Relations that are seldom or never signaled

[Claudia Soria]

Dear Dagmar,

> Claudia, is the specific set-up and results for the experiment you mentioned, published somewhere? From what I found in my data so far, I have the feeling that the results of such experiments are very sensitive to the set-up, i.e. are the texts produced spontaneously or are they somehow prepared, even if only by having produced the story once before. What was produced first, the spoken or the written version? What other kind of markers did you find (cf., for instance, 'just' in ex.5)?

The experiment and results are published in the Proceedings of the Workshop "Discourse relations and Discourse markers", Coling-ACL 1998, Montreal, Canada. I agree that the results may have been influenced by the experiment set-up, as this is always the case. The specific aim of the experiment was to compare spoken and written language, reducing as much as possible the amount of variation induced by topic or genre. The cartoon story seemed to suit these requirements, providing a common topic which was neutral to the specific mode of expression. The texts were produced spontaneously, that is subjects were given the
story and immediately asked to tell it. After the spoken version was produced, then subjects had to produce a written version of the same story; thus in a sense there was a certain degree of preparation from the spoken to the written version. For what concerns other kind of markers used to signal coherence relations, I must have a look at the data; I'll keep you informed if you're interested.

Cheers,

Claudia

N.B.: What kind of data do you have?