AbstractResults yield by conversational data are compared with those generated by elicited grammaticality judgments on the issues of topic, focus, and word order. On the one hand, most of the sentence types produced by elicited grammaticality judgments are confirmed by empirical conversational data. On the other, research utilizing grammaticality judgments detects only prototypical constructions. The cause is that invented sentences, upon which grammaticality judgments are based, are cognitively biased to be prototypical. Therefore, elicitation methodology does not provide the analyst with the whole range of possible constructions. This type of data is simplified in the sense that it consists mainly of prototypical instances placed in a context of exemplification. Conversational data, on the other hand, include the human factor, conversational and pragmatic factors, as well as the real context where a particular utterance occurs. For this reason, it is argued that syntax studies based on conversational data allow for the possibility of finding new unexpected cases that may offer new perspectives.