What prosody measures should I include? I am thinking of assessing intersentential pause duration; pausal intrusion duration; number of pausal intrusions; pitch (F0), declination; and intonation contour. Any views would be appreciated!
You might take a look at the work by Jared Bernstein and Jennifer Balogh. They approach it from automated techniques for measuring reading fluency, which sounds what like you are looking for. For example, see: http://www.ghosh.org/jennifer/ and http://www.versanttest.com/technology/papers.jsp for links to a bunch of papers
and http://epm.sagepub.com/content/72/3/435.abstract and http://aclweb.org/anthology//W/W11/W11-1406.pdf for links to relevant publications.
I do not investigate pauses in reading but rather in writing. However, one way for assessing fluency from pauses may consist in calculating the ratio between total pausing time and total reading time, or by the number of pauses in a particular syntactic or semantic structure (e.g., the number of pauses per sentences, per clause...).
One other possibility is to find fluent and hesitating periods, as in writing and speaking. For this you can plot on the y axis the cumulated number of words read and on the x axis time. Fluent periods should be evidenced by rapidly growing sections of the curve, while hesitation periods are growing less rapidly.
Thanks Thierry. I think that plotting the data as you describe may give extremely useful information. But since you mentioned spelling, I am interesting in learning exactly how you measure pauses during handwriting. I would really appreciated if you could recommend some reading material on that. You are using specific hardware and software? Are you taking video during the procedure?
With handwriting, we record pauses while writers are writing on a digitizing tablet with the 'Eye and Pen' program, which has been developed in my team here in Poitiers, France, by Chesnet and Alamargot.
Check the website of the program (www.eyeandpen.net) where you will find useful information.
Else, check http://link.springer.com/article/10.3758/BF03192780 for the paper describing the program.
Alternatively, there is also the Ductus program but it is more focused on low-level analyses of handwriting (strokes, velocity peaks, etc...)
With typing, we usually use Inputlog (www.inputlog.net).
Thanks a lot Thierry! That is really useful. I will read the material thoroughly! The software is available for buying? May I contact you in case I need more information?
It is best you directly contact the authors of the programs. I know them personally and I am sure their will help you if you encounter some problems for using their programs. But if you have questions not related to the programs, do not hesitate to ask.