5.6 Development 161Unlike the matrix clause hypothesis , this proposal does not involve a rever-
sal of syntactic hierarchy (main clause > parenthetical, subordinate clause >
main clause). It merely involves deletion of an adverbial connective in cases
where the logical connection of main and adverbial clause is contextually
inferable.
In the vast majority of cases, these adverbial clauses occur in medial or fi nal
position, except in cases such as (25c, d) in which they occur sentence initially
following a conjunction. This suggests that parentheticals in initial position
may indeed arise in the way that Thompson and Mulac suggest. However, it is
extremely diffi cult – especially in historical texts where the evidence is miss-
ing – to determine whether such initial constructions are main clauses with
that - deletion or parentheticals.
Support for the importance of adverbial constructions in the rise of epistemic
parentheticals is provided by López- Couso and Méndez- Naya ( 2014c ). They
fi nd that the impersonal adverbial parentheticals as/ so (it) (me) seems/ thinks
predate the “bare” parentheticals (it) (me) seems/ thinks and therefore point to
the derivation of the “bare” parentheticals from the adverbial parentheticals.^30
They also note that that- deletion with think becomes prevalent only in the fi f-
teenth century and with seem in the later sixteenth century (207– 208), well after
the rise of the parentheticals, thus going against Thompson and Mulac ’s claim
of that - less forms being the necessary “bridging” context: “the rise of the zero
complementizer ... could be seen as a consequence, rather than a cause of the
grammaticalization of the parenthetical it seems ” (López- Couso and Méndez-
Naya 2014c : 208– 209). Further support for the importance of the adverbial
source is provided in Brinton (2008: 228– 230, 235– 237), where I found that
for both I gather and I fi nd , the adverbial forms ( as I gather , as I fi nd ) predate
the “bare” parentheticals, by three centuries in the fi rst case. NP complements
rather than sentential complements are the norm in earlier periods. For I gather ,
the frequency of zero- complementizers in sentential complements is low and
shows no increase in frequency concomitant with the appearance of the bare
parentheticals. For I fi nd , both that- and zero- complementizers increase at a
slow rate over time.^31
30 They point to some support from structures showing wh- relative fronting (2014c: 209). Cf.
Brinton ( 2008 : 229– 230) which points to wh- fronting as a possible contributing force in the
development of I gather parentheticals, as well as the existence of sentential relatives of the
form which I gather from X.
31 Palander- Collin ’s evidence (1997: 384– 391) in regard to the development of methinks is some-
what less conclusive. She fi nds that fi rst- person subjects do seem to be associated with zero-
complementizers and parentheticals, which are both common in the EModE period, and that
the impersonal verb becomes increasingly restricted to the fi rst- person over the course of the
ME period. However, the earliest parentheticals she fi nds are not fi rst person. She concludes,
“So, if methinks is the phrase that was grammaticalized, this is not evidenced in the zero- that
and parenthetical use of [later ME] examples” (386). Palander- Collin ( 1996 ) does not commit