5.6 Development 165More recently, López- Couso and Méndez- Naya ( 2014b ) have argued for
the grammaticalization of parenthetical (it) looks/ seems/ sounds/ appears like.
The development of these newer epistemic parentheticals represents layering.
There is decategorialization from a complement- taking predicate to a par-
enthetical clause (and perhaps, when it is deleted, to a “parenthetical quasi-
adverb”). The loss of tense- aspect- mood variability and the inability to take
adverbial modifi cation are indicative of morphological fi xation. The incorpor-
ation of the complementizer like represents an interesting case of fusion (52),
and the deletion of dummy it is loss of phonological substance (53). Finally,
the forms clearly serve subjective (epistemic/ evidential) and intersubjective
(hedging, politeness ) functions (54– 55).
In accordance with these studies, I believe that the diachronic evidence
offers support for considering the development of epistemic parentheticals as
a case of grammaticalization. They increase in morphological fi xation (gener-
ally as fi rst- person present- tense forms) and they undergo decategorialization
from a subject- full verb construction to a particle- like parenthetical. Thompson
and Mulac ( 1991 : 318, 324) note that the category status of epistemic paren-
theticals is diffi cult to determine, but propose that they are best considered a
subcategory of adverb. They argue as well that an epistemic parenthetical func-
tions as a “unitary epistemic morpheme” (315) or a “single element” (318),
which suggests a kind of coalescence , though morphological fusion does not
typically occur (except in the case of methinks ). Thompson and Mulac further
observe (324– 325) that epistemic parentheticals exhibit divergence or “form/
meaning asymmetry” in that while they are grammaticalized in certain con-
texts, in other contexts they continue to be used as ordinary subject- full verb
constructions, available for negation and questioning. Epistemic parentheticals
also adhere to Hopper ’s principle of persistence (1991: 28) in that the eviden-
tial meaning (indeed the core cognitive meaning) is preserved to some extent,
even in the grammaticalized epistemic expression. The original meaning
accounts for the varying modal strengths of the different fi rst- person epistemic
parentheticals, for example, with I guess expressing more tentativeness than I
think , since guessing is less certain than thinking (Thompson and Mulac 1991 :
325), and with I know more often expressing certainty rather than uncertainty,
since knowledge is more secure than belief. Semantically, epistemic paren-
theticals follow the directions of change observed in grammaticalization: they
shift from propositional or content meaning (denoting a cognitive act) to non-
propositional meaning (denoting opinion or belief) and they acquire subjective
of specialization and the fact that the productive impersonal disappears while the fossilized
methinks remains in use is an example of divergence; i.e., the grammaticalized item is left
out of changes affecting the lexical item. However, neither of these phenomena represent my
understanding of what Hopper ( 1991 ) means by “specialization” or “divergence.”