpus is orthographically transcribed (transcription criteria).