I am currently designing an acceptability judgment experiment that contains one variable with two levels. The experiment will be a within-subject design. There will be multiple sets of target sentences. Each set contains two sentences (each sentence is a condition) that are minimal pairs for informants to judge. Do I need to perform a latin square to prevent the informants from seeing both of the two sentences of a minimal pair? For example, suppose that I have 8 sets, then each informant will see only 4 sentences for each condition instead of all of the 8 sentences if the sets of sentences are counterbalanced.
The materials and the literature that I read tend to suggest that if the experiment is a 2x2 design with 2 within-subject variables and different item sets, the trials in each set of sentences need to be counterbalanced using Latin Square. For example, if there are 8 sets of sentences, and each set contains four sentences for the four conditions, after a Latin Square treatment, each participant will only see two sentences for each condition. So I am not sure whether it is necessary to do so if I only have one variable with two levels.
Thanks in advance for any comments.