Guy Longworth at the TLS:
Austin thought that philosophical perplexities were generated by failing to attend to such distinctions and, in particular, by failing to acknowledge the importance of the illocutionary act. One example arises from attempts to treat stating something as a locutionary rather than an illocutionary act, and so as something that is achieved just by uttering a meaningful sentence. The apparent variety of things we do in speaking would then be treated as mere differences in perlocutionary effects that arise from stating something in different circumstances. Thus, for example, uttering the sentence, “I promise that I’ll be home for dinner”, would be treated as a way of stating something about oneself, rather than as a way of promising. Promising itself would be a perlocutionary effect of making such a statement about oneself. It would consequently be perplexing to try to understand what precisely one was stating about oneself, and how stating it could bring about promising. Furthermore, the locutionary treatment makes it too easy to state something, requiring only that one utter a meaningful sentence; whereas, in fact, stating, like promising, is dependent on propitious circumstances for its successful performance. The treatment leads to puzzlement about what was stated by the utterance of some meaningful sentence in cases in which absence of the required circumstances means that no such stating occurred.
more here.