I think that's the wrong take. The more frequent utterances can be fossilized precisely because of their repeated use. The reality is that any creative or less frequent use of language (thus carrying more information) is carefully constructed in regular ways whose rules can be discovered.