In Sections I and II of this book, we considered general issues regarding

Variation between spoken and written academic
discourse: Multidimensional analysis
One key aspect of variation in academic discourse is between written 
and spoken modes. In this section, we see how variation in mode can be 
investigated using multidimensional analysis (MDA) (Biber, 1988). MDA 
is an approach which uses statistical techniques to identify clusters of co-
occurring linguistic features in large corpora. Texts which share patterns 
of linguistic co-occurrence are assumed to share common communicative 
functions and thus to belong to the same register or genre. MDA allows us 
to characterize the features of a set of texts and to compare it with any other 
set to determine how they vary.
Biber (2006) uses MDA to describe the characteristics of a range of discourse 
types that students encounter at university, including both academic genres 
like seminars and administrative ones like brochures. His findings show that 
there are consistent differences between spoken and written discourse, what-
ever its purpose. Thus all the written discourse is dense in information and 
uses elaborated reference, features often achieved through the use of noun 
phrases and nominalization (detailed consideration of this example) (see the 
section below for more detail on nominalization). Written language also uses 
an impersonal style, marked by the use of conjuncts (however), passive verb 
forms without by and subordinators (because). Spoken academic discourse 
tends to be characterized by features that show the personal involvement 
of the speaker, such as the use of contractions and first and second person 
pronouns (I’m, you’ve). It is also marked by situation-dependent references, 
which are those that can only be understood within the immediate context 
and include time and place adverbials like next and earlier.
These differences are found because the circumstances of production dif-
fer between spoken and written academic discourse. Written language is 
usually permanent and pre-planned, which enables the production of dense 
text and the use of elaborated reference; it is produced outside the presence 

---

94 Teaching and assessing EAP
of a relatively unspecified reader, which accounts for its impersonal style. 
Spoken language tends to be unrehearsed, short-lived and produced in real 
time, usually in the presence of listeners. This results in features of situation-
dependent reference and a more personal, involved style. However, the dis-
tinction between spoken and written discourse is not absolute; we can view 
texts as situated along a continuum of oral to literate production. For exam-
ple, spoken discourse that is monologic and pre-planned (e.g. lectures) is 
likely to show more features of written discourse than relatively unplanned 
discourse with more than one speaker (e.g. seminars).