There is a kind of formulaic sequence pervasive in most genres of texts which has not been properly recognised and studied. They are not fixed in form and are usually lengthier than the relatively well studied two-word collocations and more irregular than fixed idioms. They appear in text as a package with a core collocation and some accompanying semantic, syntactic, or pragmatic features that influence the choices of words around the core. Stubbs (2002) proposed the term extended lexical unit for this kind of structure. In this article, the term extended collocation is used instead to better reflect its phraseological nature. It is argued that only the web itself is large enough to provide adequate instances for the investigation of extended collocations, and therefore the integration of phraseology into TESOL research and practice. Arguably the most difficult aspect of studying any formulaic sequences is the initial identification and validation of their phraseological status. This article proposes a reliable method of identifying extended collocations and other phraseological units via Google search. This method will be useful not only for research of phraseology but also for the teaching of English phraseology to speakers of other languages.