Abstract: This paper re-visits the use of the corpus evidence in the US supreme court in the case FCC v. AT&T, from the perspective of the corpus driven approach, and focuses our discussion on the use of etymology, dictionaries, lemma and collocation analysis attending in particular to their data retrieval and procedures. The original result is also put to verification using different corpora. The paper cautions that in using corpora it is “easy to say there is something there, but hard to say there is not”, and that the corpus use as a powerful method and tool in expanding application areas does not warrant the reliability of the claims only from using corpora. The keystone is not whether one uses corpora, but how.