Advancing the corpus linguistics, looking specifically at tragedy, the emphasis is on banish and banishment because of the political context.  Exclude/exclusion are not encountered at all.  (Corpora obtained from ProjectGutenberg and run through TextSTAT).

Duchess of Malfi: banished x5; banishment x2. Inc. ANTONIO: ‘My banishment, feeding my melancholy…’

The White Devil: banish/ed 3; banishment 3.  Opening with Ludovico’s exclamation ‘Banished!’.

Although the term is not encountered in A Woman Killed with Kindness, banishment is effectively the fate of Anne (Nan).

Banishment was the sense invoked by urban authorities in their exclusion of their own members who had allegedly contravened expected norms.

So to other obvious observations.  OED has two strands for exclusion: one is tantamount to structural exclusion (i.e. not being admitted); the other to removal – more regularly banishment or ostracism.  The noun exclusion seems fairly rare in literature (I did a brief concordance analysis of Jonson, for example; OED has earliest occurrence in 14th century).  The first concept leads in many directions about rituals of inclusion and exclusion, qualifying status, and access to resources.  The second can be analyzed on various levels from Durkheim’s sacredness of society through functionalism to the latter’s insistence on reintegration and the restoration of harmony – or, more likely, enmity and low-level self-help, continuing ostracism and acrimony.

Here’s another thought prompted by work being conducted by Charmian Mansell.  It returns me to thinking about Social Network Analysis.  One precept which is being neglected now is Time, which used to be taken into account in the earlier literature (esp. in Social Forces).  Time is important for establishing social networks, esp. as support networks.  One might then a priori expect a difference in the capability of domestic servants and servants in husbandry to establish support networks in their place of occupation.  Domestic servants have the Time, but servants in husbandry do not (one-year contracts).  This thought is also apposite for some work which I am undertaking on bankrupts who migrated (London Gazette notices), for their failures disallowed them from making strong support networks and compelled them to move on or, quite often, circular migration back to their place of origin where they had kinship support networks.  What to do about Mark Granovetter’s ‘weak links’ is a constant challenge.

As this Warwick conference approaches, my thoughts return to social interactionism and symbolic interactionism, although the latter (Mead, Blumer) may be a stretch too far.  This consideration has been stimulated too by a recent event in my life, subjective as it may seem in the historical prospectus.  Recently, my college alma mater communicated that this year it would entertain 50th-anniversary recruits (matriculands, as it has it) for its annual dinner.  The insistence, however, on black tie is anathema to some of us (well, me).  I consequently declined to attend.  Is this self-exclusion?  I suggest not entirely.  In part, the culpability rests with the college which is inflexible in allowing lounge suits (those who wish could still retain black tie).  The exclusion is a consequence of interactionism or, as I re-read it, Goffman’s Behavior in Public Places (1963).  It approximates to the expected rules which are not legislative but to which people are expected to conform.  It is not quite a question of Goffman’s ‘situational proprieties’, which are matters of etiquette.  It would be interesting if the participants in their papers address this wider issue: ‘situational proprieties’ (how people who do not respond in the anticipated fashion in face-to-face (public) situations become excluded).