Speaker-orientation meaning and positional shifts of discourse structuring markers

Full Text

1. Introduction

Discourse markers (DMs) in general have long been known to have strong preference of occurrence position, favoring the clause-initial position (Traugott 2018, Heine et al. 2021). The preference of initial position is often so robust that Keller (1979: 222) lists it among the three defining characteristics of DMs. However, numerous recent studies report that DMs show greater positional variability (Heine et al. 2021: 89–90). Similarly, discourse structuring markers (DSMs), a subset of pragmatic markers consisting of monofunctional DSMs and multifunctional DMs (Traugott 2022: 5), the “connectors that allow the speaker/writer to signal what relationship they wish the addressee/reader to deduce from the linking of discourse segments in a non-subordinate way” (Traugott 2022: 4), also typically occur at the clause-initial position (Traugott 2022: 16).

Of particular interest is that while it is true that DSMs typically emerge at the clause-initial position, some of them also occur at the clause-medial position, whereas some of them only occur at the clause-initial position. Furthermore, the positional variability is not arbitrary, i.e., DSMs do not randomly occur at any morphosyntactic position preserving their functions (see section 2 for more). From this state of affairs, there arises a question as to whether certain DSMs have indeed undergone positional shifts from their clause-initial position in the course of their development. A large body of research addressing positionality of DMs and DSMs notwithstanding, the issue of what enables positional variability has not been studied in earnest to date. Thus, the present study intends to fill this research gap by answering the following two research questions: (a) Is there indeed an initial-to-medial change attested in the development of some DSMs?, and (b) If the answer to question (a) is “yes”, how do we account for the change, i.e., what enables such positional shifts? The answers to these questions will be sought by investigating the diachronic changes of some Chinese DSMs in this study. The goal of this research is to identify the factors involved in DSM positionality from the diachronic investigation of the DSMs and their contemporary distributional patterns.

In order to pursue the answers to the two key questions, this paper collected historical and contemporary data from various sources in Chinese, paying special attention to the data used by the DM and DSM researchers. Similarly, for discussion of comparable phenomena in English, the data were collected from historical and contemporary corpora as well as earlier research. The data cited in discussion of potential crosslinguistic validity were taken from the reliable native-speaker informants from field work (see section 3 for more).

This study is structured as follows: Section 2 discusses the theoretical background. Section 3 briefly describes data collection. Based on the collected data, Section 4 illustrates two groups of Chinese DSMs that developed in the clause-initial position. It argues that the shènzhì (甚至) ‘even’ group expressed speaker-oriented meanings (and thus are also speaker-oriented adverbials) and acquired the clause-medial usage in later stages of development; the bùguò (不过) ‘however, but’ group did not indicate speaker-oriented meanings (and thus are not speaker-oriented adverbials) and did not acquire the usage in later stages of development. Based on the results of the analysis of Chinese DSMs in Section 4, Section 5 examines the enabling factors of positional shifts and expands the scope to other languages in order to ascertain the potential crosslinguistic applicability of the analysis. Drawing upon Long et al. (2022), we propose a hypothetical initial-to-medial change to account for the similar positional shifts of DSMs expressing speaker-oriented meanings in Chinese and English, further making brief reference to other languages. Section 6 offers the conclusion.

2. Theoretical background

In her seminal work, Traugott (2022) offers a comprehensive analysis on the development of DSMs. Among a number of DSMs in English addressed in the work, but, parenthetically, and back to your point are exemplified in (1):

(1) a. Countries separate, they break relations, they … leaders don’t speak to each other. But there’s always a coming back together.(ABC Nightline 1990 [COHA BLOG], quoted in Traugott 2022: 4)
     b. I just typed way too much so I’ll stop here. Parenthetically, I saw a short interview you did.(Blog maverick 2012 [COCA], quoted in Traugott 2022: 151)
     c. He was designing the B–1 bomber at Wright-Patterson Air Force Base. But back to your point. So customers are demanding of these authentic flavors.(NPR Here and Now 2014 [COCA], quoted in Traugott 2022: 160)

In the present research, the notion of ‘speaker-oriented adverbials’ (SpOAs) plays a crucial role in DSM semantics. Following Ernst (2009: 497–502), we use SpOAs to refer to three groups of adverbials:^[1] (i) discourse-oriented adverbials that “modify propositions and can be paraphrased by ‘I say ADV that P’” (p. 498; see English honestly in (2a)), (ii) modal adverbials “which indicate, roughly speaking, speakers’ degree of commitment to the truth of P” (p. 498; see English probably in (2b)), and (iii) evaluative adverbials “which represent speakers’ evaluation of the fact represented by P” (p. 498; see English luckily in (2c)) (for other similar classifications of SpOAs, see Ifantidou (1994: 130–155) and Brinton (2008: 8)).

(2) a. Honestly, I don’t know what you mean.(Quoted in Ernst 2009: 498)
     b. Karen is probably going to dance a tango.(Quoted in Ernst 2009: 498)
     c.  Luckily, Aaron did not fall off his bicycle.(Quoted in Ernst 2009: 498)

Brief remarks on DSMs and SpOAs are in order. These two closely-related terms are not mutually exclusive. One may find, for example, that some linguistic structures may serve to connect two linguistic segments (and thus are DSMs) and at the same time indicate speaker-oriented meanings (and thus are SpOAs);
see (3a–c) for the examples of English even, Chinese shènzhì (甚至) ‘even’, and Korean simcie ‘even’.

(3)   a     English
               Mary got a full fellowship from State U. They’re even paying her way out to visit the Department.
               (Quoted in Kay 1990: 74)

        b        Chinese圣如孔子，还假装生病，哄走了儒悲，孟子甚至对齐宣王也撒谎装病。

Shèng	rú	Kǒngzǐ,	hái	jiǎzhuāng	shēngbìng,	hǒng
sage	like	Confucius	still	pretend	be.sick	coax

zǒu	le	Rúbēi,	Mèngzǐ	shènzhì	duì	Qí
go	pfv	Rubei	Mencius	even	to	Qi

Xuānwáng	yě	sāhuǎng	zhuāng	bìng.
King.Xuan	also	lie	pretend	be.sick

‘A sage like Confucius still pretended to be sick and coaxed Rubei to go; Mencius even lied to King Xuan of Qi Nation and pretended to be sick.’
(Wéi Chéng [Fortress Besieged], quoted in Liu 2012: 256)

c     Korean
       Ku-nun   yelsimhi     ilha-y        sungcin-ul           ha-yss-ko     hwuey   ku-nun
      he-top    seriously   work-csl  promotion-acc   to-pst-and   later     he-top

    simcie     hoycang-i    toy-ess-ta.
    even       CEO-nom   become-pst-decl
    ‘He worked hard and was promoted, and later he even became the CEO.’ (Seongha Rhee, personal communication)

Take English even (see (3a)) as an example. It is a DSM because it serves to connect two clauses and indicate the relationship between clauses. It is also a SpOA because it expresses a meaning that violates the expectation of the addressee/reader (see Kay 1990: 82–84). Without even, there will be no expectation-violating meaning, and it will also be difficult for the addressee/reader to deduce the exact relationship between the two clauses (see (4), a modified example of (3a)).²

(4)   Mary got a full fellowship from State U. They’re paying her way out to visit the Department.

The literature has conducted a lot of studies on the grammatical changes of linguistic structures that may be seen both as DSMs and SpOAs; see Traugott (1982, 1989, 1995, 2020a, 2020b, 2020c, 2022), Lewis (2000, 2007, 2011), Traugott & Dasher (2002: 152–175), Leuschner (2006), Brinton (2008: 203–218, 2017: 251–283), Lenker (2010, 2014), Heine (2013), Haselow (2015), Heine et al. (2021), and others. However, there are few authors discussing the correlation between position differences (see the examples of clause-initial DSMs in (1a–c) and examples of clause-medial DSMs in (3a–c)) and the expression of speaker-oriented meanings in these structures.

Traugott (1982, 1989, 1995, 2020a, 2020b, 2020c, 2022), Lewis (2000, 2007, 2011), Traugott & Dasher (2002: 152–175), Leuschner (2006), Brinton (2008:
203–218, 2017: 251–283), Lenker (2010, 2014), Heine (2013), Haselow (2015), Heine et al. (2021), and others study the development of a number of English DSMs from circumstance adverbials and argue that the development typically took place in the initial position of the clauses that they connect.³ According to Traugott (1995: 9–10), for example, English in fact was originally a circumstance adverbial meaning ‘in actuality’ (see (5a)). It developed into a SpOA expressing counter-expectation meaning in the clause-initial position (see (5b)), and later into a DSM that is “primarily elaborative in function” in the same position (see (5c)).⁴

(5)  a   But it is evident in fact and experience that there is no such universal Judge, appointed by God over the whole World, to decide all Cases of temporal Right.(Sermons Tillotson. 1671: 445 [HC], quoted in Traugott 1995: 9–10)
      b  Whence comes it then, that, in fact, the utmost a wise magistrate can propose with regard to popular religions, is, as far as possible, to make a saving game of it ...?(Hume Dial. Nat. Relig. Pt.12, publ. 1779: 223, quoted in Traugott 1995: 10)
      c   I should not have used the expression. In fact, it does not concern you – it concerns only myself.(Austen, Emma, Vol. 3, Chap. 10, 1815: 393, quoted in Traugott 1995: 10)

Dik (1997: 406–408) has suggested a Relator Principle (see the following) to account for the positional preferences of grammatical elements “which serve to link two constituents together, and/or to mark the function(s) of a constituent”. Chu & Tao (2008), Dong (2012), Peng (2012), Fan (2015), Gao (2016), Shi & Hao (2018), Li (2020), Yu (2020), and others have extended the principle to account for the positional preference of DSMs in Chinese and the languages in the region of China.⁵

Relators have their preferred position (i) in between two relata; (ii) at the periphery of the relatum with which they form one constituent (if they do so). (Dik 1997: 406)

The Relator Principle may explain the positional preference of the English DSM in fact in (5c) because its preferred position (i.e., the initial position of one of the clauses that it connects) may also be explained as between two clauses (or two relata). What it cannot explain, however, is the fact that apart from the clause-initial position, some English DSMs may also appear in a clause-medial position; see contrasted pairs in (6a–b), (7a–b), and (8a–b).

(6)    a    The umpires’ decision is final. Therefore, so far as the ICC is concerned, the matter is closed.
            (The Daily Mirror. 5437 s-units [BNC])
        b This necessarily entails longer term assistance in comparatively stable situations. We therefore particularly value our partnership with SCF through TRANSAID ...
            (Transport. u.p., n.d., pp. ??. 645 s-units [BNC])

(7)     a     We have the additional problem that the shops in Saint Mary’s Street, with the majority of shops in Saint Mary’s Street, that is the main street, have no rear accesses. Consequently heavy lorries make deliveries and             other heavy lorries passing through cause congestion.
            (Suffolk County Council Highways and Transport Committee meeting
            (Pub/instit). Rec. on 17 Nov 1992 with 2 partics, 746 utts [BNC])
        b     In February 1934 Nizan was in the Soviet Union. He consequently witnessed the February days, the potential civil war at a distance.
        (Paul Nizan communist novelist. Scriven, Michael. London: Macmillan Press Ltd, 1988: 1–120. 1550 s-units [BNC])

(8)     a     As a result of all this, Members know that any effort which goes into the preparation of a draft Bill prior to the ballot is likely to be wasted. In consequence, many Members who put their names down will have only illformed ideas about what they want and it frequently happens that Bills presented are ill-prepared or misconceived.
        (An introduction to British constitutional law. Calvert, Harry. London: Blackstone Press, 1985: 43–148. 1543 s-units. [BNC])
        b     Employers will be given a yardstick against which to measure teachers. Parents will, in consequence, have more faith in state schools.
        (What is happening to our primary schools. Pluckrose, Henry. Oxford: Blackwell, 1987: 1–138. 1757 s-units. [BNC])

Interestingly, even a developed DSM like in fact, which preferably occupies a clause-initial position according to the Relator Principle, may also appear in a clause-medial position (see (9)).[6]

(9)   I should not have used the expression. It in fact does not concern you – it concerns only myself

One may notice that one reason why Traugott (2022) rejected the terms including pragmatic marker (see Brinton 1996: 33, 2017: 2–8, Fraser 1996: 167, Aijmer 2013: 4, and others) and discourse marker (see Blakemore 1987: 141, Schiffrin 1987: 31, Fraser 1999: 931, Heine 2013: 1206–1213, and others) in favor of the term DSM was to emphasize the significance of the positional differences of these structures (see Traugott 2022: 61–63). Adopting a different nomenclature, however, does not solve the obvious problem, i.e., Traugott (2022) still cannot fully account for the clause-medial occurrence of some DSMs like English in fact in (9). To account for it, one may need to postulate that there has occurred an initial-to-medial change in the course of the development.

In this study, by examining the historical developments of some Chinese and English DSMs, as well as some incidental examples of DSMs in other languages, we take on an unresolved issue in Traugott (2022, n.d.), that is, for some DSMs that developed in the clause-initial position, why do they occupy the clause-medial position in later stages of development?

3. Data sources

The present study has adopted data (including historical data) from Chinese, English, and other languages. Era divisions of Chinese data have followed Sun (2006: 15–20), which are: Old Chinese (771 BCE to 220), Middle Chinese (220 to 960), Early Modern Chinese (960 to 1900), and Modern Chinese (1900 to present). We use the label “historical Chinese” to refer to any period of Chinese that is not Modern Chinese. All the data on historical Chinese cited in this study are based on previous studies on the same topics.

English data cited in this study has five sources: (a) previous studies on the same topics, (b) the British National Corpus (BNC), which contains 100 million words of text from a wide range of genres (e.g., spoken, fiction, magazines, newspapers, and academic), (c) the Corpus of Contemporary American English (COCA), which contains more than one billion words of text (over 25 million words each year from 1990 to 2019) from eight genres (spoken, fiction, popular magazines, newspapers, academic texts, TV and movies subtitles, blogs, and other web pages), (d) the Corpus of Historical American English (COHA), which contains more than 475 million words of text from the 1820s to the 2010s, and (e) the Oxford English Dictionary (OED), which contains over 600,000 entries of the English language. It provides detailed definitions, word origins, and historical usage examples, and is an essential resource for understanding the evolution and nuances of English.

Unless specifically indicated otherwise, data on DSMs in languages other than Chinese and English were collected by the first author through fieldwork. All the informants for the data are experienced linguists who have had more than five years of experience studying the languages consulted. The names of the informants are indicated in parentheses after each sentence example. The sentence example of historical Korean (see (23a)) should be particularly mentioned because it was not collected through fieldwork; rather, it was adopted from AKORN (https://akorn.bab2min.pe.kr), an online resource platform dedicated to providing Korean literature and ancient texts. The site includes several key features: digitized versions of numerous ancient Korean texts, making them accessible for researchers and enthusiasts; modern Korean dictionaries, grammar books, and other resources to aid in the learning and understanding of the Korean language; multiple language support, facilitating use by international users; and a wealth of scholarly documents and materials, beneficial for academics engaged in Korean studies.

4. Result: Two groups of Chinese DSMs in clause-initial position

The data collection from the historical sources as well as existing literature and their analysis renders significant results as to the development of DSMs. A number of DSMs in Modern Chinese may have developed in the clause-initial position. According to Fang & Jiang (2009) and Liu (2012), the Modern Chinese DSM shènzhì (甚至) ‘even’ developed from a predicate meaning ‘so hard/serious as to reach (the level/position of)’; see (10a) and (10b). It originally took an NP as its complement in Old Chinese (see (10a)), and later developed to take a predicate as its complement in Early Modern Chinese (see (10b)).

(10)    a    肾不生，则髓不能满，故寒甚至骨也。

Shèn

bù

shēng,

zé

suǐ

bù

néng

mǎn,

kidney

neg

live

thus

bone.marrow

neg

can

be.full

gù		hán	shèn		zhì		gǔ			yě.
so		cold	hard		reach		bone			fp

‘If there are no kidneys, there will be inadequate marrow in the bones, and the coldness will be so severe as to reach the position of the bones.’
(Huángdì Nèijīng, Sùwèn [Huangdi Neijing, Suwen], quoted in Liu 2012: 250; no later than the first century CE)

       b    若或父母坚不从所谏，甚至怒而挞之流血。

Ruò

huò

fùmǔ

jiān

bù

cóng

suǒ

jiàn,

if

sometimes

parents

firmly

neg

follow

nmlz

advise

shèn	zhì			nù		ér		tà	zhī		liúxuè.
hard	reach			be.angry		and		beat	him		bleed

‘Sometimes the parents insistently rejected his advice. (The situation became) so difficult as to reach the level that they beat him until he bled.’
(Zhūzǐ Yǔlèi, Dì Èrshíqī Juàn [Zhuzi’s Language Category, Vol. Twenty-Seven], quoted in Liu 2012: 250; 1270)

In the clause-initial position, it developed into a DSM with speaker-oriented meanings (thus also a SpOA) in Early Modern Chinese no later than the early 17^th century; see (11). In an example like (11), the speaker (narrator) uses shènzhì to indicate that the condition described by the following clause (i.e., cháotíng gōngwéi zhīzhōng yǒushí zhàoyòng ‘he was sometimes summoned to the court and inner palace’) is deemed to be against the principle of conduct by the speaker/writer (for similar arguments, see Liu (2012: 252–253)).

(11)        所以公卿大夫都有信着他的，甚至朝廷宫闱之中有时召用。

Suǒyǐ	gōngqīng dàfū		dōu	yǒu	xìnzhe	tā	de,
therefore	high-ranking.official		all	have	trust	him	nmlz

shènzhì	cháotíng	gōngwéi	zhī	zhōng	yǒushí	zhàoyòng.
even	Court	inner.palace	poss	inside	sometimes	summon

‘Therefore, the nobles and high-ranking officials all trusted him, and he was even sometimes summoned to the court and inner palace.’
(Chūkè Pāi’àn Jīngqí, Dì Sānshíjiǔ Juàn [First Series of Marvelous Tales at the Pavilion, Chapter Thirty-Nine], quoted in Liu 2012: 253; the early 17^th century CE)

In Modern Chinese, the DSM and SpOA shènzhì has developed to occupy a clause-medial position; see (12) (= (3b)).

(12)   圣如孔子，还假装生病，哄走了儒悲，孟子甚至对齐宣王也撒谎装病。

Shèng	rú	Kǒngzǐ,	hái	jiǎzhuāng	shēngbìng,	hǒng
sage	like	Confucius	still	pretend	be.sick	coax

zǒu	le	Rúbēi,	Mèngzǐ	shènzhì	duì	Qí
go	pfv	Rubei	Mencius	even	to	Qi

Xuānwáng	yě	sāhuǎng	zhuāng	bìng.
King.Xuan	also	lie	pretend	be.sick

‘A sage like Confucius still pretended to be sick and coaxed Rubei to go; Mencius even lied to King Xuan of Qi Nation and pretended to be sick.’
(Wéi Chéng [Fortress Besieged], quoted in Liu 2012: 256)

This kind of change is rather common in Chinese. Our investigation reveals that the following DSMs express speaker-oriented meanings (thus are also SpOAs); see (13), as indicated by their glosses. They have developed in the clause-initial position, and have all occupied a clause-medial position in later stages of development.

(13)	a	gùrán (固然)	‘no doubt’	(Li 2017: 106)
	b	guàibùdé (怪不得)	‘no wonder’	(Jiang & Luo 2019: 60–62)
	c	guǒbùqírán (果不其然)	‘it really happens, as expected’	(Ye 2016: 193–200)
	d	guǒrán (果然)	‘it really happens, as expected’	(Long et al. 2022)
	e	hékuàng (何况)	‘not to say’	(Li 2014)
	f	jìrán (既然)	‘now that’	(Jiang 2010: 103)
	g	nánguài (难怪)	‘no wonder’	(Xie & Zuo 2009: 30–31)
	h	suīrán (虽然)	‘although’	(Xu & Jiang 2010)

Notice that not all the Chinese DSMs developed in a clause-initial position may later develop to occupy a clause-medial position. For instance, according to Shen (2004: 34) and He (2016), Chinese bùguò (不过) ‘however, but’ was first a verbal phrase meaning ‘not exceed’. It originally took an NP as a complement (see (14a)), and later developed to take a predicate as a complement (see (14b)) in Old Chinese.

 

(14)	a	内官不过九御，外官不过九品。
		Nèiguān	bù	guò	jiǔyù,	wàiguān	bù	guò	jiǔpǐn.
		inner.official	neg	exceed	ninth.grade	out.official	neg	exceed	ninth.rank
		‘The inner officials do not exceed the ninth grade, and the outer officials do not exceed the ninth ranks.’ (Guóyǔ, Zhōuyǔ [Discourses of the States, Discourse of Zhou State], quoted in He 2016: 104; the 5th century BCE to the 4th century BCE)
	b	公输子之意，不过欲杀臣。
		Gōngshūzǐ	zhī	yì,	bù	guò	yù	shā	chén.
		Gongshuzi	poss	intention	neg	exceed	want.to	kill	me
		‘Gongshuzi’s intention does not exceed killing me (is merely to kill me).’ (Mòzǐ, Gōngshū [Mozi, Gongshu], quoted in He 2016: 104; the 5th century BCE to the 4th century BCE)

Shen (2004: 36) argues that bùguò developed into a DSM meaning ‘but, however’ in the clause-initial position in Chinese around the early 20^th century (see (15)). It is noteworthy, however, that throughout the development, bùguò rarely expressed speaker-oriented meanings (thus was not a SpOA).

(15)  我也没有长策。不过这种事情，其势已迫，不能计出万全的。

Wǒ	yě	méiyǒu		chángcè.		Bùguò	zhè	zhǒng	shìqíng,
I	also	not.have		long-term.plan		but	dem	kind	thing

qí	shì		yǐ	pò,	bùnéng	jì	chū	wànquán	de.
its	condition		already	urgent	can’t	solution	come.out	perfect	nmlz

‘I also do not have a long-term plan, but in this kind of situation, the matter is urgent, and it is impossible to come up with a completely foolproof solution.’
(Lǎocán Yóujì, Dì Shíliù Huí [The Travels of Laocan, Chapter Sixteen], quoted in Shen 2004: 36; 1903)

In contrast with shènzhì and the other DSMs listed in (13), the DSM bùguò has never developed to occupy a clause-medial position in either historical Chinese or Modern Chinese. This pattern is common in Chinese. Our investigation reveals that the following DSMs, listed in (16), do not express speaker-oriented meanings (thus are not SpOAs). They have all developed in the clause-initial position but have not developed a clause-medial usage in either historical Chinese or Modern Chinese:

(16)	a	dànshì (但是)	‘however, but’	(He 2016)
	b	kěshì (可是)	‘however, but’	(Long & Zhou 2024)
	c	rán’ér (然而)	‘however, but’	(Zhu 2012)
	d	zhǐshì (只是)	‘however, but’	(He 2016)
	e	bùrán (不然)	‘otherwise’	(Wu 2021)
	f	érqiě (而且)	‘and, what’s more’	(Li, Liu & Jiang 2015)
	g	nàme (那么)	‘then’	(Liu 2024: 21–22)

The collected data of DSMs in two different categories, i.e., the shènzhì-type and the bùguò-type, raise a theoretically important issue on DSM positionality. In other words, if the shènzhì-type DSMs exhibit clause-initial and clause-medial occurrences whereas the bùguò-type DSMs are restricted to the clause-initial position, one may wonder what factors may contribute to the positional differences of these two groups of Chinese DSMs. We will come to this in the next section.

5. Discussion: Initial-to-medial change of speaker-oriented DSMs

In Section 4, we have described the diachronic changes of two groups of Chinese DSMs developed in the clause-initial position: one group, exemplified by the DSM shènzhì, expressed the speaker-oriented meanings (thus are also SpOAs) and may occupy the clause-medial position in later stages of development; the other group, exemplified by the DSM bùguò, did not express the speaker-oriented meanings (thus are not SpOAs) and have not occupied the clause-medial position in later stages of development. If our arguments in Section 4 are right, one may naturally hypothesize that DSMs developed in the clause-initial position may occupy the clause-medial position in later stages of development if they express speaker-oriented meanings (thus are SpOAs).

If the expression of speaker-oriented meanings (i.e., thus also being SpOAs) is indeed an important factor triggering the initial-to-medial changes of DSMs developed in the clause-initial position, one might wonder why this is the case. To address this issue, one may need to re-examine the typical positions for the occurrence of SpOAs. The literature generally assumes that SpOAs can occur in multiple positions within a clause, with the clause-initial and clause-medial positions being the most typical (see examples (17a) and (17b); also refer to Quirk et al. (1985: 490–491), Hoye (1997: 148), Biber et al. (1999: 892), Clift (2001), Lenker (2014), Suzuki & Fujiwara (2017: 831), and others, for similar arguments).

(17)   a      Possibly they may have been sent to London.
                  (Quoted in Hoye 1997: 148)
         b     They possibly may have been sent to London.
                 (Quoted in Hoye 1997: 148)

The literature also generally argues that SpOAs developed in a clause-initial position; see Traugott (1995, 2022), Traugott & Dasher (2002: 152–175), Brinton (2008: 203–218, 2017: 251–283), Haselow (2015), and Heine et al. (2021), among others. If this is indeed the case, one might naturally wonder whether SpOAs developed in this typical position should further evolve to occupy the other typical position, namely, the clause-medial position. Long et al. (2022) argue that they indeed do. The authors argue that, cross-linguistically speaking, SpOAs developed in the clause-initial position may also occupy the clause-medial position in later stages of development. Take the English SpOA perhaps as an example (see (18a–c)). According to Molencki (2021: 417), it developed from a circumstance adverbial meaning ‘by chance’ in Middle English (see (18a)), and the whole change took place in the clause-initial position (see (18b)). In Modern English, it may also occupy a clause-medial position (see (18c)).

(18)	a	Parhap y wrussen fye. ‘Perhaps I might flee.’ (Origo Mundi in Norris Anc. Corn. Drama (Bod 791)1352, ? a1450 (? 1350–1375), quoted in Molencki 2021: 417 and Long et al. 2022: 231)
	b	Perhaps some Merchant hath inuited him. (Shakespeare Com. Err. ii.i.4, 1590, quoted in Molencki 2021: 417 and Long et al. 2022: 231)
	c	Horatio perhaps lost his mind. (Quoted in Long et al. 2022: 231)

We have argued in Section 4 that the first group of Chinese DSMs all expressed speaker-oriented meanings and should also be seen as SpOAs.⁷ Considering the SpOA status of these Chinese DSMs, one may naturally expect them to follow the hypothetical initial-to-medial pathway like the other SpOAs discussed in Long et al. (2022). This also explains why the English DSM in fact, which typically occupies a clause-initial position (see (19a) = (5c)), may also occupy a clause-medial position (see (19b) = (9)) in later stages of development.⁸

(19)	a	I should not have used the expression. In fact, it does not concern you – it concerns only myself. (Austen, Emma, Vol. 3, Chap. 10, p. 393, 1815, quoted in Traugott 1995: 10)
	b	I should not have used the expression. It in fact does not concern you – it concerns only myself.

On the other hand, one may also find that there are English DSMs that do not express speaker-oriented meanings. If they developed in a clause-initial position, following the arguments above, one will naturally expect them not to acquire the clause-medial usage in later stages of development. Traugott (2022: 159–160), for example, argues that the English back to X point (see (20b)) developed from the phrase to come back to the point (see (20a)) in the clause-initial position. Confirming our expectation, this DSM has not acquired a clause-medial usage either in historical English or Modern English.

(20)	a	The great body of people in every country are idolaters. They worship the image or form rather than God, the living principle of goodness. But, to come back to the point, I believe God does mark (‘note’) the conduct of men. (New Englander and Yale Review [COHA], Traugott 2022: 159, 1843)
	b	He was designing the B–1 bomber at Wright-Patterson Air Force Base. But back to your point. So customers are demanding of these authentic flavors. (NPR Here and Now [COCA], Traugott 2022: 159, 2014)

In the present research, we have looked into the developments of other English DSMs that developed in a clause-initial position and have found that if they did not express speaker-oriented meanings, they usually did not acquire a clause-medial usage in the later stages of development. The interesting part is with English DSMs like incidentally and parenthetically (see (21a) and (21b)). According to Traugott (2022: 150), these two DSMs have “very sparse” examples of being used to express speaker-oriented meanings, and they almost exclusively appear in the clause-initial position in Modern English (see (21a) and (21b)).

(21)	a	“They talked about refrigerators, clearly.” Misha rolled his eyes at Arkady. “Incidentally, you don’t happen to know any murderer repairmen who owe you a favor?” (Smith, Gorky Park [COHA], quoted in Traugott 2022: 150, 1981)
	b	I just typed way too much so I’ll stop here. Parenthetically, I saw a short interview you did. (Blog maverick [COHA BLOG], quoted in Traugott 2022: 151, 2012)

As indicated above in Section 2, numerous researchers assert that the clause-initial position is a typical position for the development of DSMs. Also as briefly alluded to in Section 2, according to the Relator Principle, the clause-initial position is the typical position for the occurrence of DSMs; similar arguments can be found in Lehmann (1988), Kortmann (1998: 545), Ghezzi & Molinelli (2014), Tsunoda (2018), and others. It seems safe to assume that cross-linguistically speaking, the clause-initial position may be a preferred position for both the development and the occurrence of DSMs.

On the other hand, DSMs occupying the clause-medial position are also attested in a number of languages; see (22a–g) for examples of DSMs occupying a clause-medial position in German, Italian, Greek, Hungarian, Korean, Liujiang Zhuang, and Jieyang dialect of Southern Min.

(22)   a   German

Er     ist             reich,   das       ist             aber   nicht         wahr.

3sg    cop.3sg    rich      dem      cop.3sg    but     neg          true

‘He is rich, this however is not true.’ (Bernd Heine, fieldwork)

b    Italian

È          ricco.      Questo      peró          forse     non    è             vero.

be.3sg  rich        this           however    maybe  not     be.3sg    true

‘He’s rich. This however maybe is not true.’ (Francesco Ursini, fieldwork)

c   Greek

Aftos    ine          plusios,     ine            omos         poli           filikos.

3sg      be.3sg    rich           be.3sg      however    very          friendly

‘He is rich; he is however very friendly.’ (Brian Joseph, fieldwork)

d    Hungarian

Hatalmas   zivatar    kerekedett    és      mi    persze         bőrig          áztunk.

large         storm      arose.3sg     and    we   of.course   skin.till       soaked.1pl

‘A fierce storm arose and we of course got wet through.’

(Quoted in Vaskó 2012: 471)

e    Korean

Ku-nun      yelsimhi       ilha-y           sungcin-ul        ha-yss-ko     hwuey  ku-nun

he-top      seriously      work-csl     promotion-acc to-pst-and   later     he-top

simcie       hoycang-i       toy-ess-ta.

even         CEO-nom      become-pst-decl

‘He worked hard and was promoted, and later he even became the CEO.’ (Seongha Rhee, personal communication)

f    Liujiang Zhuang (Tai, Kra-Dai)

Lak⁸   ka:m¹    ni⁴      ʔi³        ʔdi¹     kɯn¹,      mɯŋ²    tsi⁶     ka:i⁵     kɯn¹     le⁵.

clf    citrus   dem   neg      good   eat         you      thus   don’t    eat       fp

‘The citrus is not delicious. You thus don’t eat it.’ (Fengyu Qin, fieldwork)

g   Jieyang dialect of Southern Min (Min, Sinitic, Sino-Tibetan)

I³³     sui³³ziaŋ^55-22   hoʔ^2-5    no^213-53lak⁵,       taŋ^35-21si^35-21             seŋ^55-22tseʔ²

3sg    although        very     endeavor          but                     score

tiam^22-21 tiam^22-21      mo⁴².

always                     not.good

‘Although he is very diligent, his scores are still not good.’ (Yanxuan Huang, fieldwork)

Based on the observations above, we may expand our scope further and attempt to obtain a crosslinguistically valid generalization. In other words, we may ask if crosslinguistically some of the DSMs that developed in the clause-initial position also express speaker-oriented meanings (and thus are also SpOAs), they would acquire a clause-medial usage in later stages of development. Considering that Chinese and English are genetically unrelated languages, if the above argument is true for Chinese and English, we may also expect it to be (at least partly) true for some other languages.

Confined by the limited data, we currently cannot give a definite answer to the question, but whenever there are historical data supporting the argument that a DSM from a language other than Chinese and English developed in a clause-initial position and at the same time indicated speaker-oriented meanings (and thus are also SpOAs), it indeed seems that we can always expect it to acquire a clause-medial usage in later stages of development. Take the Korean DSM simcie ‘even’ cited in (22e) (= (3c)) as an example. Our investigation of historical Korean reveals that it developed from an adverbial meaning ‘such that’ (see (23a)) into a DSM meaning ‘even’ (see (23b)) in the clause-initial position. It was also a SpOA because it exhibited a speaker-oriented meaning indicating that the condition described by the clause is beyond regular people’s assessment, as indicated by the translations of (23a) and (23b). Simcie has followed the hypothetical initial-to-medial change and has developed a clause-medial usage in Modern Korean (see (22e) = (3c)).

(23)     a    Syeyson-i                   cangin-ul                 tyoha-si-myen      chyengwen-ul          sayo-ko,

crown.prince-nom     father-in-law-acc  like-hon-if           [title]-acc                hate-and

simcie         syeyson-i                    Song           sansak-si-nola,

such.that      crown.prince-nom      Songsa        dissipate.eliminate-hon-purp

 paskuy    naka-si-myen         Song-ul          ta           sayo-ni…

outside   go.out-hon-if        Songsa-acc   all           hate-as

‘When the prince loved his father-in-law, she hated him. (Her jealousy was) such that if the prince goes outside in order to neutralize the bad effects from the bad words or characters occurring in Songsa (The History of Song Dynasty), (she) even hated all the volumes in Songsa, so....’ (Hancwunglok 384 [AKORN]; 1801–1805)

b   Ku-nun      yelsimhi       ilha-y          sungcin-ul           ha-yss-ta.         Simcie   ku-nun     

he-top      seriously      work-csl   promotion-acc  do-pst-decl     even     he-top    

hoycang-i         toy-ess-ta.

CEO-nom        become-pst-decl

‘He worked hard and was promoted. He even became the CEO.’ (Seongha Rhee, fieldwork)

Another piece of supporting data comes from Hungarian. The Hungarian SpOA persze may also be used as a DSM indicating concessive (see (24a)) or elaborative (see (24b)) meanings; see Furkó (2011) and Vaskó (2012). According to Vaskó (2012), the DSM and SpOA persze developed from the particle persze that serves to give an affirmative response to the previous utterance (see (25)), and this development occurred in the clause-initial position.

(24)

a

Nem

olyan

jó,

mint

a

könyv,

persze

azért

izgalmas.

not

so

good

as

the

book

of.course

still

exciting

‘It is not as good as the book, (but) still exciting, of course.’ (Quoted in Vaskó 2012: 475)

b

Ez

a

letiltott

csekk.

Nem

az

igazi

persze,

csak

másolat.

that

the

canceled

check

not

the

real

of.course

only

copy

‘That’s the canceled check. Not the real one, you know, it’s a copy.’ (House M. D. © NBC Universal Television; quoted in Furkó 2011: 103)

 

(25)	A:		Ismered know		azt that		a the		fickót? guy
	B:		Persze. of.course
	‘A:		Do you know that guy?’
	‘B:		Of course.’ (Quoted in Vaskó 2012: 468)

Furkó (2011: 100) and Vaskó (2012: 467) argue that the Hungarian DSM and SpOA persze may also appear in the clause-medial position (see (22d)). Although the two authors did not explicitly state it, we can still infer from their arguments that they support a hypothetical initial-to-medial change for the development of Hungarian DSM and SpOA persze.

If the above descriptions of Korean simcie and Hungarian persze are correct, it seems reasonable to assume that our arguments concerning positional shifts of clause-initial speaker-oriented DSMs may also be true in Korean and Hungarian. In any case, this is an issue that deserves further study, especially more thorough studies focusing on languages other than Chinese and English. However, at the current level of our understanding based on available data from a few languages, the correlation between speaker-oriented meaning of DSMs and their positional shift from the clause-initial to clause-medial position seems to be robust.

As a final remark, we would like to stress that our discussions have left one issue undiscussed, namely that the clause-initial position is not the only position where DSMs may develop. DSMs may also develop in other positions, including the medial position of a clause. For example, according to our investigations, the three English DSMs occupying both a clause-initial and a clause-medial position that we mentioned in Section 2 first acquired the discourse structuring usage in the clause-medial position; see (26a–c), respectively, for the early examples of the clause-medial DSMs therefore, consequently, and in consequence.

(26)	a	Ysmael pleide hard gamen; Sarra was ðor-fore often wroð. ‘Ysmael played hard games. Sarra was therefore often angry.’ (Gen. & Ex. 1215 [OED], c 1250)
	b	Vaccination had not been very extensively adopted among the poor, and there were consequently fewer cases in proportion, of the varioloid disease among the vaccinated, than in Scotland. (Small-Pox and Vaccination, North American Review: October 1821: 286–310 [COHA], 1821)
	c	In short, a thousand things may take place to prevent Maria from ever again seeing you. She, in consequence, sinks into an agonizing suspense concerning your fate. (The Wilderness; or, Braddock’s Times. A Tale of the West, Volume 1 [COHA], 1823)

These DSMs do not express speaker-oriented meanings, either. They have apparently developed the clause-initial usage in later stages of development. Since they did not develop from the clause-initial position, their developments do not contradict the generalizations drawn in this study. For a fuller understanding of DSM positionality more in-depth studies on multiple languages are warranted.

6. Conclusion

In this study, we have discussed the diachronic developments of two groups of Chinese discourse structuring markers (DSMs) that developed in the clause-initial position. We have shown that one group which expressed speaker-oriented meanings (and thus are SpOAs), exemplified by the DSM shènzhì (甚至) ‘even’, has acquired a clause-medial usage in later stages of development. On the other hand, the other group which did not express speaker-oriented meaning (and thus are not SpOAs), exemplified by the DSM bùguò (不过) ‘however, but’, has not acquired a clause-medial usage in later stages of development.

Based on our discussions of the developments of Chinese DSMs, we argued that the expression of speaker-oriented meaning is an important factor triggering the initial-to-medial changes of DSMs that developed in the clause-initial position, an analysis also borne out with the English DSMs. Furthermore, drawing upon the historical data in Korean and Hungarian, we have shown, albeit within limited scope of the data, that the generalization is applicable to the DSMs in these languages. Admittedly, the positionality of DSMs involves multifarious issues that cannot be easily resolved, and more concrete and crosslinguistically robust generalizations call for both microscopic and macroscopic studies across languages.

Abbreviations

The abbreviations in this study follow the standard Leipzig Glossing Rules (https://www.eva.mpg.de/lingua/pdf/Glossing-Rules.pdf). They are: 1 = first person; 3 = third person; acc = accusative; adv = adverbial; clf = classifier; cop = copula; csl = clause linker; decl = declarative; dem = demonstrative; fp = final particle; hon = honorific; neg = negation; nmlz = nominalizer; nom = nominative; pfv = perfective; pl = plural; poss = possessive; pst = past tense; purp = purposive; sg = singular; top = topic marker.

 

_{¹ Adverbial structures are notoriously amorphous, falling somewhere between being adverbial phrases and being adverbs. In this study, following Traugott (1995, 2022), Bussmann (2006: 1060), Swan & Breivik (2011), and Long et al. (2022), we use the term adverbials as a cover term for all forms of adverbial structures.}

_{² English even (see (3a)) and its counterparts in Chinese (i.e., shènzhì ‘even’, see (3b)) and Korean (i.e., simcie ‘even’, see (3c)) are “largely contentful/semantic” DSMs according to Traugott’s (2022: 3–4) classification. English parenthetically (see (1b)) also falls into this category according to the author.}

_{³ Following Traugott (2022: 70), we use circumstance adverbial to refer to an adverbial “with semantics of space, time, manner, instrument, etc.”}

_{⁴ It is important to note that the development of the DSM in fact, as illustrated from (5a) to (5c) and even to (9), is fundamentally based on Traugott (1995). We suspect that, although not explicitly stated, Traugott (2022) may have viewed this development differently. This is a significant issue, and we will reserve it for discussion in future publications.}

_{⁵ We thank Prof. Bernd Heine for rightfully pointing out that Dik’s (1997: 406–408) Relator Principle was not intended to account for the positional preference of DSMs. These authors’ arguments are actually extensions of the principle.}

_{⁶ One may wonder if the clause-medial in fact in (9) could be construed as a DSM. We argue that it conforms to Traugott’s (2022) definition of a DSM (see Section 1) and should be seen as a DSM; see Cliff (2001: 247), Aijmer (2015: 94), Alonso-Almeida & Álvarez-Gil (2021: 12), and others for arguments supporting clause-medial occurrence of English DSMs such as in fact, actually, etc., and see Lenker (2014: 21–25) and Traugott (2022: 212–213) for general arguments supporting clause-medial occurrence of English DSMs.}

_{⁷ In fact, when discussing the development of these DSMs, most linguists cited in (13) viewed them as SpOAs or as a hybrid of DSM and SpOA.}

_{⁸ Our arguments here may give the impression to the authors that we endorse the view that the DSM in fact developed in a clause-initial position and later moved to a clause-medial position. We admit that this is a plausible pathway, but we would like to emphasize that the issue calls for further investigation following the publication of Traugott (2022).}

About the authors

Haiping Long

Sun Yat-sen University

Author for correspondence.
Email: longhp@mail.sysu.edu.cn
ORCID iD: 0000-0003-2465-8723

professor at the School of Foreign Languages at Sun Yat-sen University, specializes in grammaticalization, linguistic typology, historical linguistics, and discourse grammar. He has authored over 50 papers in journals such as Studies in Language, Functions of Language, Linguistics, Journal of Historical Pragmatics and Language and Linguistics. He has led four national social science projects, including a major funding project, and has been awarded the Lü Shuxiang Linguistics Second Prize as well as a provincial award for Philosophy and Social Sciences.

Guangzhou, P.R. China

Lei Wang

Sun Yat-sen University; University of Konstanz

Email: lei.4.wang@uni-konstanz.de
ORCID iD: 0000-0001-6622-5163

PhD student at the School of Foreign Languages, Sun Yat-sen University. She is sponsored by the China Scholarship Council (CSC) and is currently a visiting student in the Department of Linguistics at the University of Konstanz. Her research interests include discourse grammar, Chinese dialects, grammaticalization, and the syntax-prosody interface. She has co-authored six papers in journals such as Functions of Language, Language and Linguistics and SAGE Open.

Guangzhou, P.R. China; Konstanz, Germany

References