Hello everybody,
I'm looking at the recognition of words in spoken French by Danish learners (n= 48) and French native speakers (n= 12).
I have a linear mixed effect model with one dependent variable (reactiontime in ms) and the following fixed effects: sentence context, target phoneme, position of target phoneme in word, voice/unvoiced target phoneme and group of subjects. I also want to include two random effects in the model: person and word.
As far as I have understood from the S-news archives, if I don't want nested random effects but crossed random effects (and I think crossed effects is what I want), I have to use the pdBlocked(list(pdIdent(~person-1),pdIdent(~ord-1))) formula.
It seems to work fine when I include all my subjects and items in the model, but when I want to work on a subset of subjects (the Danish learners) and/or items (French words vs. non-words), apparently something goes wrong. Even though I choose a different dataset, one that doesn't include any native speakers or any non-words, the model still includes all of the subjects and all the items in the random formula.
Can anyone tell me what to do to avoid this?
Thank you.
Lise Hedevang
Ph.d.-studerende
Institut for Sprog, Litteratur og Kultur, Fransk
Aarhus Universitet
Jens Chr. Skous Vej 5, bygn. 1461, 425
DK - 8000 Århus C
(+ 45) 89 42 64 34
romlh@hum.au.dk
http://person.au.dk/romlh@hum
|
|