288 Concluding Remarks: Pathways of Change
original main clause is lowered to become parenthetical and the original com-
plement clause assumes the role of the main clause. The change is dependent
upon the frequency of occurrence of think and guess with fi rst- person subjects
and complement clauses without that. Although intended to describe only the
synchronic source of these parentheticals, Thompson and Mulac ’s proposal
has been used to account for the diachronic origins of a number of paren-
theticals, including methinks / I think , pray (you/ thee)/ prithee , I’m afraid , I’m
sorry , I promise , and you know/ see/ say. Brinton ( 2006 ) explores this hypoth-
esis in detail in respect to the diachronic development of a large number of
comment clauses, fi nding that in the majority of cases the rarity of the source
construction – i.e., the matrix clause with that- full complement and later the
matrix clause with that- less complement – at the time at which the parenthet-
ical develops (or at any time) argues against the Thompson and Mulac pro-
posal. The fact that deletion of that does not proceed regularly in the history of
English, as well as more recent work suggesting that the presence or absence
of that does not distinguish between main- clause and comment- clause uses of
initial clauses both serve to undercut the validity of this hypothesis (see Section
1.4.1.2 for further discussion).
Chapter 5 is an extended examination of the history of epistemic parentheti-
cals in English. Epistemic parentheticals built on a wide variety of know - verbs,
verbs of cognition and verbs of direct perception ( believe , deem , guess , know ,
leve , suppose , think , trow , trust , understand , undertake , be aware , wene , woot ,
and woot , and impersonal seem and think ), are numerous and fully developed
in Middle English. They serve, as they do in Present- day English, as subjec-
tive epistemics and evidentials , with intersubjective and politeness functions.
Given their well- formedness in Middle English, the matrix clause hypothesis
would suggest that we should be able to trace them back to main clauses with
that - complements in an earlier stage of the language. However, the marking
of epistemic modality, especially low- probability evaluation, in Old English is
extremely limited; apart from a small class of truth- intensifying adverbs and
a handful of periphrastic expressions ( wen is þæt ‘it is doubtless’ is the most
common), only four forms seem to function potentially as epistemic parentheti-
cals: ic wene (þæt) ‘I suppose, believe that,’ ic teohhi(g)e (þæt) ‘I presume,
think that,’ ic hopige (þæt) ‘I suppose,’ and me þincð (þæt) ‘I think.’ While
Old English has a fairly substantial class of know - verbs taking that - comple-
ments, deletion of that is rare, and even where it occurs, the form of the attached
clause often indicates its continued syntactic dependency. Furthermore, even
in Middle English, we do not fi nd a high frequency of fi rst- person subjects
and that - deletion with know - verbs, a situation upon which the matrix clause
hypothesis depends. In fact, the rates of that - deletion continue to be low
throughout the Middle English period. Thus, the existence of potential sources
for Middle English epistemic parentheticals in Old English – as dictated by