I am also attaching an Excel spreadsheet with data from COCA on particular verbs and expressions used by writers in the differing disciplinary domains to describe their knowledge-making activities, both as personal and impersonal actions, and including various kinds of reference to the document itself ("this paper" etc.). With this data, one could give a fine-grained analysis of the sweeping generalization that academic writing is impersonal.
There also then follow sets of terms that are pretty good markers for each of the 9 domains and a set that occur smoothly in all domains—ie.e., are 'academic' as such.
The figures are the COCA frequencies per million words.