Constraints upon the structure of possible languages are presumably associated with the machinery available within the person to process utterances. This machinery may reveal its nature through observations that are not themselves meaningful linguistically. That is, two sentences may both be grammatical, and may be logically equivalent, and yet may differ in the speed with which people understand them, or the accuracy with which they are remembered. Such findings then shed light on the biological constraints that limit the possible forms of communication available. Two particular examples can be taken from studies of working memory and of identification of single isolated words. Both fields reveal important differences in the mechanisms for processing heard and read communication; they also suggest constraints both of a syntactic and of a semantic kind.