Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

Language Sciences 87 (2021) 101400

Contents lists available at ScienceDirect

Language Sciences
journal homepage: www.elsevier.com/locate/langsci

From clause to discourse marker: on the development of


comment clauses
Bernd Heine a, Gunther Kaltenböck b, *
a
University of Cologne, Nonnenwerthstr. 48, 50937 Köln, Germany
b
University of Graz, Heinrichstrasse 36, 8010 Graz, Austria

a r t i c l e i n f o a b s t r a c t

Article history: This paper proposes an alternative account for the development of comment clauses (e.g. I
Available online xxx think, I mean, I admit), which differs from previous accounts centring on grammaticali-
zation, pragmaticalization, or lexicalization. Based on the framework of Discourse
Keywords: Grammar and building on Heine et al. (2021a, b), it is argued that their development in-
Comment clause volves a stage of cooptation, whereby a text piece is transferred from the sentence level to
Discourse marker
the metatextual level of discourse processing thus acquiring new grammatical properties,
Grammaticalization
viz. independence from the host clause in terms of meaning, syntax and prosody, meta-
Cooptation
Pragmaticalization
textual function, and positional freedom. All these changes are hard to reconcile with
Complement-taking predicate grammaticalization. At the same time, however, grammaticalization does play a role in the
process once cooptation has taken place, affecting mainly the internal form of coopted
expressions.
Ó 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC
BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction

The origin and development of comment clauses (Quirk et al., 1985: 1112f), such as the ones in (1) and (2), has been the
subject of considerable discussion in recent decades, with different source constructions and diachronic pathways having
been proposed.

(1) Paul, I am getting a hit to 10 Senate Democrats, which I admit is a little bit like a solar eclipse. (COCA2017_SPOK_Fox_Journal Editorial Report)1
(2) Well, I mean, the prosecution was disappointed in the outcome in this case. (COCA2017_SPOK_NPR_Morning)

The crucial question to be answered is how a clausal (lexical) unit, a so-called complement-taking predicate (CTP) clause
(Thompson 2002), eventually ends up being used as a syntactically independent and positionally flexible discourse marker,
that is, as a comment clause.2

* Corresponding author.
E-mail addresses: bernd.heine@uni-koln.de (B. Heine), gunther.kaltenboeck@uni-graz.at (G. Kaltenböck).
1
COCA refers to the Corpus of Contemporary American English (Davies 2008).
2
Discourse markers are defined as follows: Discourse markers are (a) invariable expressions which are (b) syntactically independent from their envi-
ronment, (c) typically set off prosodically from the rest of the utterance, and (d) their function is metatextual, relating a text to the situation of discourse,
that is, to the organization of texts, the attitudes of the speaker, and/or speaker–hearer interaction (Heine et al., 2021a, (1)). Comment clauses are a subclass
of discourse markers. According to Brinton (2008: 241), comment clauses differ from some other parentheticals and discourse markers (‘pragmatic
markers’) in their “clausal origin”.

https://doi.org/10.1016/j.langsci.2021.101400
0388-0001/Ó 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/
4.0/).
2 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

This paper offers an alternative account to the existing proposals in terms of grammaticalization, pragmaticalization, or
lexicalization. By drawing on historical data from existing studies, it argues that the development of comment clauses in-
volves an operation called cooptation, whereby pieces of a text are deployed for use on the metatextual level (cf. Heine 2013;
2018; Heine et al., 2017). Cooptation essentially lifts elements out of Sentence Grammar thereby creating a parenthetical,
extra-clausal comment clause, which subsequently is subject to grammaticalization. Although the focus here is on comment
clauses in English, it is assumed that the account proposed holds for other languages as well.
The paper is structured as follows. After having presented a brief outline of the paper in this section, Section 2 introduces
the topic of the paper, namely the syntactic category of comment clauses. In Section 3, the reader is familiarized with previous
research on the subject matter of the paper by providing an overview of the different processes and pathways that have been
proposed to explain the emergence of comment clauses, namely grammaticalization, pragmaticalization, and lexicalization.
Section 4 then draws attention to problems associated with previous research, highlighting the challenges involved in ac-
counting for the development of comment clauses particularly in terms of a process of grammaticalization. In order to deal
with these problems, Section 5 offers an alternative account based on the framework of Discourse Grammar. Section 5.1 first
outlines the basic assumptions of this essentially dualistic framework and Section 5.2 then discusses cooptation as the main
mechanism involved in the development of comment clauses, and its relationship to grammaticalization. Section 5.3 applies
the proposal to a test case: I admit. Section 6, finally, explores the ambiguous status of initial comment clauses and how they
can be accounted for in a dualistic framework. Section 7 offers a brief conclusion.

2. The category: comment clauses

The term “comment clause” (Quirk et al., 1985: 1112–1118) or “reduced parenthetical clause” (Schneider 2007) is used to
refer to parenthetical disjuncts with clausal structure, which may occur initially, finally, or medially with respect to a host
clause, as illustrated in (3)–(5). These structures have also been referred to as “parenthetical verbs” (Urmson 1952: 484),
“parentheticals” or “disjuncts” (Espinal 1991), and “discourse markers” (Biber et al., 1999: 1086), and are characteristic of
spoken language.

(3) There were no other applicants, I believe, for that job.


(4) There were no other applicants for that job, I believe.
(5) I believe, there were no other applicants for that job.

Six structural types of comment clauses can be distinguished, namely the following:

Type (i): They take the form of a main clause,


Type (ii): They take the form of an adverbial finite clause (as you know),
Type (iii): They take a nominal relative clause (what is more upsetting),
Type (iv): They take a to-infinitive (to be honest),
Type (v): They take an -ing clause (speaking as a layman), and
Type (vi): They take an -ed clause (stated bluntly).

The most important, according to Quirk et al. (1985: 1113), is type (i), and it is also the one focussed on in this paper. In this
type, the comment clause is “like the matrix clause of a main clause” (Quirk et al., 1985: 1112), as in (3)–(5). Structurally, this
type “generally contain[s] a transitive verb or adjective which elsewhere requires a nominal that-clause as object” (Quirk
et al., 1985: 1113). Typical examples are I believe, I guess, I think, I expect, I feel, I assume, I understand, I’m glad to say etc.
Semantically, comment clauses are “metatextual” (e.g., Traugott 2018: 27) or “metacommunicative” (Peltola 1983: 103),
expressing speaker “stance” (Biber et al., 1999: 972), with type (i) comment clauses having been classified as expressing (a)
the speaker’s tentativeness over the truth value of the matrix clause (e.g. I guess), (b) the speaker’s certainty (e.g. I know), (c)
the speaker’s emotional attitude (e.g. I’m glad to say), or (d) claiming the hearer’s attention (e.g. you see) (Quirk et al., 1985:
1114–5).
Although the examples of comment clauses discussed in the literature are all rather formulaic or “stereotyped” Quirk
et al. (1985: 1114) also acknowledge the existence of “freely constructed” ones, which permit “variations of subject,
tense, and aspect, or additions of adjuncts, etc.” As an example of such a non-formulaic comment clause they give the
following:

(6) The Indian railways (my uncle was telling me some time ago) have always made a profit.

Formulaic comment clauses have also been associated with so-called “complement-taking predicates” (or CTPs), a term
originally introduced by Noonan (1985) and later used by Thompson (2002) to refer to epistemic/evidential/evaluative
fragments introducing a finite clause. More specifically, Boye and Harder (2007) discuss comment clauses such as I think
under the heading of grammatical CTPs, which are seen as derived from lexical CTPs in clause-initial position (see Section 3.1).
There is indeed a close relationship between CTP-clauses and comment clauses, although Quirk et al. (1985: 1114) are not so
clear about the type of emotional predicates they include in the category, citing only those that are followed by a to-infinitival
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 3

verb of speaking (e.g. I’m glad to say, I’m happy to tell you).3 Thompson, by comparison, fully embraces emotive/evaluative
expressions such as (don’t) give a shit, (be) glad, (be) (a) good (thing), like it, (doesn’t) matter, (be) too bad, (be) a shame.
Conversely, Thompson (2002: 151–2) focuses particularly on the formulaic nature of CTPs, while Quirk et al. more readily
admit non-formulaic expressions (see above).

3. The proposals: explaining the emergence of comment clauses

The close correspondence between clause-initial CTPs (or matrix clauses) and comment clauses has been noted not only
on a synchronic level, but also diachronically, with the former having been identified as a possible source for the latter.
Various pathways and processes have been proposed for different comment clauses, which will be briefly reviewed in the
following.

3.1. Grammaticalization

Grammaticalization is by far the most frequently invoked process in the development of comment clauses (e.g. Thompson
and Mulac 1991; López Couso 1996; Brinton 1996; 2008; 2017; Palander-Collin 1999; Wischer 2000; Traugott and Dasher
2002; Akimoto 2002; Kärkkäinen 2003; Brinton and Traugott 2005; Van Bogaert 2009; 2011; Boye and Harder 2007;
Diewald 2011a; 2011b; see also López-Couso and Méndez-Naya 2014) with the parameter of decategorialization being most
often cited in support of this hypothesis (e.g. Brinton 2017: 165).
The most widely discussed pathway is the one suggested by Thompson and Mulac (1991: 313–329), who argue that
epistemic parentheticals like I guess, I think, or I suppose arise as matrix clauses with dependent that-complement clauses.
Deletion of the that-complementizer then creates an indeterminate structure where the original matrix clause is reanalyzed
as a parenthetical and moved to medial or final position of a clause (see also Palander-Collin 1999). The different stages of this
hypothesis, referred to by Brinton (2006) as ʽthe matrix clause hypothesisʼ, are illustrated in (7).

(7) Thompson and Mulac’s (1991) matrix clause hypothesis (see Brinton 2017: 157)
a Stage 1: I think that the world is flat.
b Stage 2: I think Ø the world is flat.
c Stage 3: The world is flat, I think.

This hypothesis, which is based on synchronic data only, has also met with criticism, particularly the claim of an inter-
mediate stage of that deletion. In particular, it has been argued that there is no compelling historical evidence to show that
there necessarily was a complementizer (that) which was subsequently deleted (e.g. Aijmer 1997: 6; Rissanen 2008; Brinton
(1996: 239–253, 2008: 44–46, 133–54, 2017: 155–162, Fischer 2007a: 105).
An alternative hypothesis, based on a detailed analysis of diachronic data, is proposed by Brinton (1996: 239–254), who
argues that first-person epistemic comment clauses such as I think, I guess, I suppose have originated not in a matrix clause but
an adjoined adverbial/relative structure of the type as I think, as illustrated in (8).

(8) Development of first person epistemic comment clauses (Brinton 1996: 252)
Stage I: They are poisonous. That I think.
Stage II: They are poisonous, {that I think, I think that/it, as/so I think}.¼‘which I think’
Stage III: They are poisonous, I think. OR
They are poisonous, as I think.¼‘as far as I think, probably’
Stage IV: I think, they are poisonous. They are, I think, poisonous.

Analyzing English I mean, Brinton (2008: 127) presents a similar five-stage trajectory. Unlike the matrix clause hypothesis,
Brinton’s scenario does not involve a ʽreversalʼ of syntactic hierarchy (main clause > parenthetical, subordinate clause > main
clause), but “deletion of an adverbial connective in cases where the logical connection of main and adverbial clause is
contextually inferable” (Brinton 2017: 160) (for a critical review see Fischer 2007a).
Brinton, however, does not discard the matrix clause hypothesis altogether, observing that it accounts for some comment
clauses but not for others. For instance, she finds support for the matrix clause hypothesis in the case of I/you (modal) admit, I
say, I see, and I’m just saying (Brinton 2008, 2006, 2017: 289).4
Another grammaticalization approach is that by Boye and Harder (2007), who offer a usage-based take on the devel-
opment of comment clauses such as I think by explicitly linking the notion of structural subordination with that of usage, i.e.

3
Examples aside, they do however mention expressing “the speaker’s emotional attitude” as one semantic characteristic of comment clauses and are
even prepared to include interjections such as God knows and Heaven forbid on account of their emotive attitude (Quirk et al., 1985: 1114–5).
4
Moreover, Brinton (2017: 18) notes that the following could also be covered by the matrix clause hypothesis: Parenthetical second person forms (e.g.
you say/ see/ know), impersonal constructions (e.g. it seems/ appears/ is said/ is rumored, that/ it is to wit), and third-person forms (e.g., one hears, there’s no
doubt, God knows).
4 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

primary vs. secondary discourse status.5 They account for the change from fully lexical clause (matrix clause)6 to a gram-
matical element (comment clause) by means of a scale of three categories for CTPs: (1) primary lexical CTPs, (2) secondary
lexical CTPs, and (3) secondary grammatical CTPs. The classification incorporates both the structural and the usage status,
each of which is described by a binary set of values: lexical vs. grammatical structural status and primary vs. secondary usage
status. The three categories are seen as different stages in the diachronic development of CTPs, as illustrated in (9).

(9) Developmental pathway of CTPs (Boye and Harder 2007: 590)


A: lexical and primary CTP
Y usage reanalysis
B: lexical but secondary CTP
Y structure reanalysis, grammaticalization of CTP
C: grammatical and secondary CTP

While the first stage is easily identifiable as matrix clause and the last stage as (clause internal and final) comment clause,
the intermediate second stage is a hybrid category which is marked by a discrepancy between usage status and structural
status and as such is descriptively ambiguous (see Boye and Harder 2007: 586).
According to Boye and Harder (2007: 592), their proposed diachronic scenario is compatible with Thompson and Mulac’s
(1991) matrix clause hypothesis, as illustrated by the adapted version in (10).

(10) [I think] [I love her]


A: [main lexical CTP clause, primary status] [complement clause]
Y usage reanalysis
B: [main lexical CTP clause, secondary status] [complement clause]
Y structure reanalysis, grammaticalization of CTP
C: [grammatical CTP clause, secondary status] [main clause]

At the same time, Boye and Harder’s hypothesis is also argued to be compatible with Brinton’s (1996) account, with her
stage I CTP clause being only lexical and primary, stage II lexical but secondary, and stages III and IV being equivalent to
secondary and grammatical (see Boye and Harder 2007: 593).

3.2. Pragmaticalization

The concept of pragmaticalization has been proposed as a possible alternative to grammaticalization. The term was first
proposed by Erman and Kotsinas (1993: 79) and has been defined as “a process of linguistic change in which a full lexical item
. or grammatical item . changes category and status and becomes a pragmatic item, that is, an item which is not fully
integrated into the syntactic structure of the utterance and which has a textual or interpersonal meaning” (Dostie 2009: 203).
This view thus takes into account that the outcome of the process is that of a pragmatic marker, that is, not one of the
traditional grammatical word classes, and that grammaticalization cannot readily explain a number of features such as their
optionality, extra-clausal status, non-truth-conditional meaning, lack of fusion, and increase in semantic-pragmatic scope
(see Section 4 for discussion).
On this view, then, lexical or other expressions develop into discourse markers either without an intermediate stage of
grammaticalization or via a kind of grammaticalization that differs from ʽcanonicalʼ grammaticalization in that it does not
lead, e.g., to syntactic integration, the result being markers “mainly serving as textstructuring devices at [non-sentential]
levels of discourse” (Erman and Kotsinas 1993: 79; see Brinton 2008: 61). It has also been suggested (e.g. Aijmer 1997,
Dostie 2004, Günthner and Mutz 2004: 98, Claridge and Arnovick 2010: 167) that grammaticalization describes the emer-
gence of grammatical markers belonging to sentence grammar while the notion of pragmaticalization would be more
appropriate to describe the emergence of text structuring discourse markers.7
The concept of pragmaticalization has been evoked, for instance, in the development of the comment clause you know,
which Erman and Kotsinas (1993) see as delexicalized with no involvement of grammaticalization, and I think with its
expression of the speaker’s attitude to the hearer or the message (see Aijmer 1997).
As an explanatory concept for the rise of discourse markers, pragmaticalization has, however, also been the subject of
some debate. For example, while acknowledging that pragmaticalization constitutes a relevant mechanism needed to un-
derstand the diachronic behaviour of discourse markers, some authors propose to include it as a subclass of grammaticali-
zation (Prévost 2011: 391, Brinton 2010a: 64). Again other authors such as Traugott (2007: 151) and Brinton (2017: 35)
question the need for pragmaticalization as a separate type of change, arguing that “some degree of pragmaticalization will be

5
Primary and secondary CTPs differ in terms of discourse centrality, which can be related to addressability (e.g. retorts such as really? or no way!).
6
The term ‘lexical clause’ refers to the structural (i.e. morphosyntactic) properties of a typical clause, such as containing a VP and the ability to take
dependents in the form of a subordinate clause (Boye and Harder 2007: 591). In their framework ‘lexical’ contrasts with ‘grammatical’, which is equated
with secondary discourse function.
7
Some authors avoid the term ʽpragmaticalizationʼ but propose frameworks that can be related to this concept (cf. e.g. Waltereit 2002: 988; 2006,
Ocampo 2006: 307, Claridge and Arnovick 2010: 183).
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 5

found in any change involving function as well as form, including the development of standard examples of grammaticali-
zation such as auxiliaries” (Traugott 2007: 152).8

3.3. Lexicalization

Yet another explanation provided for the peculiar behaviour of comment clauses, and discourse markers more generally, is
that of lexicalization, which focuses on the end product of their development: Comment clauses are transparently derived
from clausal expressions but end up as mono-morphemic, unanalyzable forms – in other words, they move towards uni-
verbation (Lehmann [1982] 2015: 89, Haspelmath 2011).
The lexicalization hypothesis is most strongly associated with the work of Wischer (2000) and Fischer (2007a; 2007b), but
also mentioned in other studies (e.g. Schiffrin 1987: 319, Schourup 1999: 227, Claridge 2013: 180). Using the discourse marker
methinks as an example, Wischer (2000: e.g. 363) observes that an originally productive construction consisting of the
combination of clause-initial recipient plus impersonal verb becomes fossilized and unproductive, stored as a whole in the
lexicon as a ʽsymbolʼ, and classified as an ʽadverbʼ (see also Brinton 2017: 166). At the same time, Wischer is also aware that the
process cannot be reduced to lexicalization alone but that there is also grammaticalization since it may also involve coa-
lescence and fossilization.
Another form of the lexicalization hypothesis is proposed by Fischer (2007a, b), who draws attention to the fact that
phrases like I think do not follow the usual parameters of grammaticalization. Based on the observation that in non-
standardized languages, epistemic parentheticals are likely to form one lexical unit in the course of time, and that
morphological bonding may occur, Fischer (2007a: 308–311, 2007b) argues for lexicalization rather than grammaticalization
in the case of I think type comment clauses: “as formulaic tokens . they lose some referential content, being narrowed down
to a more epistemic, evaluative meaning” (Fischer 2007a: 117).
Reservations about the lexicalization hypothesis have been expressed mainly by proponents of the grammaticalization
hypothesis. Brinton (2017: 166), for instance, argues that there is insufficient evidence for treating epistemic parentheticals as
ʽlexical unitsʼ, as they do not constitute a major lexical category, and what appears to be involved is univerbation rather than
lexicalization. She also points at the increase in productivity, which is typical of grammaticalization but not of lexicalization. It
has to be noted, however, that comment clauses differ from grammatical units, for example, in terms of their morphosyntactic
independence – a point that had been emphasized already in Fischer’s work (2007a, b). This is an issue that we will turn to in
the following section.

4. The conundrum: accounting for a ‘quantum leap’

As noted above, the concept of pragmaticalization was proposed out of a concern that grammaticalization cannot satis-
factorily account for the development of discourse markers, including comment clauses. Indeed, it can be observed that up to
a certain stage in their development, discourse markers are in accordance with established parameters of grammaticaliza-
tion.9 But then they take a kind of ‘quantum leap’, which is elusive to these parameters, even though at a later stage their
development is shaped by the gradual evolution characterising grammaticalization. With the label ‘quantum leap’ we are
referring to the operation of cooptation, whereby segments of sentence grammar, such as main clauses without their
complements, are taken out of sentence grammar and transferred to the metatextual level of discourse processing to give rise
to comment clauses and discourse markers in general (Heine 2013). This operation, which does not involve parameters of
grammaticalization, has been described in detail in Heine et al. (2017).
To identify those problematic features in the development of comment clauses let us briefly summarize their main
grammatical properties in (11) (for a more detailed discussion see Heine 2013; Heine et al., 2021b in print).

(11) Grammatical properties of comment clauses


a. Meaning: Their meaning is not part of the meaning of the sentence they are associated with.
b. Function: Their function is metatextual.
c. Syntax: They are not a syntactic constituent of the sentence in which they occur.
d. Prosody: They are likely to be set off prosodically from the rest of the sentence.
e. Semantic-pragmatic scope: They have scope over and beyond the sentence.
f. Placement: They are flexible in their placement behavior, most commonly appearing in sentence-initial position.

The features in (11) are difficult to explain with the help of grammaticalization. As argued for instance by Ocampo (2006:
317) and Norde (2009: 23), grammaticalization is movement towards syntax and morphology whereas discourse markers
move precisely to the opposite end, namely outside of syntax and towards discourse (cf. also Waltereit 2002: 1006). Various
other authors have expressed similar dissatisfactions. Fischer (2007a: 280–297), for instance, draws attention to the fact that

8
An anonymous reviewer rightly points out that there may also be purely formal circumstances leading to processes of grammaticalization, in particular
in cases involving analogy (e.g., Fischer 2007a, b, De Smet 2012).
9
The set of parameters proposed here are: (a) Context extension, (b) desemanticization (ʽsemantic bleachingʼ), decategorialization (loss of morpho-
syntactic properties), and (d) erosion (loss of phonetic substance) (Heine and Kuteva 2007: 33–46).
6 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

word order evidence suggests that there is no direct line from adjunct to disjunct adverbial, that is, from sentence grammar
unit to parenthetical. Brinton (2017: 29) presents a list of features that are hard to reconcile with a grammaticalization hy-
pothesis, and Traugott and Trousdale (2013: 109–12) point out that there are features found in discourse markers that were
shown by proponents of pragmaticalization to be problematic for an analysis based on grammaticalization (see Furkó 2014:
293–5 for an insightful discussion).
A survey of studies on this issue suggests that it is most of all the following changes in the development of discourse
markers, including comment clauses, that are difficult to reconcile with observations commonly made in grammaticalization:

(12) Properties observed in the development of comment clauses, and discourse markers more generally
a. From meaning as part of the sentence to meaning outside the sentence
b. From sentence-grammar function to metatextual function
c. From syntactic constituent of the sentence to syntactically unattached status
d. From prosodically integrated to unintegrated or less integrated status
e. From more restricted semantic-pragmatic scope to wider scope
f. From positionally constrained to less constrained placement

Let us briefly look at these six changes in turn. Property (a) captures the fact that a discourse marker does not affect the
propositional content of a sentence or an utterance (Jucker 1993: 436, Fraser 1996, 1999: 936, Hansen 1997: 161, Lenk 1998:
52, Schourup 1999: 227) and can therefore easily be omitted. By contrast, meanings arising in grammaticalization shape the
meaning of sentences, be that within a phrase or a clause, or between clauses, but not normally beyond the level of sentences
(see Kuteva et al., 2019 for examples).
Property (b) refers to the discourse marker’s function of being anchored in, and relating the meaning of a sentence to, the
situation of discourse, rather than being anchored in a sentence (Heine 2018: 34). Various such metatextual (Traugott 1995: 6;
2018: 27) functions have been identified: For instance, signaling the speaker’s commitment with regard to the text segment
over which they have semantic-pragmatic scope (cf. Fraser 1988: 22) or providing “instructions from the speaker to the
hearer on how to integrate the host unit into a coherent mental representation of the discourse” (Hansen 1997: 161). Such
features contrast sharply with those that have been proposed for grammaticalization processes (e.g. Kuteva et al., 2019):
Expressions commonly providing the source for discourse markers serve functions of structuring sentences or parts of them,
and they retain this property when undergoing grammaticalization, but not when turning into discourse markers.
Property (c) refers to the conspicuous clause-external status of discourse markers (Hansen 1997: 156), being generally not
part of or only loosely attached to the syntactic structure (Jucker 2002: 212), being disjunctive (Tabor and Traugott 1998: 254)
or syntactically disjunct (Traugott 1995: 14) (cf. also Brinton 2010a: 64, 1996: 33–5; Jucker and Ziv 1998: 3). As such, they
cannot normally be questioned, they cannot become the focus of a cleft sentence, and they are not in the scope of a negated
sentence (cf. Haegeman 1991; Espinal 1991). Accordingly, the development of discourse markers has been described as
involving “an increase in syntactic freedom instead of syntactic fixation” (Norde 2009: 22). Increase in syntactic freedom and
movement outside of the sentence are, however, not compatible with any of the criteria that have been proposed for
grammaticalization (e.g. Lehmann [1982] 2015; Heine and Kuteva 2002: 2–4): Grammaticalization usually leads in the
opposite direction, towards decategorialization and bonding, almost invariably involving loss rather than gain of morpho-
syntactic independence.
Property (d) denotes the tendency of discourse markers to be prosodically set off from their host utterance (e.g. Zwicky
1985: 303, Schiffrin 1987: 328, Brinton 1996: 33, Hansen 1997: 156, 1998a: 66, Tabor and Traugott 1998: 254, Jucker and
Ziv 1998: 3, Traugott and Trousdale 2013: 110), forming a separate tone group (Brinton 1996: 33–5; Jucker 2002: 212),
also referred to as ‘comma intonation’ (Hansen 1997: 156, Tabor and Traugott 1998: 254). In this respect, discourse markers
differ from grammaticalizing items, which almost invariably lose features of prosodic distinctiveness on the way from lexical
to grammatical expressions: They tend to be integrated in the prosodic structure of their host (e.g. Haspelmath 2011: 347,
Wichmann 2011: 341).
Property (e) indicates that, as a rule, discourse markers exhibit an increase in semantic-pragmatic scope.10 This increase
leads from scope over parts of a sentence to scope over the whole sentence or over discourse beyond the sentence (e.g.
Waltereit 2002: 1005, Detges and Waltereit 2016, Brinton 2017: 29, Hansen 1998b: 236, Thompson and Mulac 1991, Traugott
1995: 1, Brinton 1996: 253, 2001: 194, 2008, Tabor and Traugott 1998: 254, Lewis 2011: 419, Traugott and Trousdale 2013: 109,
Brinton and Traugott 2005: 138). As metatextual expressions, discourse markers relate to the situation of discourse, that is,
the attitudes of the speaker, the interaction between speaker and hearer, and the organization of texts (see Heine et al., 2017).
Their scope, in other words, is determined by the discourse structure (Detges and Waltereit 2016) rather than by the
grammatical structure. Thus, what distinguishes the semantic-pragmatic scope of discourse markers from that of Sentence
Grammar expressions is that it “makes reference to discourse, not to grammar” (Waltereit 2006: 65).

10
The notions of ‘syntactic scope’ (e.g., Roberts and Roussou 2003) and ‘structural scope’ (in the sense of Lehmann [1982] 2015: 152–3) as used, for
example, by Tabor and Traugott (1998: 254), do not appear to be relevant to the rise of discourse markers for the simple reason that once an expression
gives rise to a discourse marker it is syntactically unattached and, hence, loses its ability to signal syntactic or structural scope beyond syntactic units such
as sentences (for further discussion see Heine et al., 2021b, Section 1.3.2.1).
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 7

Property (f), finally, captures the observation that discourse markers, as semantically and syntactically unattached items,
are typically unconstrained in their placement, that is, they may be found in various slots of a sentence (e.g. Tabor and
Traugott 1998: 254, Hansen 2008: 58). Brinton and Traugott (2005: 138) therefore conclude that Lehmann’s ([1982] 2015:
158–60) notion of fixation or loss of syntactic variability is problematic for discourse markers since they are “often quite
moveable” – a problem discussed by Van Bogaert (2011: 302–3) as the ʽdecategorialization paradoxʼ.
To conclude, discourse markers exhibit a number of features that are difficult to reconcile with a grammaticalization
account. These are, most importantly, the issues of disjunction and increase in scope (e.g. Traugott 1995: 14). One attempt to
overcome these problems has been to widen the concept of grammaticalization and propose a distinction between a narrow
or restricted and a wider or expanded view of grammaticalization (e.g. Traugott 1995, Hansen 1998b, Lenker 2000, Traugott
and Dasher 2002, Brinton and Traugott 2005: 136–140, Brinton 2008, Prévost 2011, Diewald 2011a, 2011b, Van Bogaert 2011,
Degand and Simon-Vandenbergen 2011: 290, Degand and Evers-Vermeul 2015: 77). Following this line of research, Traugott
and Trousdale (2013: 99–112) distinguish between two models of grammaticalization: Grammaticalization as reduction and
increase in dependency (GR) and grammaticalization as expansion (GE), with the latter covering the development of
discourse markers (see Heine 2018 for a critical discussion). Another attempt to account for discourse markers has been the
proposal of pragmaticalization as a separate process (although included by some as a subclass of grammaticalization; e.g.
Prévost 2011: 391). Yet another account for the development of discourse markers is proposed in the following section.

5. The view of Discourse Grammar: an alternative proposal

5.1. A dualistic view of grammar

The present study relies on the framework of Discourse Grammar, which distinguishes between two organizing principles
or grammatical domains: Sentence Grammar and Thetical Grammar (e.g. Kaltenböck et al., 2011; Heine et al., 2013). Sentence
Grammar is organized in terms of propositional concepts and clauses and their combination and has been the only, or the
main subject of mainstream theories of linguistics. The domain of Thetical Grammar, on the other hand, comprises elements
that are generally seen as being outside the confines of Sentence Grammar, the so-called theticals. Theticals are linguistic
discourse units which have metatextual functions and are anchored in the immediate situation of discourse. They include a
wide range of elements traditionally referred to as parentheticals or extra-clausal units, such as comment clauses and other
discourse markers, vocatives, formulae of social exchange (good morning), imperatives, interjections and the like (e.g.
Kaltenböck 2007: 31, Dik 1997).
What sets theticals apart from elements of Sentence Grammar are the grammatical properties listed in (13) (e.g. Heine
et al., 2021a, b in print), which not surprisingly correspond with the ones for comment clauses given in (11) above.

(13) Grammatical properties characterizing theticals


a. Meaning: Their meaning is not part of the meaning of sentence they are associated with.
b. Function: Their function is metatextual.
c. Syntax: They are not a syntactic constituent of the sentence in which they occur.
d. Prosody: They are likely to be set off prosodically from the rest of the sentence.
e. Semantic-pragmatic scope: They have scope over the sentence or beyond the sentence.
f. Placement: Depending on their discourse function they are flexible in their placement behaviour.

Functionally, theticals differ from Sentence Grammar elements in that they are metatextual and relate to the immediate
situation of discourse, mainly to the following areas of discourse processing: text organisation, speaker–hearer interaction,
and attitudes of the speaker.
The two domains of grammar are seen as being related in a number of ways, most crucially by a cognitive-communicative
operation called cooptation, which is responsible for creating theticals by lifting elements out of Sentence Grammar for
metatextual use. Cooptation, in other words, allows speakers to switch perspective from the sentence level to a meta-level
anchored in the situation of discourse, and can be defined as follows (see Kaltenböck et al., 2011: 874–875, Heine 2013:
1221, Heine et al., 2013: 204–205, Furkó 2014; Heine et al., 2016).

(14) Cooptation is a fully productive operation whereby a chunk of sentence grammar, such as a word, a phrase, a reduced clause, a full clause, or some
other piece of text, is deployed for use on the metatextual level of discourse processing, thereby turning into a thetical. Its functions are determined
by the situation of discourse, serving (a) to overcome constraints imposed by linearization in structuring texts, (b) to provide the source of
information, (c) to place a text in a wider perspective, e.g. by elaborating, proffering an explanation, a comment or supplementary information, (d) to
describe the attitudes of the speaker, and/or (e) to interact with the hearer.

Cooptation is an instantaneous operation and can, in principle, affect any element from Sentence Grammar. To the extent
that coopted elements become adopted by a wider speech community and conventionalize, three different types of theticals
can be distinguished (Heine et al., 2016: 188, Heine et al., 2017: 819–21), namely instantaneous, constructional, and formulaic
theticals. Instantaneous theticals are fully compositional and can be designed spontaneously any time and anywhere, and
quite a few are uttered only once and never again, as in the constructed example in (15) (cf. also Quirk et al.’s example (6)
above).
8 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

(15) Our chairman, you probably know already, will resign next month.

Constructional theticals are recurrent patterns or constructions having a schematic format and function. In the termi-
nology of Construction Grammar (Goldberg 1995; 2006; Croft 2001) they include both fully schematic and partially sub-
stantive constructions. Examples of fully schematic theticals are, for instance, comment clauses, as in (16).

(16) The bulk of your your writing was done at a time when I suppose the who dunnit the sort of Agatha Christie novel was at its most popular. (ICE-
GB:s1b-048-124)11

Formulaic theticals are unanalyzable, that is, their shape is largely or entirely invariable. They include formulae of social
exchange (e.g. Good morning!, hello!, please), interjections (e.g. boy!, fuck!, hell!) and discourse markers, such as anyway,
indeed, in fact, I mean, I think, you know, now, so, well. Compare:

(17) But I do think that Hillary, you know, upset a lot of people in the Midwest (COCA2019_SPOK_NBC_MeetPress)

5.2. Cooptation and grammaticalization

Cooptation, as outlined above, thus differs fundamentally from the process of grammaticalization. In the view of Discourse
Grammar both mechanisms are needed to reconstruct the development of comment clauses and discourse markers more
generally. Grammaticalization allows speakers to develop new patterns and forms for constructing sentences. Cooptation, on
the other hand, enables the transfer of linguistic expressions to the metatextual level of discourse processing. Cooptation is
responsible for salient grammatical properties of comment clauses, such as the ones listed in (11) above, but it marks only one
stage in their development.

Table 1
Comparing the effects of grammaticalization and cooptation.

Grammaticalization Cooptation
Meaning Change of meaning within the sentence From meaning as part of the sentence to meaning outside the sentence
Function Increasing functional dependence within the From function within the sentence to metatextual function
sentence
Syntax Increasing syntactic integration within the From integrated to detached syntax
sentence
Prosody Decrease in prosodic distinctiveness Increase in prosodic distinctiveness
Semantic-pragmatic Scope may change within the sentence From scope within the sentence to scope over and beyond the sentence
scope
Placement Decrease in freedom of placement Increase in freedom of placement

Let us briefly compare the specific effects of each of these two mechanisms. Grammaticalization, as is well known, can be
characterized as gradually leading “from lexical to grammatical forms and from grammatical to even more grammatical
forms” (Heine and Kuteva 2002: 2). Thus, the expression concerned is gradually pressed into service for processing grammar,
losing features it had as a relatively free lexical or other form and increasingly becoming dependent on its grammatical
environment. Cooptation, by contrast, has the effect of freeing an expression from earlier constraints so that it is no longer
dependent on its grammatical environment. As can be seen from Table 1, the effects of the two mechanisms are in fact
diametrically opposed to each other, with grammaticalization leading to a gradual loss of grammatical autonomy (Lehmann
2004: 155) or discourse autonomy (Hopper 1991: 30), or to stronger internal dependencies (Haspelmath 2004: 26), and
cooptation leading to gains in grammatical and discourse autonomy or independence.
If we now apply these two mechanisms to the development of comment clauses, such as I think, you know, the following
scenario presents itself. As discussed in Section 3, comment clauses can be traced back to semantically and morphosyntac-
tically transparent segments of sentence grammar. In the case of epistemic, main clause-like comment clauses (e.g. I think)
two possible source constructions have been identified in the literature: matrix clause structures such as I think that John is in
London, and adverbial clauses of the type John is in London, as/so I think. In accordance with Table 1, their development entailed
two kinds of grammatical changes: (i) In their external structure their development was one of loss of integration but gain of
independence. This can be attributed to cooptation, as a result of which they became syntactically unattached and lost their
status as a grammatical part of the sentence. This affected their external but not their internal status. (ii) In their internal
structure, their development was one of increasing integration as a result of grammaticalization. Expressions like I think, you

11
ICE-GB refers to the British component of the International Corpus of English (Nelson et al., 2002).
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 9

know, if you will, as it were lost their internal boundaries, with their constituent parts gradually coalescing and developing into
invariable markers.
On the analysis proposed here, then, two mechanisms are involved in the development of comment clauses (and other
discourse markers), each with contrasting effects: first cooptation, which leads to semantically and syntactically free forms,
followed by grammaticalization, leading to internal integration and ʽfixationʼ, the latter having, e.g., the effect that freely
combinable text segments like I think, you know, etc. gradually turn into invariable expressions as a result of late gramma-
ticalization (e.g. Heine 2018).
Although grammaticalization of discourse markers can already set in before cooptation, comment clauses undergo only
so-called ‘late grammaticalization’ (Heine et al., 2021b in print, 2.5.2), that is, after their cooptation as theticals: The
expression is no longer a constituent of the sentence, having the properties of theticals listed in (13), and it now develops
gradually into a discourse marker. Late grammaticalization is typically though not necessarily associated with ‘secondary
grammaticalization’, that is, the development from a grammatical to a more grammatical expression (Givón 1991; Breban
2015; Kranich 2015; Breban and Kranich 2015).12
In terms of the parameters of grammaticalization (Heine and Kuteva 2007: 33–46) the grammaticalization of comment
clauses involves mainly the parameters of desemanticization and context extension. With regard to desemanticization,
comment clauses lose some of their earlier lexical meaning in favour of procedural discourse functions (e.g. Crible 2017: 106),
and they are no longer affected by the truth conditions of their host sentence (e.g. Aijmer 1997: 2). Context extension is
arguably the most conspicuous effect of grammaticalization: In much the same way as they are desemanticized, their use is
extended to new contexts inviting new discourse functions. This also entails greater positional flexibility, unlike ordinary
grammaticalization. Highly grammaticalized comment clauses may also show signs of erosion, as can be seen in the phonetic
reduction of you know to y’know and I guess to guess.13 There may also be suprasegmental erosion with high-frequency
comment clauses such as I think being gradually re-integrated in the intonation group of the host clause (cf. Kaltenböck 2008).
The grammaticalization of comment clauses can be described as a continuum or chain, extending from weakly to strongly
grammaticalized. Weakly grammaticalized comment clauses have retained part of the lexical meaning they had prior to coop-
tation, express only a limited range of discourse functions, show little, if any, decategorialization and erosion, and may exhibit
some morphosyntactic variation rather than being restricted to one invariable form. An example is the comment clause I admit,
which retains a number of variants such as I admit, I will admit, I must admit, I have to admit, you admit, you will admit, you must
admit, and you have to admit (Brinton 2017: 168; see also Section 5.3). Strongly grammaticalized comment clauses, by contrast,
correspond to the prototypical notion of ʽformulaic theticalsʼ (Section 5.1): They have little, if any lexical meaning content, express
a wider range of discourse functions, and are restricted to one invariable form, such as the discourse marker I mean.
It has to be noted, however, that one and the same comment clause can be associated with different stages of context-
dependent usages, being weakly grammaticalized in some contexts but strongly grammaticalized in other. For example, I
think can serve, on the one hand, as a fairly variable thetical, being suggestive of a weak degree of grammaticalization. On the
other hand, it also shows uses as a highly grammaticalized, invariable expression, with the characteristics of a paradigm
instance of discourse markers (Kaltenböck 2007: 47, 2011, 2013).
The account proposed here is largely compatible with the pragmaticalization hypothesis but differs from it in two
important aspects: First, rather than one unitary mechanism, the present account assumes that there are two distinct
mechanisms involved, namely cooptation followed by grammaticalization. And second, unlike pragmaticalization, the
cooptation hypothesis is based on a wider perspective of discourse structuring – one that accounts not only for comment
clauses and discourse markers but also for other kinds of grammatical categories such as reporting clauses, non-restrictive
relative clauses, as well as formulaic categories like interjections, formulae of social exchange, question tags, expletives,
exclamatives, etc.14

5.3. An example: I admit

In this section we will put the cooptation hypothesis, as outlined above, to the test by drawing on historical data provided
by Brinton (2017). The comment clause investigated here is I admit as an example of a weakly grammaticalized comment
clause or constructional thetical (for a strongly grammaticalized comment clause see Heine et al.’s 2021b in print discussion of
I mean).
The speech act verb admit has a number of different uses, notably the ones illustrated in (18) and (19), which we refer to as
I admit1 and I admit2 respectively.

(18) I admit that I have done things in my life that I am not proud of (COCA2019_NEWS_New York Post)
(19) This would, I admit, give no small degree of pleasure (COCA2016_FIC_Analog)

12
Grammaticalization, which is a gradual development, is thus cut across by cooptation, which divides the grammaticalization in into two phases, one
preceding and one following cooptation, or early and late grammaticalization respectively.
13
An anonymous reviewer rightly observes that such reduced forms can also be found in fast speech in non-grammaticalized main clauses, so they are not
necessarily the result of the grammaticalization parameter of erosion.
14
Cooptation can thus account also for non-conventionalized comment clauses as discussed by Quirk et al. (1985: 1114), such as example (6) above.
10 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

I admit1 in (18) is performative and has all the features of a verb of Sentence Grammar, taking a pronominal subject and a
phrasal or a clausal complement typically introduced by that. It is fully productive in that it can be negated (I don’t admit),
questioned (do I admit?), etc. I admit2 in (19), on the other hand, is a comment clause and as such does not take complements,
is not a syntactic constituent of the host sentence, and is likely to be set off prosodically, frequently signalled in writing by
commas or other punctuation marks. I admit2, in other words, has all the hallmarks of discourse markers (theticals) identified
in Section 4. They are listed again for convenience in (20).

(20) Grammatical properties of I admit2 distinguishing it from I admit1


a. Meaning: Its meaning is outside of the sentence it is associated with.
b. Function: Its function is metatextual.
c. Syntax: It is not a syntactic constituent of the sentence in which it occurs.
d. Prosody: It is likely to be set off prosodically from the rest of the sentence.
e. Semantic-pragmatic scope: It has scope over the sentence as a whole.
f. Placement: Depending on its function it can be placed within the sentence or at its left or right periphery (cf. Brinton 2017, Table 6.2).

According to the findings of Brinton (2017) I admit1, with a speech act meaning, can be traced back to Middle English, such
as the early attestation in (21).

(21) If eny man aske whi y wole not admytte [.] þat all menal moral vertues [.] be in þe iiije table oonli .
‘if any man asks why I will not admit [.] that all ancillary moral virtues [.] are in the fourth table only .’.
(c1454 Pecock, The follower to The Donet (Roy 17.D.9) 200/ 17 [MED]; Brinton 2017: 179)

By comparison, unambiguous examples of I admit2, in medial or final position (Brinton 2010b: 295), do not appear until the
18th century, as in (22).

(22) for the three former qualities, I admit, make you worthy of happiness.
(1749 Fielding, Tom Jones [CLMET3.0]; Brinton 2017: 182)

For the development of I admit2 Brinton (2017: 187) evokes Thompson and Mulac’s (1991) matrix clause hypothesis (see
Section 3.1). On this view, I admit2 results from a grammaticalization of I admit1, that is, a matrix clause use such as in (23),15
where loss of the that complementizer leads to indeterminate structures with the original matrix clause being reanalysed as a
parenthetical.

(23) I admit the Northwest point standeth due North, and my course is to go due West, (1574 Bourne, A regiment for the sea [EEBO]; Brinton 2017: 181)

While there is clear evidence that I admit1 preceded the first occurrence of I admit2 in time and that I admit2 is historically
derived from I admit1, as argued convincingly by Brinton (2017), there is no conclusive evidence to suggest that the specific
grammatical properties of I admit2 listed in (20) evolved gradually, over a longer period of time. Instead, I admit2 seems to
have acquired them fairly suddenly in the 18th century or earlier.16
It seems obvious that I admit2 can at best be interpreted as a case of low degree of grammaticalization. Once again,
however, grammaticalization theory is hard-pressed to explain the grammatical properties of I admit2 given in (20), which are
at variance with the unidirectionality principle of grammaticalization and its prediction of grammatical change from less
dependent to more dependent linguistic expressions rather than the other way round. More specifically, in accordance with
the parameters of desemanticization and bonding (Heine and Kuteva 2007: 33–44, Lehmann [1982] 2015: 132), gramma-
ticalizing nouns or verbs will gradually become functionally dependent on their host constituent. Second, in accordance with
the parameter of decategorialization, grammaticalization leads as a rule to reduction or loss of morphosyntactic indepen-
dence, rather than to an increase in independence. And third, in accordance with the parameter of erosion, grammaticalizing
nouns, verbs, and other units are likely to lose phonetic substance, including prosodic features; jointly with their host
constituent they tend to become part of one and the same intonation unit.
A reconstruction involving cooptation, however, can account for these properties, namely independence from the host
clause in terms of meaning, syntax and prosody, metatextual function, and positional freedom. And once coopted, or – in the
terminology of Brinton (2017: 187) – “reanalyzed as parenthetical”, in the 18th century, I admit2 was ready to grammaticalize
in the direction of a discourse marker, but showing no traces of a pronounced grammaticalization like I mean. Since, as we
argue, I admit was coopted without a complementizer, we see no reason to assume that there was a grammaticalization
process leading to the loss of the that complementizer.

15
Note, however that I admit in (23), depending on prosody, would allow also for classification as a clause-initial comment clause rather than a matrix
clause, i.e. a thetical rather than a sentence grammar unit.
16
Note that this conclusion is based on the written evidence that there is. As an anonymous reviewer rightly observes, the situation might be different if
data were available on spoken language use, which unfortunately is not the case.
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 11

6. The issue of initial comment clauses: how to account for ambiguous cases?

While comment clauses in clause-medial and clause-final position are unambiguously identifiable as such, clause-initial
position, as illustrated in (24) and (25), has been controversially discussed as either a main clause or a comment clause.

(24) I think that this is a real split (COCA2019_SPOK_NBC_MeetPress)


(25) I think you have to keep at it methodically (COCA2019_SPOK_CBS_FaceNation)

Some studies assume a strict division between main and comment clause depending on the presence or absence of a that
complementizer (e.g. Thompson and Mulac 1991; Aijmer 1997). Accordingly, I think in (24) would be analysed as a main
clause, while in (25) it is a comment clause. Various syntactic and semantic tests have been proposed for such a distinction
(e.g. Aijmer 1972: 52, 1997: 8, Hand 1993: 501, Knowles 1980: 405, Asher 2000: 33, Erteschik-Shir and Lappin 1979: 46), none
of which is, however, unproblematic (see Kaltenböck 2011). Other studies maintain the traditional main clause analysis for
both cases, with and without complementizer (e.g. Kearns 2007), and again other studies, mainly from a cognitive-functional
perspective, support a comment clause analysis of initial CTP clauses irrespective of whether they are followed by a
complementizer or not (e.g. Thompson 2002; Kärkkäinnen 2003; Diessel and Tomasello 1999, 2001; Boye and Harder, 2007).
The status of that as a marker of subordination in spoken language has also been challenged by prosodic studies (e.g.
Kaltenböck 2011).
Without assessing the different positions, the question that arises for the present study is how a dualistic view of
grammatical organisation such as Discourse Grammar can account for the ambiguous (Biber et al., 1999: 197) and potentially
indeterminate status of initial CTP clauses.
While for initial CTP clauses with a that complementizer there are, arguably, structural reasons for its analysis as a main
clause, namely the presence of a subordinator, for those without a that establishing the syntactic status is a particular challenge.
To answer this question, we need to take a closer look at the relationship and interaction between the two domains
of Discourse Grammar: Sentence Grammar and Thetical Grammar. Although essentially dualistic in nature, Discourse
Grammar does allow for overlaps and links between the two domains. One category of overlap, for instance, is that of
imperatives, as discussed by Heine (2016). Fundamentally, it is however the mechanism of cooptation that bridges the
gap between the two domains. As noted above, cooptation is responsible for the transfer of an element of Sentence
Grammar to the plane of Thetical Grammar and thus for the creation of theticals. This transfer frees a linguistic
element from the constraints of Sentence Grammar, such as dependency relations with other elements. In the case of
comment clauses, for instance, they no longer require a complement and in that sense they can be described as
elliptical. However, comment clauses still require a link to a host clause or textual anchor. The link to another linguistic
unit is thus still present in a comment clause, even if in a much weaker and transmuted form, elevated from valency
to the discourse level. Put differently, a coopted unit may retain ‘memory’ traces of its previous existence as a Sentence
Grammar element, to varying extents depending on its degree of grammaticalization and its entrenchment as a
thetical.
A good way of capturing the links between the two domains is in terms of a constructional account (e.g. Goldberg 1995,
2006). From a Construction Grammar perspective constructions are independent, but not isolated entities. They are linked
with other, related constructions of different levels of schematicity in a taxonomic network of constructions. The domains of
Sentence Grammar and Thetical Grammar can each be taken to represent two different types of networks or families of
constructions (e.g. Ruiz de Mendoza Ibáñez et al., 2017; Diessel 2019). Since clause-initial CTPs share features with both
matrix clauses (i.e. Sentence Grammar) and comment clauses (i.e. Thetical Grammar) they can be represented as activating
links to both these domains. The links to Sentence Grammar are mainly formal in nature, that is, they share the same syntactic
position and subject-predicate form with matrix clauses and allow for the insertion of a complementizer. The links to Thetical
Grammar, on the other hand, are mainly functional, that is, they exhibit the same function as parenthetical comment clauses.
Accordingly, initial CTP clauses take an intermediate position between Sentence Grammar elements and theticals (cf. Boye
and Harder’s 2007 intermediate category; see Section 3.1).
For initial CTP clauses the following parent constructions can be identified. In the domain of Thetical Grammar we can
assume inheritance links to the category of (parenthetical) comment clauses, which in turn are linked to the larger class of
discourse markers, and extra-clausal elements more generally. In the domain of Sentence Grammar we can posit a link to the
Matrix clause þ Object clause construction (including a wide range of CTPs, going beyond the cognitive and communication
verbs found with comment clauses, e.g. I understand/guess/respect/regret . that John is not pleased), which in turn are linked to
the more schematic Transitive construction, instantiated most typically in the form of Subject þ transitive verb þ Object Noun
Phrase (e.g. I understand the problem).
The strength of these taxonomic links will depend on the degree of entrenchment of the parent constructions and thus
vary depending on the predicate type involved, as is illustrated for I guess and I understand in Figs. 1 and 2 respectively (where
line thickness indicates greater strength). Accordingly, we can expect a stronger link of initial I guess with Thetical Grammar,
since the parenthetical use of I guess is a well-established, high-frequency comment clause (attested with 20 non-initial
instances in ICE-GB) and thus more entrenched than the Sentence Grammar counterpart, transitive I guess, as in I guess
12 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

the number (attested with 5 instances in ICE-GB).17 Initial I understand, on the other hand, can be expected to show a stronger
link to Sentence Grammar, where its parent construction of transitive I understand, as in I understand the problem, is more
frequent than its Thetical Grammar parent construction, namely parenthetical I understand (the ICE-GB figures being 84 for
the former and 5 for the latter).

Fig. 1. Taxonomic network links for clause-initial I guess (cxn ¼ construction).

Fig. 2. Taxonomic network links for clause-initial I understand (cxn ¼ construction).

A constructional analysis of initial CTP clauses in terms of varying affinities to different parent constructions can thus
account for their intermediate status between Sentence Grammar and Thetical Grammar. Moreover, based on the assumption
of an inheritance link between initial CTP clauses and transitive object NP constructions, as outlined above, it should be
possible to account for the varying propensity of different CTPs to occur with a complementizer in terms of their relative
activation of the transitive construction. Accordingly, the higher rate of that found for some CTPs can be attributed to a
stronger link with the transitivity scheme, as measured by a greater propensity of these predicates to occur with a direct
object NP (cf. I believe a story vs. *I suppose a problem).
Evidence from corpus data (ICE-GB) suggests that this is indeed what seems to be the case (cf. Kaltenböck 2013). The CTPs
most frequently associated with a that complementizer are I understand, I believe, I realise (see Table 2 for figures in the spoken
part of ICE-GB from Van Bogaert 2009: 378). Interestingly, it is precisely these verbs (understand, believe, realise), which also
have the highest proportion of direct object NPs in ICE-GB: e.g. He just didn’t understand the situation (s1a-018-278), Foreigners

Table 2
Frequencies of initial CTP clauses þ that (in spoken ICE-GB; from Van Bogaert 2009: 378) and frequencies of verbs and their transitive use with object NP (in
total ICE-GB).

Initial CTP clause þ that (in spoken ICE-GB) Overall verb frequency (in ICE-GB) þ direct object NP (in ICE-GB)
I understand 50.00% understand (187) 46.52% (87)
I believe 46.15% believe (295) 15.93% (47)
I realise 33.33% realise (87) 21.84% (19)
I guess 9.09% guess (62) 8.06% (5)
I think 8.78% think (2,563) 0.31% (8)
I suppose 6.52% suppose (237) 0.00% (0)
I expect 0.00% expect (124) 11.29% (14)
I reckon 0.00% reckon (13) 0.00% (0)

17
As pointed out by Vandelanotte (2006: 154–155, 2009: 302f), even highly entrenched theticals such as I think may still be activated as fully
“addressable” by specific discourse configurations, e.g. I think they’ll be all right. Mm. I really do.
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 13

believe this too (s2b-035-085), It didn’t take very long to realise that (s1a-047-106). On the other hand, CTPs which rarely take a
that complementizer (viz. I reckon, I expect, I suppose, I think, I guess) show a weaker association with NP objects (see Table 2
for figures).
A constructional account as outlined above thus not only accommodates the intermediate status of initial CTPs between
Sentence Grammar and Thetical Grammar but can also shed light on the varying use of the complementizer with different
CTPs.18

7. Conclusion

The main concern of the paper was with the historical development of comment clauses. We saw in the preceding sections
that linguistic structures involving I think and related expressions can clearly be classified with reference to which domain of
grammar they belong to on the basis of the criteria listed in Section 5.1. This means, for example, that if a given expression
conforms to the properties listed there, like semantic, syntactic and prosodic detachment, it is unambiguously a thetical, such
as a comment clause or, more generally, a discourse marker. But there are a few contexts where the distinction is or appears to
be blurred on the basis of the linguistic evidence available. The case discussed in Section 6 presents a paradigm example of
this kind, where neither syntactic, prosodic nor other criteria clearly allow establishing whether I guess in examples like I
guess John is not pleased is a main clause belonging to Sentence Grammar or a comment clause of Thetical Grammar. We saw in
Section 5.2, however, that there are factors such as degree of entrenchment, constructional knowledge and frequency of use
that may guide hearers or readers in assigning an appropriate conceptual interpretation to the utterance or the discourse
concerned, as has been argued by Hansen (1997: 161).
The development described in the paper has been the subject of detailed reconstruction work, and it has given rise to a
number of hypotheses on how the development is best to be explained. The most commonly discussed hypotheses were
summarized in Section 3 and, as we saw there, the notion of grammaticalization played an important role in those discus-
sions. This research has generated a range of new insights, such as on how earlier main clauses can give rise to parenthetical
expressions (Thompson and Mulac 1991), on the potential role of bi-clausal structures in shaping the rise of comment clauses
(Brinton 2006, 2008, 2017), or on how the growth of comment clauses can be affected by shifts from primary to secondary
discourse status (Boye and Harder 2007).
As much as such insights have enriched our knowledge of the rise and development of comment clauses, their contri-
bution to understanding why comment clauses have the grammatical properties they have, such as the ones listed in Section
5.1, is limited. To be sure, the history of comment clauses from lexical main clause to discourse-structuring element has
features of a gradual process, as captured in the accounts of grammaticalization sketched in Section 3. But such accounts do
not seem to take care of the fact that comment clauses are the result of a shift from syntactically and prosodically integrated
complement-taking predicate to syntactically and prosodically unattached expression having metatextual functions.
In the present paper, an account was proposed that takes care of exactly this fact. Based on the framework of Discourse
Grammar it was argued that the development of comment clauses involves a stage of cooptation, whereby a text piece of
Sentence Grammar, such as a complement-taking predicate clause, is transferred from the sentence level to the metatextual
level of discourse processing. As a result of this operation, that text piece on the one hand acquires the new functions dis-
cussed in Section 5.1, namely serving to overcome constraints imposed by linearization in structuring texts, to provide the
source of information, to place a text in a wider perspective, to describe the attitudes of the speaker, and/or to interact with
the hearer. On the other hand it loses most of the grammatical features that it had as part of the structure of the sentence to
which it belonged and acquires a range of new grammatical properties, namely independence from the host clause in terms of
meaning, syntax and prosody, metatextual function, and positional freedom.
Such changes are not only hard to reconcile with grammaticalization, they even contradict what is commonly observed in
grammaticalization processes, as can be seen most clearly in Table 1 of Section 5.2. At the same time, we observed in that
section that grammaticalization also plays a role in the process, affecting most of all the internal form of coopted expressions:
On the way to turning into discourse markers, comment clauses are likely to lose part or even all of their lexical meaning and
internal morphological boundaries, turning into unanalyzable forms, and sometimes they also lose part of their phonetic
substance, in accordance with the grammaticalization parameters of desemanticization, decategorialization, and erosion,
respectively. The overall development of discourse markers, including comment clauses, thus can be depicted as in (26),
where the brackets signal that grammaticalization is optional here, that is, it may but need not precede cooptation while it
obligatorily follows cooptation.

(26) Hypothesized development of discourse markers (Heine 2018: 41)


(Grammaticalization >) cooptation > grammaticalization

To conclude, from the perspective proffered in this paper the development of comment clauses is shaped by two distinct
mechanisms of linguistic change. Grammaticalization may accompany their history from its beginning to its present state.

18
It is clear that this is only one of the possible conditioning factors for the use of the that complementizer. For an overview of various other factors
identified in the literature see, for instance, Shank et al. (2016), Kaltenböck (2006).
14 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

Cooptation, in contrast, marks only a short phase of this history but determines their subsequent development and accounts
for why comment clauses in particular and discourse markers in general differ substantially, both in their structure and their
functions, from grammatical forms having no history of cooptation.

Acknowledgements

We wish to thank the editors of this special issue, Lieselotte Brems, An Van linden and Lieven Vandelanotte, for their
helpful guidance in the editorial process and for organising the workshop on “The syntagmatic properties of complemen-
tation patterns: Accommodating lexical and grammatical uses of CTP-clauses” held from 9 to 10 May 2019 in Liège, as well as
all the workshop participants for the stimulating discussions. Special thanks are also due to the following colleagues for all the
input they provided while we were working on this paper: Mikyun Ahn, Kate Beeching, Laurel Brinton, Liesbeth Degand, Olga
Fischer, Maj-Britt Mosegaard Hansen, Alexander Haselow, Diana Lewis, Seongha Rhee, and Elizabeth Traugott. Our gratitude is
also due to two anonymous reviewers for all the constructive comments they made on an earlier version of the paper.

References

Aijmer, K., 1972. Some Aspects of Psychological Predicates in English. Almqvist & Wiksell, Stockholm.
Aijmer, K., 1997. I think d an English modal particle. In: Swan, T., Jansen-Westvik, O. (Eds.), Modality in Germanic Languages: Historical and Compasrative
Perspectives. Mouton de Gruyter, Berlin & New York, pp. 1–47.
Akimoto, M., 2002. On the grammaticalization of the parenthetical I’m afraid. In: Fisak, J. (Ed.), Studies in English Historical Linguistics and Philology: A
Festschrift for Akio Oizumi. Peter Lang, Frankfurt am Main, pp. 1–9.
Asher, N., 2000. Truth conditional discourse semantics for parentheticals. J. Semant. 17, 31–50.
Biber, D., Johansson, S., Leech, G., Conrad, S., Finegan, E., 1999. Longman Grammar of Spoken and Written English. Longman, London.
Boye, K., Harder, P., 2007. Complement-taking predicates: usage and linguistic structure. Stud. Lang. 31 (3), 569–606.
Breban, T., 2015. Refining secondary grammaticalization by looking at subprocesses of change. Lang. Sci. 47, 161–171.
Breban, T., Kranich, S., 2015. Introduction to what happens after grammaticalization? Secondary grammaticalization and other late stage processes. Lang.
Sci. 47, 129–131.
Brinton, L.J., 1996. Pragmatic Markers in English: Grammaticalization and Discourse Functions. Mouton de Gruyter, Berlin & New York.
Brinton, L.J., 2001. From matrix clause to pragmatic marker: the history of look-forms. J. Hist. Pragmat. 2, 177–199.
Brinton, L.J., 2006. Pathways in the development of pragmatic markers in English. In: van Kemenade, A., Los, B. (Eds.), The Handbook of the History of
English. Blackwell, Oxford, pp. 307–334.
Brinton, L.J., 2008. The Comment Clause in English: Syntactic Origins and Pragmatic Development. Cambridge University Press, Cambridge.
Brinton, L.J., 2010a. The development of I mean: implications for the study of historical pragmatics. In: Fitzmaurice, S.M., Taavitsainen, I. (Eds.), Methods in
Historical Pragmatics. Mouton de Gruyter, Berlin, pp. 37–80.
Brinton, L.J., 2010b. From performative to concessive disjunct: I/you admit and admittedly. In: Kytö, M., Scahill, J., Tanabe, H. (Eds.), Language Change and
Variation from Old English to Late Modern English: A Festschrift for Minoji Akimoto, Studies in Language and Communication, vol. 14. Lang, Bern, pp.
279–302.
Brinton, L.J., 2017. The Development of Pragmatic Markers in English: Pathways of Change. Cambridge University Press, Cambridge.
Brinton, L.J., Traugott, E.C., 2005. Lexicalization and Language Change. Cambridge University Press, Cambridge.
Claridge, C., 2013. The evolution of three pragmatic markers: as it were, so to speak/say and if you like. J. Hist. Pragmat. 14 (2), 161–184.
Claridge, C., Arnovick, L., 2010. Pragmaticalisation and discursisation. In: Jucker, A.H., Taavitsainen, I. (Eds.), Historical Pragmatics, Handbook of Pragmatics,
vol. 8. Mouton de Gruyter, Berlin, pp. 165–192.
Crible, L., 2017. Towards an operational category of discourse markers: a definition and its model. In: Fedriani, C., Sansò, A. (Eds.), Pragmatic Markers,
Discourse Markers and Modal Particles, Studies in Language Companion Series, vol. 186. Benjamins, Amsterdam & Philadelphia, pp. 99–124.
Croft, W., 2001. Radical Construction Grammar. Oxford University Press, Oxford.
Davies, M., 2008. The Corpus of Contemporary American English (COCA): One Billion Words, 1990-2019. Available online at. https://www.english-corpora.
org/coca/.
De Smet, H., 2012. The course of actualization. Language 88 (3), 601–633.
Degand, L., Evers-Vermeul, J., 2015. Grammaticalization or pragmaticalization of discourse markers? More than a terminological issue. J. Hist. Pragmat. 16
(1), 59–85.
Degand, L., Simon-Vandenbergen, A.-M., 2011. Introduction: grammaticalization and (inter)subjectification of discourse markers. Linguistics 49 (2), 287–
294.
Detges, U., Waltereit, R., 2016. Grammaticalization and pragmaticalization. In: Fischer, S., Gabriel, C. (Eds.), Manual of Grammatical Interfaces in Romance,
Manuals of Romance Linguistics. Mouton de Gruyter, Berlin, pp. 635–657.
Diessel, H., 2019. The Grammar Network: How Language Structure Is Shaped by Language Use. Cambridge University Press, Cambridge.
Diessel, H., Tomasello, M., 1999. Why complement clauses do not include a that-complementizer in early child language. In: Proceedings of the 25th Annual
Meeting. Berkeley Linguistics Society, pp. 86–97.
Diessel, H., Tomasello, M., 2001. The acquisition of finite complement clauses in English: a corpus-based analysis. Cognit. Ling. 12, 97–141.
Diewald, G., 2011a. Grammaticalization and pragmaticalization. In: Narrog, H., Heine, B. (Eds.), The Oxford Handbook of Grammaticalization. Oxford
University Press, Oxford, pp. 450–461.
Diewald, G., 2011b. Pragmaticalization (defined) as grammaticalization of discourse functions. Linguistics 49 (2), 365–390.
Dik, S.C., 1997. The Theory of Functional Grammar, Part 2: Complex and Derived Constructions. In: Functional Grammar Series, vol. 21. Mouton de Gruyter,
Berlin & New York, pp. 379–409.
Dostie, G., 2004. Pragmaticalisation et marqueurs discursifs: Analyse sémantique et traitement lexicographique. De Boeck & Larcier, Brussels.
Dostie, G., 2009. Discourse markers and regional variation in French: a lexico-semantic approach. In: Beeching, K., Armstrong, N., Gadet, F. (Eds.), Socio-
linguistic Variation in Contemporary French. Benjamins, Amsterdam & Philadelphia, pp. 201–214.
Erman, B., Kotsinas, U.-B., 1993. Pragmaticalization: the case of ba and you know. Stud. Mod. Sprakvetenskap 10, 76–92.
Erteschik-Shir, N., Lappin, S., 1979. Dominance and the functional explanation of island phenomena. Theor. Ling. 6 (1), 41–86.
Espinal, M.T., 1991. The representation of disjunct constituents. Language 67, 726–762.
Fischer, O., 2007a. Morphosyntactic Change. Oxford University Press, Oxford.
Fischer, O., 2007b. The development of English parentheticals: a case of grammaticalization? In: Smit, U., et al. (Eds.), Tracing English through Time: Ex-
plorations in Language Variation: A Festschrift for Herbert Schendl on the Occasion of His 65th Birthday. Braumüller, Vienna, pp. 103–118.
Fraser, B., 1988. Types of English discourse markers. Acta Ling. Hung. 38, 19–33.
Fraser, B., 1996. Pragmatic markers. Pragmatics 6 (2), 167–190.
Fraser, B., 1999. What are discourse markers. J. Pragmat. 31, 931–952.
B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400 15

Furkó, B.P., 2014. Cooptation over grammaticalization. Argumentum 10, 289–300.


Givón, T., 1991. The evolution of dependent clause morpho-syntax in Biblical Hebrew. In: Traugott, E.C., Heine, B. (Eds.), Approaches to Grammaticalization,
vol. 1. Benjamins, Amsterdam & Philadelphia, pp. 257–310.
Goldberg, A.E., 1995. Constructions: A Construction Grammar Approach to Argument Structure. University of Chicago Press, Chicago.
Goldberg, A.E., 2006. Constructions at Work: The Nature of Generalization in Language. Oxford University Press, Oxford.
Günthner, S., Mutz, K., 2004. Grammaticalization vs. pragmaticalization? The development of pragmatic markers in German and Italian. In: Bisang, W.,
Himmelmann, N., Wiemer, B. (Eds.), What Makes Grammaticalization? A Look from its Fringes and its Components. Mouton de Gruyter, Berlin, pp. 77–
107.
Haegeman, L., 1991. Parenthetical adverbials: the radical orphanage approach. In: Chiba, S., Ogawa, A., Fuiwara, Y., Yamada, N., Koma, O., Yagi, T. (Eds.),
Aspects of Modern Linguistics: Papers Presented to Masatomo Ukaji on His 6oth Birthday. Kaitakushi, Tokyo, pp. 232–254.
Hand, M., 1993. Parataxis and parentheticals. Ling. Philos. 16, 495–507.
Hansen, M.-B. Mosegaard, 1997. Alors and donc in spoken French: a reanalysis. J. Pragmat. 28, 153–187.
Hansen, M.-B. Mosegaard, 1998a. The semantic status of discourse markers. Lingua 104 (3–4), 235–260.
Hansen, M.-B. Mosegaard, 1998b. The Function of Discourse Particles. A Study with Special Reference to Spoken Standard French. John Benjamins,
Amsterdam & Philadelphia.
Hansen, M.-B. Mosegaard, 2008. Particles at the Semantics/Pragmatics Interface: Synchronic and Diachronic Issues. Elsevier, Oxford.
Haspelmath, M., 2004. On directionality in language change with particular reference to grammaticalization. In: Fischer, O., Norde, M., Perridon, H. (Eds.),
Up and Down the ClinedThe Nature of Grammaticalization, Typological Studies in Language, vol. 59. Benjamins, Amsterdam & Philadelphia, pp. 17–44.
Haspelmath, M., 2011. The gradual coalescence into ’words’ in grammaticalization. In: Narrog, H., Heine, B. (Eds.), The Oxford Handbook of Grammatic-
alization. Oxford University Press, Oxford, pp. 342–355.
Heine, B., 2013. On discourse markers: grammaticalization, pragmaticalization, or something else? Linguistics 51 (6), 1205–1247.
Heine, B., 2016. On non-finiteness and canonical imperatives. In: Chamoreau, C., Estrada-Fernández, Z. (Eds.), Finiteness and Nominalization, Typological
Studies in Language, vol. 113. Benjamins, Amsterdam & Philadelphia, pp. 245–270.
Heine, B., 2018. Are there two different ways of approaching grammaticalization? In: Hancil, S., Breban, T., Vicente Lozano, J. (Eds.), New Trends on
Grammaticalization and Language Change. Benjamins, Amsterdam & Philadelphia, pp. 23–54.
Heine, B., Kuteva, T., 2002. World Lexicon of Grammaticalization. Cambridge University Press, Cambridge.
Heine, B., Kuteva, T., 2007. The genesis of grammar: a reconstruction. In: Studies in the Development of Language, vol. 9. Oxford University Press, Oxford.
Heine, B., Kaltenböck, G., Kuteva, T., Long, H., 2013. An outline of discourse grammar. In: Bischoff, S., Jany, C. (Eds.), Functional Approaches to Language.
Mouton de Gruyter, Berlin, pp. 175–233.
Heine, B., Kaltenböck, G., Kuteva, T., 2016. On insubordination and cooptation. In: Evans, N., Watanabe, H. (Eds.), Insubordination, Typological Studies in
Language. Benjamins, Amsterdam & Philadelphia, pp. 39–63.
Heine, B., Kaltenböck, G., Kuteva, T., Long, H., 2017. Cooptation as a discourse strategy. Linguistics 55, 1–43.
Heine, B., Kaltenböck, G., Kuteva, T., Long, H., 2021a. On the rise of discourse markers. In: Hancil, S., Haselow, A. (Eds.), Studies at the Grammar–Discourse
Interface. Benjamins, Amsterdam & Philadelphia in print.
Heine, B., Kaltenböck, G., Kuteva, T., Long, H., 2021b. The Rise of Discourse Markers. Cambridge University Press, Cambridge in print.
Hopper, P.J., 1991. On some principles of grammaticization. In: Traugott, E.C., Heine, B. (Eds.), Approaches to Grammaticalization, Typological Studies in
Language 19, 1, vol. 1. Benjamins, Amsterdam & Philadelphia, pp. 17–35.
Jucker, A.H., 1993. The discourse marker well: a relevance-theoretical account. J. Pragmat. 19 (5), 435–452.
Jucker, A.H., 2002. Discourse markers in early modern English. In: Watts, R., Trudgill, P. (Eds.), Alternative Histories of English. Routledge, London & New
York, pp. 210–230.
Jucker, A.H., Ziv, Y., 1998. Discourse markers: introduction. In: Jucker, A.H., Ziv, Y. (Eds.), Discourse Markers: Descriptions and Theory. John Benjamins,
Amsterdam, pp. 1–12.
Kaltenböck, G., 2006. ‘...That is the question’: complementizer omission in extraposed that-clauses. Engl. Lang. Ling. 10 (2), 371–396.
Kaltenböck, G., 2007. Spoken parenthetical clauses in English. In: Dehé, N., Kavalova, Y. (Eds.), Parentheticals, Linguistics Today, vol. 106. Benjamins,
Amsterdam & Philadelphia, pp. 25–52.
Kaltenböck, G., 2008. Prosody and function of English comment clauses. Folia Ling. 42 (1), 83–134.
Kaltenböck, G., 2011. Explaining diverging evidence. The case of clause-initial I think. In: Schönefeld, D. (Ed.), Converging Evidence: Methodological and
Theoretical Issues for Linguistic Research. Benjamins, Amsterdam, pp. 81–112.
Kaltenböck, G., 2013. The development of comment clauses. In: Aarts, B., Close, J., Leech, G., Wallis, S. (Eds.), The Verb Phrase in English: Investigating Recent
Language Change with Corpora. Cambridge University Press, Cambridge, pp. 286–317.
Kaltenböck, G., Heine, B., Kuteva, T., 2011. On thetical grammar. Stud. Lang. 35 (4), 848–893.
Kärkkäinen, E., 2003. Epistemic Stance in English Conversation: A Description of its Interactional Functions, with a Focus on I Think. Benjamins, Amsterdam
& Philadelphia.
Kearns, K., 2007. Epistemic verbs and zero complementizer. Engl. Lang. Ling. 11 (3), 475–505.
Knowles, J., 1980. The tag as a parenthetical. Stud. Lang. 4 (3), 379–409.
Kranich, S., 2015. The impact of input and output domains: towards a function-based categorization of types of grammaticalization. Lang. Sci. 47, 172–187.
Kuteva, T., Heine, B., Hong, B., Long, H., Narrog, H., Rhee, S., 2019. World Lexicon of Grammaticalization, 2nd extensively revised and updated edition.
Cambridge University Press, Cambridge.
Lehmann, C., 2004. Theory and method in grammaticalization. Z. Ger. Linguist. 32 (2), 152–187.
Lehmann, C., 2015[1982]. Thoughts on grammaticalization. In: Classics in Linguistics, third ed., vol. 1. Language Science Press, Berlin.
Lenk, U., 1998. Marking Discourse Coherence: Functions of Discourse Markers in Spoken English. Narr, Tübingen.
Lenker, U., 2000. Soþlice and witodlice: discourse markers in Old English. In: Fischer, O., Rosenbach, A., Stein, D. (Eds.), Pathways of Change. Gramma-
ticalization in English. John Benjamins, Amsterdam, Philadelphia, pp. 229–249.
Lewis, D.M., 2011. A discourse-constructional approach to the emergence of discourse markers in English. Linguistics 49 (2), 415–443.
López-Couso, M.J., 1996. On the history of methinks: from impersonal construction to fossilized expression. Folia Linguist. Hist. 17, 153–169.
López-Couso, M.J., Méndez-Naya, B., 2014. From clause to pragmatic marker: a study of the development of like parentheticals in American English. J. Hist.
Pragmat. 15 (1), 36–61.
Nelson, G., Wallis, S., Aarts, B., 2002. Exploring Natural Language. Working with the British Component of the International Corpus of English. Benjamins, ,
Amsterdam & Philadelphia.
Noonan, M., 1985. Complementation. In: Shopen, T. (Ed.), Language Typology and Syntactic Description, Complex Constructions, vol. 2. Cambridge Uni-
versity Press, Cambridge, pp. 42–140.
Norde, M., 2009. Degrammaticalization. Oxford University Press, Oxford.
Ocampo, F., 2006. Movement towards discourse is not grammaticalization: the development of /claro/ from adjective to discourse particle in spoken
Spanish. In: Sagarra, N., Toribio, A.J. (Eds.), Selected Proceedings of the 9th Hispanic Linguistics Symposium. Cascadilla Proceedings Project, Somerville,
MA, pp. 308–319.
Palander Collin, M., 1999. Grammaticalization and Social Embedding: I THINK and METHINKS in Middle and Early Modern English (Mémoires de la Société
Néophilologique, Tome LV). Société Néophilologique, Helsinki.
Peltola, N., 1983. Comment clauses in present-day English. In: Kajanto, I., et al. (Eds.), Studies in Classical and Modern Philology. Suomalainen, Tiedea-
katemia, Helsinki, pp. 101–113.
16 B. Heine, G. Kaltenböck / Language Sciences 87 (2021) 101400

Prévost, S., 2011. A propos from verbal complement to discourse marker: a case of grammaticalization? Linguistics 49 (2), 391–413.
Quirk, R., Greenbaum, S., Leech, G., Svartvik, J., 1985. A Comprehensive Grammar of the English Language. Longman, London & New York.
Rissanen, M., 2008. From ‘quickly’ to ‘fairly’: on the history of rather. Engl. Lang. Ling. 12 (2), 345–359.
Roberts, I., Roussou, A., 2003. Syntactic Change: A Minimalist Approach to Grammaticalization. Cambridge University Press, Cambridge.
Ruiz de Mendoza Ibáñez, F.J., Luzondo Oyón, A., Pérez Sobrino, P. (Eds.), 2017. Constructing Families of Constructions. Analytical Perspectives and Theoretical
Challenges. John Benjamins, Amsterdam.
Schiffrin, D., 1987. Discourse markers. In: Studies in Interactional Sociolinguistics, vol. 5. Cambridge University Press, Cambridge.
Schneider, S., 2007. Reduced Parenthetical Clauses as Mitigators: A Corpus Study of Spoken French. Italian and Spanish. Benjamins, Amsterdam &
Philadelphia.
Schourup, L.C., 1999. Discourse markers. Lingua 107, 227–265.
Shank, C., Van Bogaert, J., Plevoets, K., 2016. The diachronic development of zero complementation: a multifactorial analysis of the that/zero alternation
with think, suppose, and believe. Corpus Linguist. Linguistic Theory 12 (1), 31–72.
Tabor, W., Traugott, E.C., 1998. Structural scope expansion and grammaticalization. In: Ramat, G., Hopper, A., Hopper, P.J. (Eds.), The Limits of Gramma-
ticalization. Benjamins, Amsterdam & Philadelphia, pp. 229–272.
Thompson, S.A., 2002. “Object complements” and conversation: towards a realistic account. Stud. Lang. 26, 125–164.
Thompson, S.A., Mulac, A., 1991. A quantitative perspective on the grammaticization of epistemic parentheticals in English. In: Traugott, E.C., Heine, B. (Eds.),
Approaches to Grammaticalization, vol. 2. Benjamins, Amsterdam & Philadelphia, pp. 313–329.
Traugott, E.C., 1995. The role of the development of discourse markers in a theory of grammaticalization. In: Paper Presented at the International Conference
of Historical Linguistics XII, Manchester 1995.
Traugott, E.C., 2007. Discussion article: discourse markers, modal particles, and contrastive analysis, synchronic and diachronic. Catalan J. Linguist. 6, 139–
157.
Traugott, E.C., 2018. Modeling language change with constructional networks. In: Pons Bordería, S., Loureda Lamas, O. (Eds.), Beyond Grammaticalization
and Discourse Markers: New Issues in the Study of Language Change, Studies in Pragmatics, vol. 1. Brill, Leiden & Boston, pp. 17–50.
Traugott, E.C., Dasher, R.B., 2002. Regularity in semantic change. In: Cambridge Studies in Linguistics, vol. 96. Cambridge University Press, Cambridge.
Traugott, E.C., Trousdale, G., 2013. Constructionalization and Constructional Changes. Oxford University Press, Oxford.
Urmson, J.O., 1952. Parenthetical verbs. Mind, New Series 61 (244), 480–496.
Van Bogaert, J., 2009. The Grammar of Complement-Taking Mental Predicate Constructions in Present-Day Spoken British English (Doctoral dissertation
University of Gent).
Van Bogaert, J., 2011. I think and other complement-taking mental predicates: a case of and for constructional grammaticalization. Linguistics 49 (2), 295–
332.
Vandelanotte, L., 2006. Speech or thought representation and subjectification, or the need to think twice. Belg. J. Linguist. 20 (1), 137–1608.
Vandelanotte, L., 2009. Speech and Thought Representation in English. A Cognitive-Functional Approach. Mouton de Gruyter, Berlin, New York.
Waltereit, R., 2002. Imperatives, interruption in conversation and the rise of discourse markers: a study of Italian guarda. Linguistics 40, 987–1010.
Waltereit, R., 2006. The rise of discourse markers in Italian: a specific type of language change. In: Fischer, K. (Ed.), Approaches to Discourse Markers, Studies
in Pragmatics, vol. 1. Elsevier, Amsterdam, pp. 61–76.
Wichmann, A., 2011. Grammaticalization and prosody. In: Narrog, H., Heine, B. (Eds.), The Oxford Handbook of Grammaticalization. Oxford University Press,
New York, pp. 331–341.
Wischer, I., 2000. Grammaticalization versus lexicalization: ‘Methinks’ there is some confusion. In: Fischer, O., Rosenbach, A., Stein, D. (Eds.), Pathways of
Change. Grammaticalization in English. John Benjamins, Amsterdam & Philadelphia, pp. 355–370.
Zwicky, A.M., 1985. Clitics and particles. Language 61, 283–305.

You might also like