There are a surprisingly large number of theories about the role and purpose of citations in academic texts.
Given the indebtedness function it is important that you are exhaustive. If you make specific use of a source, finding information there which have not found elsewhere, then a reference is essential. Given the invitation function it is important that you give references which are absolutely precise. There is no sense in telling your reader what a certain book is called or telling him on which page a certain quotation is to be found if the book is in fact called something else or if the quotation is 20 pages later.
Citations are not just used when you are quoting what someone else has said. Suppose you make a general statement about theories being developed and claims being made in an academic discipline, without in the first instance specifying any particular author. However, as soon as you say that such and such is a common belief, or that such and such is frequently claimed, then you commit yourself to providing your reader with an example, and preferably more than one. You can do this by formulating a reference as follows:
Sociologists have shown that communication systems are heterogeneous and multi-layered (cf. Labov 1972; Hymes 1973).
It has become apparent that the relation of life events to symptoms and health status is not strong (see for example Rabkin and Struening 1976).
Whether you integrate the reference into the text or supply it separately in a note depends on which mode you are writing in. Legal publications in Belgium, for instance, never have in-text citations. Those style sheets which are important in the field of both linguistics and literature (the MLA, stylesheet for linguistics, and the likes) consider in-text citations the standard.