To appear in: Acta Linguistica Academiae Scientiarum Hungaricae, 1991

Inferential Markers, which sig­nal that the current utterance conveys a message which is, in some sense, conse­quential to some aspect of the foregoing. Examples are presented in (18):
(18) Inferential Discourse Markers: accordingly, as a conse­quence, as a result, con­sequently, hence, in this/that case, of course, so, then, there­fore, thus
Contrary to the other three groups, there is no obvious sub-grouping, although there are subtle differences. For exam­ple:
(19) a) John is remaining. So (?in that case) I am leaving.

b) I don't want to talk with you. But I will. Thus (?as a result) sit down

Final Remarks
The foregoing is intended to be a preliminary ex­am­ination into distinctions within English discourse markers. Based on my analy­sis, there appear to be some definite types: topic markers; dis­course activity markers; and mes­sage re­lation­ship markers. Within these, there are various, more spe­cif­ic sub­divi­sions, some clear­ly more prob­lematic than others.
I have, in addition, touched upon several areas where immed­i­ate future research is required. The notion of a sequential dis­course re­­lationship requires clarification, as do the details a core mean­­ing for each individual discourse markers. One might also consider whe­ther or not discourse markers should be viewed as a grammati­cal cate­gory, a part of the grammar of a language, as I have pro­posed, or as a part of a larger group of interactive mark­ers such as "Then there was the one about...," which in­tro­duc­es a joke, or "Let us pray," which introduces a moment of prayer.
Remaining for future research, also, is an adequate account of the interrelation­ship between performative sentences and the pre­sence of discourse markers. For example, we find "More to the point, I should have done it" conveying a speaker admission while "More to the point, I admit that I should have done it" and "I ad­mit that more to the point, I should have done it" are ques­tion­able. Not unre­lated to this problem is that of the se­quenc­ing of discourse mark­ers--which can co-occur and in what order. For ex­ample, we find sentences such as "And so you agree" but not "So and you agree." And finally, there is the question of the occur­rence of various discourse markers with syntactic structures and message types. We do not find, for example, "By way of explana­tion, who are you?," while "By way of introduction, who are you?" is fine. I leave these prob­lems for another time.