Ahn - Ha Yap - 2013 - On - The - Development - of - Korean - SAY - Evidentials - and - Their - Expected - Grammatical - Functions

John Benjamins Publishing Company
This is a contribution from Diachronica 31:3

© 2014. John Benjamins Publishing Company
This electronic file may not be altered in any way.
The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be
used by way of offprints, for their personal use only.
Permission is granted by the publishers to post this file on a closed server which is accessible
only to members (students and faculty) of the author’s/s’ institute. It is not permitted to post
this PDF on the internet, or to share it on sites such as Mendeley, ResearchGate, Academia.edu.
Please see our rights policy on https://benjamins.com/#authors/rightspolicy
For any other use of this material prior written permission should be obtained from the
publishers or through the Copyright Clearance Center (for USA: www.copyright.com).
Please contact [email protected] or consult our website: www.benjamins.com
Articles / Aufsätze
On the development of Korean SAY evidentials

and their extended pragmatic functions*
Mikyung Ahn and Foong Ha Yap

Hankuk University of Foreign Studies / Hong Kong Polytechnic University
This paper examines the development of five hearsay evidential markers in

Korean, namely, tako, tamye, tamyense, tanun and tanta, and traces their ex-
tended pragmatic functions in discourse. We first identify their functions over
time, from Middle Korean to Modern and Contemporary Korean, then quanti-
tatively analyze the usage frequency of these functions, diachronically from the
16th century to the early 20th century using the UNICONC historical corpus,
and synchronically in present-day Korean using the Sejong contemporary writ-
ten and spoken corpus. From a pragmatic perspective, we examine how Korean
speakers use these hearsay evidential markers to convey the interpersonal and
intersubjective stances of interlocutors in natural conversations. Based on the
differential rates of grammaticalization of these markers, and on their usage fre-
quency, we also examine the relationship between evidentiality marking and fi-
niteness; more specifically, we analyze the sequences and mechanisms of change
whereby different types of non-finite evidential structures develop into finite
evidential constructions. Our findings have broader theoretical and crosslinguis-
tic implications for understanding the mechanisms of insubordination, whereby
dependent structures become independent, and whereby lexically transparent
constructions develop into grammaticalized markers of speakers’ stance.
Keywords: hearsay evidential markers, diachronic corpus analysis, epistemic

strength, phonological reduction, verbal elision, main-clause ellipsis,
insubordination, finiteness
* We wish to gratefully acknowledge support from Hankuk University of Foreign Studies

Research Fund (20141160001) awarded to the first author and from the Research Grants Council
of Hong Kong (General Research Fund, #PolyU 5513/10H) awarded to the second author. The
first draft of this paper was presented at the conference on “The Nature of Evidentiality” (TNE),
June 14–16, 2012 at Leiden University, The Netherlands, and this paper is an expanded, revised
version. We wish to thank the participants at this conference for stimulating discussions that
have further shaped our understanding of the nature of evidentiality. We also wish to thank the
editors and anonymous reviewers and the following scholars for their invaluable comments:
Hilary Chappell, Sonja Gipper, Kaoru Horie, Changmin Jeon, Mary Shin Kim, Chungmin Lee,
Jinho Park, Seongha Rhee, Mizuho Tamaji, Tak-Sum Wong and Vivien Ying Yang.
Diachronica 31:3 (2014), 299–336. doi 10.1075/dia.31.3.01ahn

issn – / e-issn – © John Benjamins Publishing Company
300 Mikyung Ahn and Foong Ha Yap
1. Introduction
Speakers often indicate the source of their information when they make a claim;
this linguistic phenomenon is referred to as ‘evidentiality marking’ (Anderson
1986, Chafe & Nichols 1986, Willett 1988, Johanson & Utas 2000, Aikhenvald &
Dixon 2003, Aikhenvald 2004, Aikhenvald & LaPolla 2007, Gipper 2011, inter
alios). Speakers generally have at their disposal a variety of linguistic devices to
convey such evidential information. In English, these devices include lexical and
grammatical constructions, some of which primarily serve as perceptual eviden-
tial markers (e.g., I hear, people say and it is said), while others can further serve
as inferential evidential markers (e.g., it looks like, apparently and evidently) (see
Fox 2001).
Recent trends in evidentiality studies have shown that evidential markers of-
ten do more than identify the source of information; in many languages, these
markers often indicate the degree of accessibility to information as well (Chung
2006, Ekberg & Paradis 2009, Nuyts 2001, Shibatani 1990). In some languages,
evidential markers are also used to modulate the strength of a speaker’s epistemic
claim (see Cornillie 2009, Fox 2001, Hill & Irvine 1993, M.S. Kim 2011, Rudolph
1993). These new, more pragmatically-oriented directions in evidentiality re-
search have called for more discourse-based studies. In this paper, we examine
the development of say-derived evidential markers in Korean to shed light on the
mechanisms whereby a grammatical category develops into a pragmatic one —
more specifically, how an evidential marker also comes to function as a pragmatic
marker expressing a speaker’s stance.
The Korean suffix -te has received considerable attention and has long been
regarded as the evidential marker ‘par excellence’ in this language since it displays
directly perceived information (Kwon 2009, J. Lee 2010, Lim 2010). Against this
backdrop, other evidential markers have been somewhat neglected, in particular
the say-derived evidential markers such as tako, tamye, tamyense, tanun and tanta,
all of which express hearsay evidentiality, albeit with their own distinctive seman-
tic nuances as illustrated in (1) below. Their extended uses as pragmatic markers
have also been barely recognized, with the notable exception of M.S. Kim’s (2006,
2011) conversational analytic study of the evidential marker tamye.1
1. M.S. Kim (2006, 2011) recently showed that participants in discourse select an evidential
marker based on the relationships between the speaker, the addressee and the information be-
ing conveyed. Participants also mutually monitor one another’s epistemic claims to knowledge
moment by moment and shift their choices of evidential markers according to their updated
assessments. In this way, Kim observes, evidential markers can function as an interactive device
for redistributing the participants’ epistemic rights and reorganizing the participation frame-
work of the moment.

All rights reserved
On the development of Korean say evidentials 301
(1) a. Tom-i cip-ul naka-ss-tako?

Tom-nom home-acc leave-pst-evid
“It is said Tom left home, isn’t it true?”
b. Tom-i cip-ul naka-ss-tamye?
“Tom left home, I hear, isn’t it true?”
c. Tom-i cip-ul naka-ss-tamyense?!
“Tom left home, I hear, isn’t it true?!”
d. Tom-i cip-ul naka-ss-tanun
“Tom left home, I hear, (but I don’t know about it personally).”
e. Tom-i cip-ul naka-ss-tanta.
“(I’m afraid to tell you this but) Tom left home, I hear.”
While a growing number of studies treat the development and semantics of tako,
tamye and tamyense as sentential endings (e.g., K.H. Lee 2006, Rhee 2008a, Nam
2010), few studies deal with the evidentiality-marking functions of these parti-
cles.2 Moreover, previous studies have not explicitly identified the -ta(ha)- form
(derived from a combination of declarative (dec) sentence final particle -ta in
the preceding complement clause and the complement-taking utterance verb ha
“say” in the matrix clause, i.e., a dec + “say” construction) as a significant source
of hearsay evidentials in Korean. In light of this gap, the present study investigates
the development of the evidential markers mentioned above and their extended
uses as structurally-constrained and context-induced pragmatic markers. This
study, then, provides a diachronic perspective on the frequently intimate relation-
ship between grammaticalization and pragmaticization.
More specifically, we focus on Korean hearsay evidential markers that have
been derived via phonological elision of the say verb ha (e.g., -ta ha-ko > tako; -ta
ha-mye > tamye; -ta ha-myense > tamyense; -ta ha-nun > tanun; -ta ha-n-ta > tanta)
and also via syllable fusion of the remaining sentence final particle -ta from the
2. More specifically, K.H. Lee (2006) has recently examined how quotatives such as tamyense,
tamyen, taney, canikka and cako, among others, develop into sentence final particles, claiming
that these sentential endings have undergone /h/ deletion, followed by segment reduction and
syllable fusion. In a similar vein, Rhee (2008a) and Nam (2010) have investigated the diachronic
developments of the sentence final particles (SFPs) tako, nyako, lako and cako. In particular,
Rhee (2008a) noted that the emphatic meaning associated with these sentential endings is large-
ly derived from their original function as reportatives. In synchronic studies from a pragmatic
perspective, Jeon (1996) and Park (2008) found that tako-ending repercussive questions express
confirmation-seeking and counterexpectation-marking functions.

All rights reserved
preceding complement clause and the various postverbal suffixes (e.g., connec-
tives -ko, -mye and -myense; attributive -nun; present-tense conclusive -n-ta). As
these -ta(ha)- constructions with their phonetically elided say verb ha lack an
unambiguous interpretation on their own, their postverbal suffixes come to be
associated with pragmatic values induced from either conventionalized or specific
discourse-cum-situational contexts. This triggers the evolution of a new gram-
matical system (Eckardt 2006, Hopper & Traugott 2003), in this case, a hearsay
evidential paradigm.3
The development of these evidential markers is of theoretical interest because
they provide evidence of a high incidence of verbal elision among complement-
taking verbs in Korean that facilitate the emergence of a wide range of pragmatic
markers. In the present study we focus in particular on the elision of the utterance
verb ha “say” and its contribution to insubordination phenomena that give rise to
sentence final evidential particles with pragmatic functions.4
This line of study can shed light on similar phenomena in other verb-final
languages (e.g., Japanese, Manchu-Tungusic, Mongolian, Turkic, Tibeto-Burman)
and pave the way for typological comparisons with say-derived evidential mark-
ers in verb-medial (e.g., Sinitic) and verb-initial (e.g., Austronesian) languages.
This would allow us to identify not only robust morphosyntactic restructuring
strategies that enhance similarities across languages but also culture-specific pref-
erences that give rise to pragmatic variation.
Research on quotative-derived sentential endings has tended to focus either
on diachronic change or on pragmatic functions. We combine both traditions
to better identify the grammaticalization pathways and the semantic factors and
morphosyntactic mechanisms that contribute to the reanalysis of reportative verbs
as hearsay evidential markers, often with pragmatic nuances.
The paper is organized as follows. §2 reviews literature on evidential markers
derived from say constructions, particularly in Asian languages. In §3, we trace
the diachronic development of say-derived evidential markers in Korean, specifi-
cally the grammaticalization of tako (§3.1), tamye (§3.2), tamyense (§3.3), tanun
3. Some previous studies have included the notion of ‘emphasis’ among the pragmatic values
(Hopper & Traugott 2003, Eckardt 2006). Others include the notion of ‘activation’ (Schwenter
2006, Mosegaard Hansen & Visconti 2007). Our discussion is somewhat related to the notion
of activation, which comprises a reformulation of the notion of presupposition as information
that is accessible to the hearer (Dryer 1996). The accessibility of a proposition comes from being
explicitly used in the antecedent context, or being inferable from accommodating constructions
or contextual relations (Larrivée 2011).
4. ‘Insubordination’ is the process whereby a syntactically dependent structure (e.g., an embed-

ded complement clause) is reinterpreted as a syntactically independent one (Evans 2007).

All rights reserved
(§3.4) and tanta (§3.5), and we show how these five hearsay evidential markers are
derived via say verb elision and the subsequent insubordination of the comple-
ment clause. We also trace how these say evidential markers develop extended
pragmatic uses. In §4, we investigate the effects of usage frequency on the gram-
maticalization of these say-derived hearsay evidential markers. §5 concludes with
a summary of findings on the development of say-derived evidential construc-
tions in Korean and their extended pragmatic uses, with implications for say evi-
dential markers in other languages as well.
2. Previous studies on evidential say constructions
Many languages recruit their say verbs to signal hearsay evidential information
(e.g., Aikhenvald 2004, Heine & Kuteva 2002). For Sinitic, for example, Chang
(1998) has traced the extension of Taiwanese kong “say” into quotative, evidential
and counterexpectation markers. The term ‘quotative’ is typically used when the
author of a quoted report is identifiable from the discourse or the context, while
the term ‘evidential’ is used instead when the source of information is someone
other than the speaker and hearer and furthermore is left unspecified, for exam-
ple in constructions like People say or It is said (see also Faller 2002, Aikhenvald
2004: 177). A similar extension has also been observed for Mandarin shuo “say”
(e.g., Wang et al. 2003, Su 2004). Similar developments have likewise been report-
ed for Cantonese waa “say” (Chui 1994, Matthews 1998, Leung 2006, 2009, Yeung
2006).
In Amdo Tibetan, Sun (1993) reports that the evidential marker se is derived
from Old Tibetan zer “say” via devoicing of z and loss of the -r, both due to pho-
nological weakening.5 In Austronesian languages, Hsieh (2012: 489) found the
Kavalan verb zin “say” has gradually lost its lexical meaning and has developed
into a pragmatic marker that indexes “the speakers’ commitment to the source of
information,” conveys the speaker’s “attitude and belief ” and captures the speak-
er’s “attention to the ‘self ’ of the addressee.”
Some scholars have also suggested that Japanese evidential marker tte has its
origins in a say construction. Suzuki (2007) posits that this hearsay evidential
marker evolved from the converbal to itte “say” construction, with tte further de-
5. Amdo is a major Tibetan dialect spoken in many closely related varieties in northwestern
Sichuan, southern Gansu and eastern Qinghai, China (Sun 1993).

All rights reserved
veloping a wide range of pragmatic uses (also Miura 1974, S. Suzuki 1999, Tamaji
& Yap ms.).6
In the case of Korean, Sohn (2011) has traced the development of the lexical
verb ha “say” into the quotative marker hako (derived from hA-ko “say-conn”),
with kho and ko as phonological variants, the latter further developing into com-
plementizer tako (see §3.1).7 Yap & Ahn (2011) recently further traced the devel-
opment of tako into a hearsay evidential and counterexpectation marker. In the
next section, we extend that analysis to include the development of four additional
say-derived evidential markers in Korean, namely, tamye, tamyense, tanun and
tanta. This will enable us to better understand how hearsay evidential markers
emerge in Korean and subsequently develop pragmatic nuances, with possible im-
plications for other languages as well.
3. On the grammaticalization of evidential say constructions and their

extended pragmatic functions in Korean
In this section, we trace the development of the evidential markers tako, tamye,
tamyense, tanun and tanta, with special attention to the elision of the say mor-
pheme ha in the rise of these markers. We also examine how these five markers
differ from each other morphosyntactically.
3.1 On the grammaticalization of tako
We begin by examining how the Korean complementizer tako has developed hear-
say evidential and counterexpectation readings from say constructions. The pro-
cess first involves the grammaticalization of ha-ko, a combination of the lexical
verb ha “say” and connective -ko “and,” as shown in (2a). This combination then
developed into quotative hako ~ kho as in (2b) and (2c), respectively, as noted in
Sohn (2011). Quotative uses of hako were attested as early as the 15th century
(with kho as an occasional phonological variant), and through elision of the say
6. An alternative account posits instead that tte is derived from tote, via phonological elision of
the say verb from an erstwhile to V-te frame and concomitant syllable fusion of the remaining
complementizer to and converbal linking particle -te (Yuzawa 1954, Konoshima 1973, Saegusa
1997). Tamaji & Yap (2014) provide diachronic evidence that shows that tte is phonologically
derived from to itte via the pathway to ipite > to ifite > to ihite > to itte > totte > tte, with tote devel-
oping similar functions much earlier in Old and Middle Japanese.
7. The vowel A in hA is called ‘lower a,’ and is written as a dot, positioned beneath the conso-
nant. A was used until the 19th century and was replaced by a.

All rights reserved
verb ha, quotative hako was phonologically reduced to -ko, which in combination
with the declarative sentence final particle -ta from the preceding complement
clause, was reanalyzed as a reportative complementizer tako in the 18th century,
as in (2d) (see also Rhee 2011).
wang-i
(2) a. nwiuchy-e pulli-si-ntay [timaniho-ita] ha-ko
king-nom regret-and call-sh-and [apologize-dec] say-and
ani o-nila
not come-dec
“The king regretted and called them (his sons) again, but they didn’t
come, saying: ‘We apologize.’”(Welinsekpo 14598; cited in Sohn 2011: 129)
b. taycwung-tAlh-i ta wuz-umye wang-to wuz-umye
people-pl-nom all laugh-and king-also laugh-and
nilo-tAy [i aki kongtek-ul teunos-ta] hAko
speak-and [this baby charity-acc practice-dec] qt
nilo-tAy nay sto samsip man nyang kum-ulo
say-and 1sg also thirty ten.thousand (unit) gold-with
motAn cwung-nim-nay-lAl kongyanghA-mye
all monk-hon-pl-acc serve-and
“Everyone laughed; the king also laughed, and said: ‘This baby does
(Buddhist) charity’ and he said: I will also serve all monks with three
hundred thousand nyang of gold.”
(Sekposangcel 1447; cited in Sohn 2011: 132)
c. ku pskuy pAlAm-i pwul-e pen-i mwi-enAl
that time wind-nom blow-and banner-nom move-and
hAncyung-un nilo-tAy [pALAm-i mwi-nA-ta]-kho
onemonk-top say-and [wind-nom move-ind-dec]-qt
hAn cyung-un nilo-tAy [pen-i mwui-nA-ta] hAya
one monk-top say-and [banner-nom move-ind-dec] say
uylon-Al mati-ani-khenul
discussion-acc stop-neg-and
“At that time a wind blew and so a banner was moving; one monk said:
‘Wind is moving’; another monk said: ‘Banner is moving.’ They didn’t
stop arguing …”
(Yuckco 1496: 57; cited in Sohn 2011: 133)
d. [sAsAloi syo-lAl cwuk-yes]-tako ha-l cyay is-kenal
[for.personal.gain cow-acc kill-pst]-comp say-adn case exist-sfp
“There was a case where it was said a thief killed a cow for his own
personal gain.” (Congteksinphyen 1758)
8. This is from Welinsekpo 1459.

All rights reserved
Earlier uses of quotative hako ~ kho during the 15th century, as in (2b) and (2c)
above, could be introduced by an utterance verb such as nilo-tAy “say-and” in the
preceding clause. Later uses in the 18th century shifted in favor of utterance verbs
such as hal “say-adn” in clause-final position, more specifically, after the reportative
complementizer -tako, as in (2d). Whereas the speaker in a quotative construction
is identifiable, for example wang “the king” in (2b) or hAn cyung “a monk” in (2c),
the speaker is left unspecified in (2d). At this stage, the reportative complementizer
tako in vp-tako hal constructions such as (2d) could yield a hearsay evidential in-
terpretation, but the construction was still lexically transparent and functioned as
an evidential strategy, and tako had not yet evolved into an evidential marker.9
By the 19th century, the complementizer tako could be followed by a verb of
cognition such as pota “see, consider” as well as say verbs, as seen in (3a), and
it could further be used to set up topics to introduce the speaker’s subsequent
comments or evaluations, crucially in contexts without an (explicit) reportative
or quotative function, as in (3b), where tako serves instead a converbal function
with a general “and” as well as a causal connective reading (the latter equivalent
to English adverbial subordinator because).10 This extension into non-reportative
contexts would be a natural one given that the morpheme ha “say” need not be
overtly expressed with quotative ko (<hako ~ kho) (also Sohn 2012).
lankan
(3) a. tol-i wumcak-ye mwulnena-s tako
balustrade stone-nom move-as collapse-pst comp
po-nun skAtAlk-ey …
see-adn reason-cause
“Because it is seen (considered) that as a stone in the balustrade moved,
it collapsed …” (Toklipsinmwun 1896)
b. [hAnphen silsyu hA-yes]-tako naynom-uy
[once mistake make-pst]-comp 2sg-gen
mal-pelus-I kul-ha-l-syu-ka is-ta-n malka
word-way-nom such-do-adn-way-nom exist-dec-adn sfp
“Just because I made a mistake once, how could you say such a thing!”
(Sincayhyopansolisasel2 496 19th century; cited in Nam 2010: 121)
The extension from quotative to complementizer is also widely attested in the
languages of the world (Heine & Kuteva 2002: 265–267, Güldemann & Roncador
9. We employ the term ‘evidential’ in its narrow sense to denote grammaticalized elements that
indicate ‘source of information’ (Michael 2008: 135–140) and ‘evidential strategy’ to denote their
relatively ungrammaticalized counterparts, largely following Aikhenvald (2004).
10. It has been noted (e.g., M. Kim 2011) that in the case of versatile connectives, the discourse
and situational context help determine the intended meaning conveyed by the connectives (e.g.,
the textual connective -mye “when” > subjective textual-and-pragmatic connective “because”).

All rights reserved
2002, Güldemann 2008, Chappell 2008). The distinction between them lies in the
nature of the complement-taking matrix verb: quotatives are often followed by
utterance verbs, while complementizers are followed instead by non-utterance
verbs, though the distinction is sometimes blurred because cognitive verbs such
as think and wonder have utterance-like uses as well. However, as pointed out in
Heine et al. (1991), these unclear, ambiguous cases form the ‘bridge contexts’ for
semantic extensions that give rise to new grammatical functions.
For our present focus, it is important that in contemporary Korean, quotative
tako has further developed into a hearsay evidential marker with the meaning
“People say” or “It is said” as in (4a), where a worried husband shares something
he had heard with his wife, and then makes a sarcastic comment about this alarm-
ing development. Often, the hearsay evidential marker tako is also used to signal
a confirmation-seeking function in interrogative contexts with meanings such
as “Did you say X?,” as in (4b), and also to signal a counterexpectation-marking
function in mirative contexts (DeLancey 2001) that depicts the speaker’s surprise,
shock, disbelief, dismay, etc., as in (4c). The confirmation-seeking and counter-
expectation-marking functions of the hearsay evidential marker tako point to the
lingering influence of an implicit say morpheme such as ha, with the counterex-
pectation interpretation emerging in contexts where the speaker responds with
incredulity at what has just been said, heard or perceived.
[Ilpon-un tongpyeng
(4) a. cyunpi-ka ta
[Japan-top collecting.army preparation-nom all
toy-yass]-tako. I-kes cham camiisketoy-yas-kona
become-pst]-evid this-nmlz excl be.interesting-pst-sfp
“People say /It is said that Japan finished the preparation for recruiting
the army. This (situation) has become very interesting.”
(Twukyenseng 1912)
b. A: Yenghuy-ka wuli-pan-eyse ceyil yeypp-e.
Yenghuy-nom 1pl-class-in most be.pretty-ie
“Yenghuy is the most beautiful in our class.”
B: [Yenghuy-ka wuli-pan-eyse ceyil yeyppu]-tako?
[Yenghuy-nom 1pl-class-in most be.pretty]-cs
“Did you say (that)
Yenghuy is the most beautiful in our class?” (Jeon 1996: 901)
c. [yeca-ka wucwu pihayngsa-lul ha-keyss]-tako!?
[woman-nom astronaut-acc do-be.willing]-ce
“How dare a woman become an astronaut!?” (Yu 2002: 116)
[lit. “It’s said a woman is willing to become an astronaut!?”]
(our reconstruction)

All rights reserved
Factors such as intonation and discourse context are important for the pragmat-
ic interpretation of tako. A low tone at the end of declaratives yields a hearsay
evidential interpretation, while a rising tone at the end of interrogatives yields a
confirmation-seeking interpretation, and a high tone plus high pitch at the end
of mirative expressions yields a counterexpectation interpretation. In addition,
depending on discourse context, tako constructions such as (4b) could be inter-
preted with a counterexpectation-marking function instead of a confirmation-
seeking one, while a reverse interpretation could emerge in tako constructions in
(4c). For example, in (4b), if the speaker thinks that Yenghuy is the ugliest girl in
her class, the speaker will experience surprise, shock, disbelief, dismay, etc., and
as such the sentence final prosody on tako will yield a counterexpectation reading.
On the other hand, if interrogative instead of exclamative prosody is used at the
end of (4c), the tako construction can be understood as a confirmation-seeking
construction (“Did you say a woman is willing to become an astronaut?”) rather
than a counterexpectation-marking one (“How dare a woman become an astro-
naut?”). In sum, the illocutionary force of the utterance (whether declarative, in-
terrogative or exclamatory), is in large part realized through the speaker’s sentence
final prosody, with clause-final tako as its convenient host. What remains to be
explained is how a clause-chaining and non-conclusive tako construction comes
to be reanalyzed as a stand-alone conclusive (or finite) construction, where tako is
reinterpreted as a sentence final evidential or pragmatic marker.11
Tako has also developed into an emphatic marker without evidential nuance,
as in (5). The emphasis is derived from the speaker’s repeated presentation of a
proposition that has been previously uttered.
11. In casual and familiar speech, especially among the younger generation, tako is often pro-
nounced as takwu (Jeon 1996: 899). Takwu signals not only that the information uttered is
something that the speaker (often) hears people say, but also that the speaker is confident of the
reliability of the statement, as illustrated in the dialogue below:
A: ttak ikey nwuwu-si-n-ke-ya kuntey, salam-tul-i kulen mal isscanha,

just like.this lie-hon-adn-nmlz-sfp by.the.way man-pl-nom such word dm
tolaka-si-l-ttay-ka toy-myen cengsin-i wa-ss-ta ka-ss-ta
die-hon-adn-time-nom become-when mind-nom come-pst-dec go-pst-dec
ha-n-takwu
do-impfv-evid
“He is lying just like this. By the way, people usually say like this, you know.
I heard (and know for a fact) that when a man is about to pass away, the mind strays.”
B: Ung
“Yes.” (Sejong Contemporary Spoken Corpus)

All rights reserved
(5) A:oppa ce yeyppe-ci-n ke

older.brother 1sg(honorific) be.pretty-become-adn nmlz
kath-ci anh-a-yo?
look-nmlz neg-ie-pol
“Older brother, don’t you think that I look prettier than before?”
B: twul ta cokum te elunsulewe-ci-n ke
both all a.bit more be.mature-become-adn nmlz
kath-ay
look-ie
“Both of you look a bit more mature (than before).”
(referring to both A and her friend who is also present but is quiet
during the conversation)
A: e?
“What?”
B: twul ta elunsulewe-ci-n ke kath-tako!
both all be.mature-become-adn nmlz look-emph
“Both of you look more mature!” [lit. “(What I said is) that both of you look
more mature!”] (Sejong Contemporary Spoken Corpus)
Tako can also be used to emphasize a proposition even when speakers present it for
the first time. In this context, the emphasis is derived from the speaker’s presenta-
tion of a proposition as if it had been previously uttered (see Rhee 2008a). Using
this strategy, the speaker can frame the conversation to serve his own purpose. In
(6), for example, the speaker is engaging in some form of self-compliment (prais-
ing his own mother), normally a dispreferred move. By adding an emphatic tako
in sentence-final position, the speaker implicitly sets up a mock ‘prior-context’ in
which someone has just commented favorably about his mother and he is merely
affirming the truth of that compliment. In this way, tako can be used to attenuate a
potential face-threat to himself as he gets away with self-praise.
(6) wuli12 emma-ka elmana yeyppu-tako!
we mother-nom how be.pretty-emph
“My mom is really pretty!” [lit. “(It truly is the case) that my mom is really
pretty.”] (S.T. Kim 2005: 117)
We have thus far seen how the quotative marker -ko combined with the de-
clarative sentence final particle -ta from the preceding complement clause to
form a complementizer tako that has further developed into a hearsay eviden-
tial marker, often with a confirmation-seeking function in interrogative contexts,
12. In Korean, the first-person plural (e.g., wuli “we”) is used in place of the first-person singular
(e.g., nay “I”) to express kinship.

All rights reserved
or a counterexpectation-marking function in exclamatory contexts, as well as an

emphatic-marking function in contexts involving repetition of a prior utterance.
Of particular interest is the role of ha “say” in the development of these various
evidential and extended pragmatic functions of tako. In the early stages of the de-
velopment of quotative marker hako ~ kho ~ ko, the morpheme ha “say” was elided
in the process of phonological reduction that gave rise to the complementizer tako.
More specifically, we see the following reductions: elision of /ha/, followed by sylla-
ble fusion of the declarative sentence-final particle -ta and phonologically reduced
quotative -ko, such that the quotative construction involving -ta hako is reana-
lyzed to yield complementizer tako. These reduction processes notwithstanding,
the inherent semantics of the morpheme ha “say” continue to exert an influence
on the more grammaticalized uses of tako as a hearsay evidential marker, includ-
ing its pragmatic uses in confirmation-seeking, counterexpectation-marking and
‘emphatic repetition’ contexts as seen in (4b), (4c) and (5) above.
However, whereas overt realization of the ha “say” morpheme is possible for
quotative uses of tako (i.e., vp-ta hako + utterance verb), re-insertion of the ha
morpheme is no longer grammatical in the case of evidential and pragmatic uses
of tako, largely because they are now semantically bleached and also function as
phonologically-reduced sentence-final mood particles.13
These developments are summarized in Figure 1. The quotative uses of tako
emerged earlier than the evidential uses, a difference of about 200 years (18th vs.
20th century), with extended pragmatic functions emerging in the late 20th cen-
tury. Essentially, it takes a further step involving insubordination of the (quoted)
complement clause for tako to be reanalyzed as a grammaticalized hearsay eviden-
tial marker.
This extension of functions from the textual domain (quotative and comple-
mentizer) to the pragmatic domain (hearsay evidential marker, often with confir-
mation-seeking, counterexpectation-marking and emphatic-marking functions)
13. a. ‘Na-nun pap-ul mek-ess-ta’ hako malha-yss-ta

1sg-top meal-acc eat-pst-dec qt say-pst-dec
“ ‘I had a meal,’ I said.”
b. Na-nun pap-ul mek-ess tako malha-yss-ta
1sg-top meal-acc eat-pst qt say-pst-dec
“I said I had a meal.”
c. *ne pap mek-ess-tahako?
2sg meal eat-pst-evd
“You had a meal, I hear, isn’t it true?”
d. ne pap mek-ess-tako?
2sg meal eat-pst-evid
“You had a meal, I hear, isn’t it true?”

All rights reserved
appears to be a fairly robust pathway in Korean. Below we identify some parallels

in the development of quotative tamye and tamyense (§§3.2–3.3).
Middle Korean Modern Korean Contemporary Korean

Lexical ha-ko “say-and”
Quotative hako ~ kho (15th c.)
Quotative ko (17th c.)
Complementizer tako (18th c.)
Evidential tako (early 20th c.)
Confirmation-seeking tako (late 20th c.)
Counterexpectation tako (late 20th c.)
Emphatic tako (late 20th c.)
Figure 1. The grammaticalization of tako in Middle, Modern and Contemporary Korean14
3.2 On the grammaticalization of tamye
As seen in (7a), -mye was a productive connective in Middle Korean, much like
-ko. Connective -mye often combined with the lexical verb ha “say” to form quota-
tive hamye, as in (7b), a phenomenon attested from the late 16th century.
(7) a. kecusmal hA-mye kkwumyun mal hA-mye15
lie say-conn made-up word say-conn
nAm kwucicu-mye…
others scold-conn
“(Ten evil acts:) tell a lie and tell a made-up story and scold others …”
(Welinsekpo 1459)
16
b. Taykap -ay kAlo-tAy [i hanals pAlkun myeng-ul
pn(book)-in say-and [this sky bright command-acc
kohA-ta]-hAmye
think.over-dec]-qt
“Taykap said ‘I will think over sky’s good command.’ ”
(Tayhakenhay 1590)
14. Scholars divide Korean into the following stages: Old Korean (pre-14th century), Middle
Korean (14th–16th century), Modern Korean (17th–19th century), Contemporary Korean
(20th century to the present).
15. Kkwumyun mal hA-mye in (7a) is a verb + object construction (lit. “say” + “made-up word
or lie”) and so this is lexical ha “say” plus connective -mye “and.” (7b) has a say verb kAlo-tAy
preceding the quote, and so hAmye is used as reinforcement and reanalyzed as a quotative.
16. Taykap is an edited volume of Seokyung, one of five scriptures of Chinese Confucianism.

All rights reserved
As seen in (8a) and (8b) from the 18th century, quotative hamye began to in-
creasingly merge with the declarative sentence-final particle -ta from the preced-
ing complement clause, and after elision of the say morpheme ha, this yielded a
new quotative variant, tamye. In other words, we see the following reanalysis: V-ta
hamye > V-tamye. This process is analogous to the development noted earlier for
tako (i.e., V-ta hako > V-tako).
[pwuin-uy toli-nun pwumo-ka epsu-myen
(8) a. pamey
[wife-gen duty-top parents-nom neg.exist-when at night
tang-ey nayli-ci anh-nun]-tamye toli-lul
house-loc get.out-nmlz neg.do-impfv]-qt duty-acc
cikhita-a. cwuk-ess-ta
keep-(nf) die-pst-dec
“Saying (that) the duty of a wife is not to get out of the house at
night when there are no parents around, she died doing her duty.”
(Pingpingdyen 18th century)
b. [amoto chinhA-n salAm-to ep]-tamye
[nobody be.close-adn person-also neg.exist]-qt
sikol-sa(l)-nAn ahAy-ka mwusun polil-i] iss-e
country-live-adn child-nom what thing-nom]
iss-e olnaw(a)-as-nAnya?
exist-seq come-pst-q
“While saying there is no close friend, what has brought you, the child
living in the country, here?” (Komokwha 1908)
Similar to tako, contemporary Korean tamye has also developed a hearsay evi-
dential function, albeit with a subtle difference, seen in (9). According to M.S.
Kim (2011: 436), the evidential reading associated with tamye highlights not only
that “the speaker has acquired the information through hearsay” but also that “the
speaker assumes the information s/he reports is already known to the addressee.”17
(9) A: hanpen manna-pwa(<po-a)!
once meet-see-ie
chinceng-ulo tolaw(<o)-ass-tamye.
home.of.parents.of.married.woman-to return-pst-evid
“(Just) meet her once! I heard that she returned to her parents’ home.”
B: silh-e!
hate-ie
“I hate (meeting her)!” (Sejong Contemporary Spoken Corpus)
17. M.S. Kim (2011: 436) further notes that tamye “also serves as an interactive resource which
functions to manage participants’ epistemic rights.”

All rights reserved
Another function that tamye has in common with tako is the confirmation-
seeking function, as seen in (10). This confirmation-seeking function emerges
as an extended function of evidential tamye in interrogative contexts (M.S. Kim
2011: 439).
(10) A: wuli oppa yeki wassess-tamye?
1pl older.brother here had.been-cs
“I heard my older brother had been here, is it true?”
B: ung
“Yes.” (Sejong Contemporary Spoken Corpus)
Unlike tako, often deployed for more emphatic effect, tamye is typically used to
downgrade (rather than upgrade or assert) the strength of the speaker’s epistemic
claim to knowledge, and at the same time acknowledge the co-participant’s epis-
temic right. This dual property of tamye comes in handy since, in Korean conver-
sation, speakers are sensitive to co-participants’ epistemic rights to information.
That is, speakers often select an evidential marker based on an assessment of other
participants’ epistemic rights, as seen in (11), reproduced from M.S. Kim (2011).
In this conversation involving two female friends, Speaker E brings up the topic
of Speaker J’s plan to purchase a house. Because Speaker J has epistemic primacy
over the information regarding herself, Speaker E uses tamye to downgrade her
own speaker’s epistemic right and simultaneously acknowledges co-participant J’s
epistemic right (see M.S. Kim 2011: 439).
(11) E: uh ani cip sa-n-tamye?
uh by.the.way house buy-impfv-evid
“Uh, by the way, you’re buying a house, [I hear, is it true?]”
J: e ah kulayse kwangju pwumo-nim-i
yes ah so (name) parents-ht-nom
ton-ul ponay cwu-si-nuntey:,
money-acc send:for-sh-nuntey
“Yes, ah, Kwangju’s parents are sending money for (us)”
(M.S. Kim 2011: 439)
In mirative contexts, as in (12) below, tamye often further serves a counterexpec-

tation-marking function. That is, tamye also signals the speaker’s surprise, shock,
disbelief, dismay, etc. upon receiving new or unexpected information that violates
the speaker’s prior expectation; it also often indicates that the new information
has not yet been integrated or assimilated to the speaker’s previous knowledge. We
thus see a strong association between the use of tamye and the expression of the
speaker’s intersubjective (or interpersonal) stance (see Nuyts 2001).

All rights reserved
(12) [ku-ka manhi aphu]-tamye!?

[sg-nom very be.sick]-ce
“I heard he is very sick, is it true!?”
(=“I can’t believe he is very sick!”) (K.H. Lee 2006: 248)
The development of connective -mye into quotative tamye, and subsequently into
hearsay evidential marker tamye, often with extended confirmation-seeking and
counterexpectation functions, is summarized in Figure 2 below.

Connective -mye (15th c.)
Lexical ha-mye “say-conn” (15th c.)
Quotative hamye (16th c.)
Quotative tamye (18th c.)
Evidential tamye (20th c.)
Confirmation-seeking tamye (20th c.)
Counterexpectation tamye (20th c.)
Figure 2. The grammaticalization of tamye in Middle, Modern and Contemporary Korean.
3.3 On the grammaticalization of tamyense
Connective -myense emerged in Modern Korean around the beginning of the 18th
century, and came to be used often as a more emphatic counterpart of connective
-mye (M. Kim 2011).18 As seen in (13a), -myense could be used as a temporal
connective equivalent to English and when. When it combines with utterance verbs
such as ha “say,” it produces a quotative converbal construction with the structure
[Complement clause +ha-myense], meaning “When/While saying X, …,” as in (13b).
(13) a. syeysyok-i hwangnyen-kwa kamcho
brother-in-law-nom oriental.medicine-and oriental.medicine
talhi-n mul-lo ahAy kAsna-myensye cyekcyek
boil-adn water-with child give.birth.to-when much
mek-y-e pAysok-uy teleon kes mek-un
give-let-seq stomach-gen dirty nmlz eat-adn
stong-ul nu-i-nAni
dung-acc defecate-let-sfp
18. Seongha Rhee (p.c., October 2011) pointed out the following developmental sequence: ta-
mye > tamyese (more emphatic) > tamyense (emphasis phonologically reinforced with epenthetic
/n/). A similar discussion of the more emphatic nature of tamyense is found in Ahn (2006: 189)
and M. Kim (2011: 97–116). Basically, epenthesis with /n/ blocks lenition, which explains the
more emphatic force of tamyense compared to tamye.

All rights reserved
“When I gave birth to a child, my brother-in-law let me drink a lot of

water boiled with oriental medicine, and this helped cleanse my stomach
[lit. let me defecate dung]” (Dwuchangkyeng 1711)
b. [ilen myengkwan-ul ileskhey ha-koya phyenanha-lya]
[such good.officer-acc like.this do-conn be.comfortable-q]
ha-myense tulkes-ey tamakaci-ko nam-uy cip-ulo
say-conn stretcher-on put-and other-gen home-loc
oytoha-ye
carry-conn
“While saying, ‘Can’t I be comfortable treating such a good officer
like this?’, he placed (the officer) on the stretcher and carried him to
someone’s home.” (Toklipsinmwun 1896)
Converbal quotative hamyense, similar to hamye discussed in §3.2, also came to

increasingly merge with the declarative particle -ta in the preceding complement
clause to produce a phonologically reduced quotative variant, namely, tamyense.
This process involves verbal elision, followed by phonological fusion: V-ta ha-
myense > V-tamyense. As seen in (14a) below, VP-ta hamyense serves a function
as converbal quotative, with the meaning “While saying/claiming (that) X, ….”
Crucially, the phonologically reduced converbal quotative tamyense in (14b) ex-
presses the same meaning as well.
[kwuksA-to thayphyenghA-ko minsim-to
(14) a.
[national affair-also be.peaceful-conn people-also
anlakhA-ta] hA-myensye
be.comfortable-dec] say-conn
“While claiming that the country is peaceful and the people too are
comfortable…” (Toklipsinmwun 1896)
b. [inmin-ul pohohA-ya-cwu-n]-tamyense
[people-acc protect-nf-ben-pres]-qt
ile-n kes-ul pAlkhy-e-cwuci-an-nAn kes-un
like.this-adn thing-acc clarify-nf-ben-nmlz-neg-adn thing-top
“While claiming that they (as civil servants) are protecting the people,
that they do not clarify this kind of matter (taxes imposed on Koreans
by Chinese) is (neglecting their job)” (Toklipsinmwun 1896)
Similar to tamye, in Contemporary Korean, quotative tamyense has also devel-

oped into a hearsay evidential marker, as seen in (15a), sometimes extended to a
confirmation-seeking function in interrogative contexts as in (15b), and a coun-
terexpectation-marking function in mirative contexts as in (15c).

All rights reserved
(15) a. phikonha-ci anh-a-yo?

be.tired-nmlz neg-ie-pol
ecey-to cam han-swum mos-ca-ss-tamyense.
yesterday-also sleep one-breath neg-sleep-pst-evid
“Aren’t you tired? I heard you didn’t sleep at all yesterday as well.”
(Sejong Contemporary Spoken Corpus)
b. mac-a-yo. mal-i kulekhey manh-tamyense-yo?
correct-ie-pol word-nom so much-cs-pol
“(That’s) correct. (They) talk (about others) so much like that, I hear,
isn’t it true?” (M.S. Kim 2011: 451)
c. A: yenghwa-po-le-ka
movie-see-conv-go
“I’m going to the movies.”
B: mwe? [ne aphu]-tamyense?!
what? [2sg be.sick]-ce
“What? Didn’t you say you are sick?!” (Rhee 2011)
Again like tamye, the hearsay evidential marker tamyense can also be used to
downgrade the speaker’s epistemic right while upgrading the addressee’s epis-
temic right. In terms of illocutionary force, however, tamyense is more emphatic
than tamye (though not as emphatic as tako). In (16), two female students are
engaged in a conversation about dates and relationships. Speaker A tells Speaker B
that she has again met a man she was not supposed to meet. Speaker B then uses
tamyense to rebuke Speaker A for meeting the man again. Speaker B’s use of tamy-
ense has the effect of downgrading her own speaker’s epistemic right to knowledge
about the rendezvous and about what is right and best to do, because Speaker B
recognizes that Speaker A has epistemic primacy over the information concern-
ing her own (i.e., Speaker A’s) situation. At the same time, tamyense can also send
a signal that Speaker B disapproves of the rendezvous, and implicitly is blaming
Speaker B for the way she is handling the situation.
(16) A: ku namca-lul tasinun an-manna-l-ke-ya
the man-acc again neg-meet-adn-nmlz-sfp
“I won’t meet him again.”
B: way kulay?
why so
“What happened?”
A: ecey ku namca-lul tto manna-ss-e
yesterday the man-acc also meet-pst-ie
“I met him yesterday, again.”

All rights reserved
→ B: ku namca-lul tasinun an-manna-n tamyense?

the man-acc again neg-meet-impfv evid?
“Didn’t you say that you will never meet him again?
(Why did you meet him?)” (Sejong Contemporary Spoken Corpus)
Figure 3 below summarizes the development of tamyense as a hearsay evidential

marker, often used in confirmation-seeking and counterexpectation marking con-
texts. Note that although the lexical and quotative uses of ha-myense and tamyense
were attested a century or two later than their precursors ha-mye and tamye, the
extended uses of these say-derived constructions as evidential markers with prag-
matic nuances all occurred around the same time in the 20th century.

(not attested)
Connective -myense (18th c.)
Lexical ha-myense “say-conn” (19th century)
Quotative hamyense (late 19th c.)
Quotative tamyense (late 19th c.)
Evidential tamyense (20th c.)
Confirmation-seeking tamyense (20th c.)
Counterexpectation tamyense (20th c.)
Figure 3. The grammaticalization of tamyense in Middle, Modern and Contemporary
Korean.
3.4 On the grammaticalization of tanun
Yet another quotative that also underwent phonological reduction on the way to
grammaticalizing into a hearsay evidential marker is tanun. More specifically, we
again see elision of the say morpheme ha, followed by syllable fusion: V-ta ha-
nun > V-tanun. One important difference, however, is that V-ta ha-nun does not
end with a connective such as -ko, -mye or -myense but instead with an erstwhile
adnominal -nun reinterpreted as a present tense marker in attributive form (Rhee
2008b, Shi & Kim 2011).19 For this reason, unlike tako, tamye and tamyense, the
phonologically reduced attributive tanun is not used to express either a confirma-
tion-seeking or a counterexpectation-marking function. Instead, tanun develops a
19. Similar reinterpretations of a nominalizer or adnominal as a tense marker are also attested
in other languages (see Grunow-Hårsta 2011 on Magar, a Tibeto-Burman language). In Korean,
this (ad)nominalizer-to-tense marker reanalysis has given rise to a paradigm of attributive tense
markers that include not only -nun (present tense) but also -ul (future tense) and -n (anterior or
perfect marker) (e.g., Rhee 2008b, Shi & Kim 2011, Shin 2005, Yap & Matthews 2008).

All rights reserved
different type of pragmatic function, signaling the speaker’s detachment from his
own utterance. Below we examine how this different pragmatic function extends
from the quotative and evidential uses of tanun.
First attested from the 18th century, the say verb ha sometimes followed a
clause ending with dec -ta and then further combined with adnominal -nun to
form attributive quotative vp-ta ha-nun (VP-dec “say”-adn).20 As an attributive
construction, its function is to introduce a complement clause that modifies a head
noun, such as kes “thing” in (17) below. In other words, we obtain a [complement
clause]-ta ha-nun + head noun construction, which in effect is a type of relative
clause (hence adnominal) construction.
For example, in (17), [senchin-ina tongsAyng-I ta kuli mos-hA-keyss-ta hA-
nun] kes is not simply an ordinary relative clause meaning “the thing that my late
father or my younger brother couldn’t do.” Rather, the relative clause VP-ta hanun
further includes quotative information, thereby yielding the richer interpretation
“the thing that my late father and my younger brother said they could not do.”
(17) senchin-ina tongsAyng-i ta kuli
late.father-or younger.brother-nom all so
mos-hA-keyss-ta hA-nun kes-Al
neg-do-vol-dec say-adn nmlz-acc
“the thing that my late father or my younger brother said they could not do”
(Hancwunglok 1795)
In the 19th century, attributive quotative VP-ta hanun underwent phonologi-

cal reduction to form attributive evidential tanun, as seen in (18), with hearsay
evidential meaning implied despite the non-overt manifestation of the say mor-
pheme ha. At this stage, similar to attributive quotative VP-ta hanun, attributive
evidential VP-tanun was still adjoined to an overt head noun. In other words, at-
tributive evidential VP-tanun constructions in the 19th century were still relative
clauses, hence embedded and dependent (i.e., still subordinate).
(18) koyu-myengsa-nun heyllae wen-mwun-taylo phyokiha-n-tanun
proper-noun-top Hebrew original-text-as write-impfv-adn
thukcing-ul cini-ko iss-ta
feature-acc retain-seq exist-dec
“(The Bible) has a feature that is said to write a proper noun as is in the
Hebrew original text.” (Yeyswusengkyocense 1887)
When used in sentence-final position from the 20th century onward, tanun con-
structions can appear as ‘headless’ constructions (i.e., no longer relative clause
20. Some gloss -nun as attr (for attributive form) instead of adn (for adnominal function).

All rights reserved
constructions and hence no longer needing to be referential), and as seen in (20),

sentence-final tanun can now be reinterpreted as a conclusive (or finite) hearsay ev-
idential marker. In other words, the previously nominal [VP-ta hanun (N)] struc-
tures as in (17) are now often realized as insubordinate ‘stand-alone’ constructions
via ellipsis of the head noun (N), and with the phonologically reduced attributive
evidential marker tanun reinterpreted as a conclusive (i.e., finite) evidential marker
as seen in (19) (see Ahn & Yap 2014).21 Such uses of insubordinated tanun con-
structions often have the flavor of mirative ‘stand-alone’ nominalization construc-
tions (Noonan 1997, Evans 2007, Watters 2008, Yap et al. 2011 and papers therein).
These highly subjective uses of hearsay evidential tanun constructions are more
often seen in internet communication, particularly among the younger generation.
(19) ecey kangnam-yek kunche-ey
yesterday name-station near-loc
pwul-na-ssess-tanun
fire-break.out-pst.perf-evid
“It was said that a fire had broken out near Kangnam station yesterday.”
(Google)22
Horie (2012) points out that attributive quotative tanun has also evolved into a
‘quotative’ sentence-final particle that allows the speaker to present his/her own
thought or experience with a sense of detachment, as in (20a). In other words,
by using tanun, speakers are able to express their own experience as if they are
reporting something that is said or told by someone else. Horie further notes that
speakers often choose the attributive form in order to suppress the force of their
assertion. As seen in (20b), speakers can also deploy tanun to present the their
thoughts or feelings with a sense of detachment. Essentially, in contexts such as
(20a) and (20b), sentence-final tanun serves as an impersonal evidential marker
to mitigate any potential face-threat to the speaker when he/she ventures to assert
new information (e.g., Brown & Levinson 1987, Goldberg1981).
(20) a. yang-i kkway manh-te-lakwu-yo.
volume-nom pretty much-evid-sfp-pol
twul-i mek-taka namky-ess-tanun
two-nom eat-while leave-pst-evid
21. Native speakers of Korean might consider sentence-final tanun constructions incomplete
sentences. That is, they recognize that there are some omissions going on in such sentences.
For instance, in (19), native speakers would think that a nominal expression (e.g., mal “word,”
somwn “rumor”) or a predicate (mal-i-ta “word-is-dec” or kes-ani-ni “thing-is.not-int”) has
been omitted.
22. Example (19) is from www.google.co.kr (accessed 28 July 2013).

All rights reserved
“The volume (of the food) was pretty much. We left (it), while two of us
ate it, (they) say …”
(in reality uttered by the speaker, who uses tanun to express detachment;
from Horie 2012)
b. nwuka kulye-ss-nunci?
who draw-pst-q
salccak yepkisulep-tanun
slightly be.bizarre-evid
“Who drew (it)? It’s a little bizarre, (they) say.”
(in reality uttered by the speaker, who again uses tanun to express
detachment; from Horie 2012)
The sense of detachment associated with tanun contrasts sharply with the emphat-
ic or assertive stance of tako. As noted earlier, the distancing effect seen in tanun
constructions arises from the weakened assertion inherent to clauses with attribu-
tive rather than conclusive endings. This weakened assertion could also be derived
from the omission or ellipsis of the complement-taking predicate or matrix clause
(e.g., mal-i-ta or kes-i-ta “it is (the fact/thing) that”); in other words, the speaker’s
feeling of detachment could be an effect of not completing the whole sentence.
The detachment effect closely associated with tanun also has other pragmat-
ic consequences. For example, tanun is used neither as a confirmation-seeking
marker nor a counterexpectation marker. Unlike sentence-final tamye and tamy-
ense, found only in interrogative and mirative contexts, sentence-final tanun does
not occur in these environments. Differences in the choice of suffixes (e.g., conver-
bal connective -mye/-myense vs. attributive -nun) thus contribute to pragmatic dif-
ferences, with tamye and tamyense constructions being oriented more toward ad-
dressee engagement, while tanun constructions are oriented more toward speaker
detachment, a type of disengagement.
Figure 4 summarizes the development of attributive quotative VP-ta hanun
into evidential tanun, as well as ‘impersonal evidential’ sentence final particle ta-
nun.

(not attested; no connective function)
Attributive quotative ha-nun (18th c.)
Attributive evidential tanun (19th c.)
Sentence final evidential tanun (20th c.)
‘Impersonal evidential’ tanun (20th c.)
*Confirmation-seeking tanun (not attested)
*Counterexpectation tanun (not attested)
Figure 4. The grammaticalization of tanun in Middle, Modern and Contemporary Korean.

All rights reserved
3.5 On the grammaticalization of tanta
The development of the hearsay evidential marker tanta was slightly different from
the other four hearsay evidential markers discussed earlier. Although tanta was
similarly derived from a say construction via verbal elision and syllable fusion,
specifically, via the VP-ta ha-n-ta > VP-tanta pathway, neither ha-n-ta nor tanta
had quotative uses. The source of information in the VP-ta ha-n-ta and VP-tanta
constructions was always non-specific, often involving indefinite referents such as
‘people’ or ‘someone,’ sometimes expressed explicitly but more often simply im-
plied, hence inducing a reportative or hearsay evidential function rather than a
quotative one, as in (21a) and (21b), respectively. In (21a), the still-lexical use of
ha “say” produces an evidential strategy. In (21b), on the other hand, as a result of
elision of ha “say” and phonological fusion, the more grammaticalized tanta has
become a hearsay evidential marker.
Okchyen-sye-nan i-tal
(21) a. polum-nal nah-a-ni
Okchyen-loc-top this-month 15th-day bear-pst-as
atal naha-ta ha-n-ta
son bear-dec say-pres-dec
“As in Okchyen, (a woman) bore a child on the 15th of this month, and
it is said that she bore a son.” (Swunchenkimssienkan 1502)
b. hyeng-a etAy ka-s-tAnta
older.brother-nom somewhere go-pst-evid
“It is said that older brother went somewhere.” (Pakthongsaenhay 1677)
From an interpersonal perspective, speakers use tanta to present information that is

new and unfamiliar to the addressee in a casual and friendly rather than formal or
serious manner. The purpose is to intentionally not assert the new information, and
so mitigate any potential threat to the face needs of the addressee, as seen in (22).
(22) A: payktwu-san-i hankuk-eyse ceyil
pn-mountain-nom Korea-loc most
noph-tanta
be.high-evid
“Mt. Payktwu is the highest in Korea, they say.”
(speaking in a friendly manner)
B: neyey
“Yes, I see.” (Sejong Contemporary Spoken Corpus)
The deployment of tanta in this solidarity-enhancing manner has much to do with

protecting the addressee’s face. This pragmatic function of tanta differs from that
of tanun, which focuses more on protecting the speaker’s face, as seen in (20a) and
(20b) above. Interestingly, this solidarity-enhancing function of sentence-final

All rights reserved
tanta is closer to the addressee-engaging function associated with sentence-final

tamye and tamyense, except that tanta seeks to enhance solidarity subtly by dis-
tancing the topic and allowing the speaker not to come across as being too asser-
tive, while tamye and tamyense seek to enhance solidarity by directly appealing to
(presumed) shared knowledge, often in confirmation-seeking contexts, as seen in
(10) and (15b) earlier.
Consider yet another example of the paradoxically detached-yet-solidarity
enhancing effect of hearsay evidential tanta in contemporary Korean. In (23), two
females are talking about themselves, and the topic soon revolves around someone
being sick. Speaker A wants to know who is sick, and Speaker B reveals that it is
she herself (Speaker B) who is sick. Speaker B delivers this information to Speaker
A using hearsay evidential tanta. Her choice of evidential conveys more than the
fact that the source of this information comes from a third party, in this case her
doctor; it signals her desire to discuss this delicate topic from a distal perspective
and in an impersonal, non-face-threatening manner to both herself and her in-
terlocutor. In this sense, hearsay evidential tanta is used to enhance intimacy and
solidarity between the speaker and the addressee.
(23) A: mwusun il iss-e? nwuka aph-untey?
what thing exist-ie who be.sick-sfp
“What’s going on? Who is sick?”
→ B: nay-ka aphu-tanta
1sg-nom be.sick-evid
“I hear it said that I am sick.” (Sejong Contemporary Spoken Corpus)
Speakers also often use tanta when boasting. In (24), the speaker uses tanta while
boasting that she is the prettiest girl in her class. The use of tanta allows the speak-
er to present the situation as if she is reporting what is said or told by someone
else; such verbal indirectness can be a useful means of minimizing potential face-
threats arising from the act of boasting. Speakers also deploy tanta to present their
thoughts or feelings, seemingly through the voice of others, and thereby again
prevent damaging the face of either speaker or addressee.
(24) nay-ka wuli-pan-eyse ceyil yeyppu-tanta
I-nom we-class-in most be.pretty-evid
“I am the prettiest (girl) in my class” (spoken in a boastful but playful manner)
(Sejong Contemporary Spoken Corpus)
The playful interpretation of (24) arises from the pragmatic incongruity associated
with self-boasting. In other words, we see the hearsay evidential marker tanta used
in a counterexpectation context for humorous effect.

All rights reserved
The grammaticalization of sentence-final tanta into a hearsay evidential mark-

er and subsequently into a solidarity marker with face-threat mitigating functions
is summarized in Figure 5.

Hearsay evidential ha-n-ta (16th c.)
Hearsay evidential tanta (17thc.)
Solidarity-enhancing tanta (20th c.)
Face-threat-mitigating tanta (20thc.)
Self-praising attenuator tanta (20th c.)
Figure 5. The grammaticalization of tanta in Middle, Modern and Contemporary
Korean.
To sum up, from a morphosyntactic or structural perspective, in this section we

have seen a recurrent strategy involving the elision of ha “say” and phonological
fusion in the development of five hearsay evidential markers. These include tako
(derived from VP-ta ha-ko constructions), tamye (from VP-ta ha-mye), tamyense
(from VP-ta ha-myense), tanun (from VP-ta ha-nun) and tanta (from VP-ta ha-
n-ta). This paradigmatic development suggests a robust VP-ta ha-X > VP ta-X
reanalysis via elision of ha “say”, giving rise to a new paradigm of say-derived
evidential markers in Korean. Future diachronic studies can examine the develop-
ment of other hearsay evidential markers within this relatively recent paradigm
(e.g., tapnita from VP-ta ha-p-ni-ta).
From a grammaticalization perspective, the evidence suggests that tako, tamye
and tamyense developed via a common strategy involving the use of their connec-
tive endings (i.e., -ko “and,”-mye “when, while, because” and the more emphatic
variant -myense which not only has temporal and causal readings but has further
developed a strong concessive reading as well). These ‘connective’-based hearsay
evidential markers frequently modulate (e.g., upgrade or downgrade) the epis-
temic strengths of declarative statements, and they are also often used in interrog-
ative contexts as confirmation-seeking markers and in mirative contexts to express
counterexpectation (e.g., surprise, disbelief, disagreement and disapproval) on the
part of the speaker.
Tanun shows a slightly different grammaticalization strategy. This could be
related to the adnominal origin of -nun, which has developed into an attributive
present tense marker and more recently in Contemporary Korean is emerging as
a conclusive (i.e., finiteness) marker for stand-alone tanun constructions (see §4,
in particular Tables 4 and 5, which show a dramatic increase in the usage of tanun
constructions that pave the way for the emergence not only of attributive but also
conclusive evidential uses of tanun in 20th-century Korean.)

All rights reserved
Tanta, on the other hand, has an inherently finite morphosyntactic structure.

Unlike the ‘connective’-based hearsay evidential markers tako, tamye and tamy-
ense or the adnominal-based attributive hearsay evidential tanun, which started
out within dependent converbal or adnominal clauses respectively, tanta instead is
derived from finite (i.e., structurally independent) say constructions. The deriva-
tion involves a VP-ta ha-n-ta > VP tanta reanalysis, via elision of the say verb ha
and phonological fusion of sentence-final particle -ta from the erstwhile comple-
ment clause and the finite sentence-final particle -n-ta (a combination of attribu-
tive present tense marker-n and declarative sentence final particle -ta for the entire
hearsay construction). The finite status of the complement-taking verb ha-n-ta
may have accounted for the earlier development of tanta as a grammaticalized
hearsay evidential marker as early as the 17th century, as in (21b), approximately
three centuries ahead of the other four hearsay evidential markers — namely, tako,
tamye, tamyense and tanun, which did not grammaticalize into evidential markers
until the 20th century.23
4. Frequency analysis of say evidentials and their pragmatic uses in Korean
In this section, we use frequency analysis to examine the development of the above-
mentioned say-derived evidential markers. The historical data for our analysis are
based on the UNICONC database (Korean historical corpus), which comprises
267,808 words from texts from the 16th century, 457,443 words from the 17th
century, 664,733 words from the 18th century, 1,824,97319 words from the 19th
century and 2,209,352 words from the 20th century. Our analysis of contemporary
Korean is based primarily on the Sejong spoken corpus, which comprises 4,204,082
words, and the Sejong written corpus, which consists of 63,632,472 words. Table 1
shows the frequency of lexical uses of tako, tamye, tamyense, tanun and tanta (i.e.,
their usage as say verbs) from the 17th century to the early 20th century based on
the UNICONC historical corpus. Table 2 shows the frequency of their quotative
and complementizer uses, and Table 3 shows the frequency of their evidential uses.
Table 1 reveals that uses of the morpheme ha “say” as a lexical verb are at-
tested from the 16th century in conclusive (i.e., finite) VP-ta ha-n-ta structures (4
tokens) and in converbal (non-finite) structures such as VP-ta ha-ko and VP-ta
23. An anonymous reviewer asked if it is possible that tanta might have acted as a catalyst for
the grammaticalization of the other say-derived forms. Our answer is yes, but the phenomenon
is rather complex. Tako may have emerged earlier as a complementizer and thus served as a
catalyst for the grammaticalization of tamye and tamyense (via analogy), while tanta may have
served as catalyst for the reanalysis of converbal and attributive uses of the hearsay evidentials
tako, tamye, tamyense and tanun as conclusive sentence-final evidential particles.

All rights reserved
Table 1. Frequency of use of lexical verb ha “say” with connectives -ko, -mye and -my-
ense, adnominal-nun and conclusive ending -nta between the 16th to early 20th centuries
(based on the UNICONC historical corpus).
Century Frequency of use of ha “say” as lexical verb
(raw tokens, with no. of tokens per 100,000 words in parentheses)
VP-ta VP-ta VP-ta VP-ta VP-ta
hako hamye hamyense hanun hanta
16th 57 (21.28) 9 ( 3.36) 0 (0.00) 0 (0.00) 4 (1.49)
17th 106 (23.17) 5 ( 1.09) 0 (0.00) 0 (0.00) 23 (0.51)
18th 192 (28.88) 48 (10.49) 0 (0.00) 1 (0.15) 1 (0.15)
19th 543 (29.75) 192 (10.52) 1 (0.06) 4 (0.22) 19 (1.04)
early 20th 911 (41.23) 300 (13.58) 2 (0.09) 11 (0.50) 10 (0.45)
ha-mye (57 tokens and 9 tokens respectively). Attributive VP-ta ha-nun is attested
in the 18th century, and the more emphatic converbal VP-taha-myense is attested
later in the 19th century. This chronological sequence helps explain why, as seen in
Table 2 below, quotative uses are attested earlier for converbal tako and tamye and
also for attributive tanun (i.e., in the 18th century) than for late-comer converbal
tamyense (whose quotative uses are not attested in the UNICONC database until
the 19th century).
Table 2. Frequency of use of tako, tamye, tamyense, tanun and tanta as quotatives be-
tween the 17th and early 20th centuries (based on the UNICONC historical corpus).
Century Frequency of use as quotative/complementizer
tako24 tamye tamyense tanun tanta
17th 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00)
18th 4 (0.60) 3 (0.45) 0 (0.00) 8 (1.20) 0 (0.00)
19th 2,254 (123.51) 9 (0.49) 1 (0.06) 62 (3.40) 0 (0.00)
20th 2,460 (111.35) 21 (0.95) 1 (0.05) 77 (3.49) 0 (0.00)
Note: Tako has both quotative and complementizer uses; tamye, tamyense and tanun have quotative uses
but not a complementizer use where the say-derived marker is followed by a non-utterance verb; tanta has
neither a quotative nor complementizer use.
24. There were only sporadic uses of tako as a quotative and complementizer during the 18th
century (with 4 tokens attested), but the frequency of complementizer use increased expo-
nentially from the late 19th century onward. This coincides with the emergence of a new col-
loquial style in Korean literature along with modern-style newspapers such as Toklipsinmun
“Independence.” Indeed, the vast majority of uses of tako as a complementizer during this period
were in Toklipsinmun (with a raw frequency of 2,211 tokens out of a total of 2,254 tokens, i.e.,
98.1%). In addition, the frequency of the [VP-ta hako] construction increased substantially in
the 18th century. Due to this increase, the grammaticalization of tako seemed to be expedited
during this period.

All rights reserved
Table 2 indicates that conclusive tanta is not attested with a quotative function.
Instead, as seen in Table 3, tanta developed into a hearsay evidential marker in the
17th century and this became productive from the 19th century onward. Crucially,
conclusive tanta developed into a sentence-final hearsay evidential marker much
earlier than either converbal tako, tamye and tamyense or attributive tanun. This,
we suggest, is closely linked to the inherently finite status of conclusive tanta, hence
the greater ease with which it is reanalysed as a sentence-final evidential marker,
whereas the converbal and attributive evidential forms need to be productively used
as insubordinate structures to be reanalysed as sentence-final evidential markers.
This would help explain the differential rates of grammaticalization for different
types of hearsay evidential markers in Korean, and highlights the close link be-
tween evidential marking and finiteness (see also Nikolaeva 2010, Bisang 2007).
Table 3. Frequency of use of tako, tamye, tamyense, tanun and tanta as evidential mark-
ers in sentence-final position between the 17th and early 20th centuries (based on the
UNICONC historical corpus).
Century Frequency of use as evidential marker
tako tamye tamyense tanun tanta
17th 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00) 2 (0.44)
18th 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00)
19th 0 (0.00) 0 (0.00) 0 (0.00) 0 (0.00) 31 (1.70)
20th 37 (1.68) 0 (0.00) 0 (0.00) 0 (0.00) 88 (3.98)
As seen in Tables 4 and 5 below, usage frequency data from the Sejong corpus for
contemporary Korean (i.e., from the late 20th century to the present) reveal fairly
productive use of tako, tamye, tamyense, tanun and tanta as hearsay evidential
markers.25 These hearsay evidential markers show some subtle differences in their
extended pragmatic uses. For example, converbal evidential markers tako, tamye
and tamyense are used in confirmation-seeking and counterexpectation-marking
contexts, with tako also often found in cleft constructions that highlight the speak-
er’s emphatic stance. The attributive evidential marker tanun, on the other hand, is
sometimes used as an impersonal evidential marker (e.g., when the speaker wishes
to convey his or her own utterance as though the information comes from some-
one else or from general hearsay), and the conclusive evidential tanta is used as a
25. Note that tako has both quotative (qt) and complementizer (comp) tokens combined. In
this analysis, quotatives are followed by utterance verbs, while complementizers are followed by
non-utterance verbs. Tamye, tamyense and tanun only have quotative uses.

All rights reserved
solidarity marker (e.g., when the speaker wishes to politely create distance from a
sensitive topic by adding a hearsay marker).26
Table 4. Frequency distribution of the functions of tako, tamye, tamyense, tanun and tanta
in Contemporary Korean (based on the Sejong contemporary spoken corpus).27
qt/comp evid cs ce emph Impersonal Solidarity
evidential marker
tako 19,907 390 50 23 327 0 0
tamye 162 31 13 2 0 0 0
tamyense 180 23 10 1 0 0 0
tanun 19,827 18 0 0 0 13 0
tanta 0 4 0 0 0 0 2
Table 5. Frequency distribution of the functions of tako, tamye, tamyense, tanun and tanta
in Contemporary Korean (based on the Sejong contemporary written corpus).
qt/comp evid cs ce emph Impersonal Solidarity
evidential marker
tako 1,028,645 1049 2 14 69 0 0
tamye 219,226 173 5 2 0 0 0
tamyense 78,242 62 1 2 0 0 0
tanun 1,028,748 0 0 0 0 0 0
tanta 0 1071 0 0 0 0 219
In terms of register variation, pragmatic uses of tako, tamye and tamyense — for
example the confirmation-seeking (cs) and counterexpectation-marking (ce)
functions — occur with higher frequency in the spoken register, highly consistent
with the role of pragmatic markers as conveyors of speaker’s subjective and inter-
personal (hence also intersubjective) stances.
Pragmatic uses of tanun show an even stronger bias for the spoken register,
with no hearsay and impersonal evidential tokens attested in the written corpus,
26. There is a dramatic increase in the use of attributive evidential tanun in both spoken and
written Korean from the late 20th century. The reason is not fully clear to us and deserves fur-
ther investigation.
27. Impersonal evidential = detached-style quotative, used by the speaker to present his or her
own quote as though the information comes from some other person; Solidarity marker = a
particle used by the speaker to distance and de-sensitize a topic that may be face-threatening to
speaker or hearer.

All rights reserved
while 18 tokens of hearsay evidential tanun and 13 tokens of impersonal evidential

tanun are attested in the spoken corpus. As discussed in §3.4, this is not surprising
given that insubordination of an embedded attributive evidential tanun that gives
rise to an independent sentence-final (i.e., conclusive) evidential tanun calls for
expressive illocutionary force that is more characteristic of the colloquial speech
domain. Specifically, it involves the reinterpretation of tanun from an evidential
marker with [+present, +attributive] features to one with [+present, -attributive]
features, and as a result yielding a conclusive ‘stand-alone’ hearsay evidential con-
struction that is further capable of developing into an impersonal evidential, as
highlighted in Figure 6 below.
Stage 1: Attributive evid [[X-ta ha-nun] Relative-clause(N)] ‘N which is said to be X’

Stage 2: Attributive evid [[X-tanun] Relative-clause(N)] ‘N which is said to be X’
Stage 3: Hearsay evid [X-tanun] Insubordinate-clause ‘X, it’s said’
Stage 4: Impersonal evid [X-tanun] Insubordinate-clause ‘X, it’s said’ (actually said by the
speaker but attributed to others)
Figure 6. Development of tanun in terms of subordinate/insubordinate clause structure.
On the other hand, pragmatic uses of tanta predominate in the written register
(with 219 tokens of solidarity marker tanta in Table 5 and only 2 in Table 4). Tanta
constructions are inherently finite, and are not used in interrogative or mirative
mood but rather are mostly found in declarative mood.
In sum, we see both usage frequency and morphosyntactic factors playing an
important role in the development of say evidentials and their extended pragmat-
ic functions. The effects of usage frequency are seen in the faster rate of grammati-
calization of tako compared to tamye and tamyense. At the same time, frequency
effects are also still constrained by morphosyntactic factors. In the case of tanta, its
inherent finite structure provides it with a headstart in developing a sentence-final
hearsay evidential function; in the case of tanun, on the other hand, its attributive
form helped delay its development into a sentence-final hearsay evidential. The
usage frequency data from the UNICONC historical corpus in combination with
the Sejong contemporary corpus has thus shown us that both morphosyntactic
facilitation and usage frequency are important in charting the course of gram-
maticalization and pragmaticization.
5. Conclusion
We have examined the development of say-derived constructions and their ex-

tended uses within the pragmatic domain in Korean discourse, specifically the

All rights reserved
development of tako, tamye, tamyense, tanun and tanta as hearsay evidential mark-
ers. Our analysis reveals that phonological reduction played a significant role in
the rise of the grammatical and pragmatic uses of these say-derived constructions
in Korean. In particular, elision of the say verb ha was prominent in the develop-
ment of all five hearsay evidential markers. The development of other say-derived
evidential markers such as tani and tapnita also appear to involve say elision, and
these markers should also be investigated in future research. Despite phonological
elision of the say morpheme ha, the semantics of the say verb is often retained
to varying degrees, with different degrees of semantic bleaching conditioned by
context of use.
Morphosyntactic factors also play an important role in the development of say
evidentials and their extended pragmatic functions. For instance, converbal tako,
tamye and tamyense constructions can be insubordinated as stand-alone (hence
finite) hearsay evidential constructions, often used with confirmation-seeking and
counterexpectation-marking functions. Often, such insubordination phenomena
are triggered by an ‘echoing’ (or repetition) function, and typically realized via ei-
ther interrogative or mirative mood. Tanun constructions are insubordinated in-
stead by means of attributive -nun being reanalysed as a tense marker. On the other
hand, tanta constructions are quite distinct from the other four say-derived con-
structions in that they are inherently finite by virtue of the attributive present-tense
suffix -n and the declarative sentence-final particle -ta, and thus tanta constructions
need not undergo insubordination the way the other hearsay evidential markers do.
In short, we see how structural independence can be achieved. One strategy
involves the insubordination of dependent (non-finite) clauses into independent
(finite) ones. This could involve the grounding of subordinate connective clauses
via the speaker’s illocutionary force (e.g., interrogative mood or mirative mood)
to form stand-alone insubordinate clauses, as in the case of the tako, tamye and
tamyense constructions. Or it could involve the reanalysis of attributive forms as
conclusive (hence finite) forms, as in the case of the tanun constructions. Another
strategy — or ‘non-strategy’ — is to be born with a silver spoon, as the saying
goes, and be inherently finite and thus independent from the very beginning, as
in the case of tanta, but even here we see the effects of phonological reduction and
syntactic scope expansion (strongly induced by high usage frequency), such that
VP-ta ha-n-ta constructions come to be realized as VP-tanta constructions, the
former still a lexically transparent evidential strategy, while the latter has gram-
maticalized with tanta reanalysed as an evidential marker.
Based on diachronic usage frequency data, we see that tako developed into
a highly versatile quotative and complementizer, with a hearsay evidential usage
emerging in contexts where the author of the reported quote is not specified. We
have also shown that this development involving say-derived hearsay evidentials

All rights reserved
is robust, with usage frequency also playing an important role, as seen in the faster
rate of grammaticalization of the highly productive tako compared to the less pro-
ductive tamye and tamyense. Yet, as is also clearly evident from the asymmetrical
rate of development of tanta and tanun, usage frequency is not the sole factor
determining the rate of grammaticalization; there is also an important role for
morphosyntactic facilitation. For example, the presence of the present-tense suf-
fix -n and the declarative mood marker -ta in tanta automatically triggers a finite
reading, and depending on whether the source of information is specified or not
specified, tanta will be interpreted as either a sentence-final quotative or a sen-
tence-final hearsay evidential marker. Tanun, on the other hand, is an adnominal
form, and tanun constructions need to be insubordinated prior to its reanalysis as
a sentence-final evidential marker; this accounts for the slightly delayed develop-
ment of tanun from non-finite to finite evidential marker.
We have also examined how these say evidential and pragmatic markers are
used in natural discourse. In particular, we have seen how they are used to addi-
tionally express the interpersonal and intersubjective stances of the interlocutors:
(i) tako has the pragmatic function of asserting the speaker’s epistemic knowledge
about a particular topic in the ongoing discourse; (ii) both tamye and tamyense
have the solidarity-enhancing function of downgrading the speaker’s epistemic
claims while acknowledging the epistemic claims of the addressee, with tamyense
being more emphatic and often associated with a hint of insinuating blame upon
the addressee for the adverse situation being discussed in the conversation; (iii)
tanun has the function of detaching the speaker from his or her own prior utter-
ance when it is potentially face-threatening for the speaker to be identified as the
source of the information and (iv) tanta is employed to intentionally not assert
the information provided by the speaker, and in this way helps to mitigate any
potential threat to the face needs of both the speaker and the addressee, and thus
enhance the solidarity between the two interlocutors.
Our findings have interesting implications for other languages as well. Some
parallels have already been observed in neighboring languages such as Japanese
tte constructions and Cantonese wo constructions, which likewise display evi-
dential and sentence-final pragmatic functions (S. Suzuki 1999, R. Suzuki 2007
for Japanese). It would be interesting to further investigate whether, and to what
extent, the say-derived evidential constructions also develop other extended prag-
matic uses, for example as evidentials with epistemic and/or attitudinal functions,
either at the right periphery as sentence final particles, or at the left periphery as
sentence adverbials. Another area for further research, based on the present study
on Korean as well as recent work on Chinese (see Yap et al. forthcoming) and on
Japanese (Tamaji & Yap ms.), is the relationship between evidential marking and

All rights reserved
finiteness marking, which can help us better understand how, as well as which,
linguistic constructions come to be realized as independent structures.
Abbreviations
acc accusative imp imperative

adn adnominal intj interjection
ce counterexpectation loc locative
comp complementizer neg negation
conn connective nmlz nominalizer
cop copula nom nominative
cs confirmation-seeking pl plural
dec declarative pol polite
dm discourse marker pres present
emph emphatic pst past
evid evidential q interrogative
excl exclamatory qt quotative
gen genitive seq sequential
hor hortative sfp sentence final particle
ie informal ending top topic
References
Ahn, Ju-Ho. 2006. A studying of connecting ending -meyenseo in Korean. Studies in Modern
Grammar 45. 179–198.
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. & Robert M. W. Dixon. 2003. Studies in evidentiality. Amsterdam:
John Benjamins. DOI: 10.1075/tsl.54
Aikhenvald, Alexandra Y. & Randy J. LaPolla. 2007. New perspectives on evidentials: A view
from Tibeto-Burman. Linguistics of the Tibeto-Burman Area 30(2). 1–16.
Anderson, Lloyd B. 1986. Evidentials, paths of change and mental maps: Typologically regular
asymmetries. In Wallace L. Chafe & Johanna Nichols (eds.), Evidentiality: The linguistic
coding of epistemology, 273–312. Norwood, NJ: Ablex.
Bisang, Walter 2007. Categories that make finiteness: Discreteness from a functional perspective
and some of its repercussions. In Irina Nikolaeva (ed.), Finiteness: Theoretical and empirical
foundations, 115–137. Oxford: Oxford University Press.
Brown, Penelope & Stephen Levinson. 1987. Universals in language usage: Politeness phenom-
ena. In Esther N. Goody (ed.), Questions and politeness: Strategies in social interaction, 256–
289. Cambridge: Cambridge University Press.

All rights reserved
Chafe, Wallace L. & Johanna Nichols (eds.). 1986. Evidentiality: The linguistic coding of episte-
mology. Norwood, NJ: Ablex.
Chang, Miao-Hsia. 1998. The discourse functions of Taiwanese kong in relation to its gram-
maticalization. In Huang Shuanfan (ed.), Selected papers from the Second International
Symposium on Languages in Taiwan, 111–127. Taipei: Crane Publishing Company.
Chappell, Hilary. 2008. Variation in the grammaticalization of complementizers from verba di-
cendi in Sinitic languages. Linguistic Typology 12(1). 45–98. DOI: 10.1515/LITY.2008.032
Chung, Kyung-Sook. 2006. Korean evidentials and assertion. In Donald Baumer, David
Montero & Michael Scanlon (eds.), Proceedings of the 25th West Coast Conference on
Formal Linguistics, 105–113. Somerville, MA: Cascadilla Proceedings Project.
Chui, Kawai. 1994. Grammaticalization of the saying verb wa in Cantonese. Santa Barbara
Papers in Linguistics 5. 1–13.
Cornillie, Bert. 2009. Evidentiality and epistemic modality: On the close relationship between
two different categories. Functions of Language 16(1). 44–62. DOI: 10.1075/fol.16.1.04cor
DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 33. 369–382. DOI:
10.1016/S0378-2166(01)80001-1
Dryer, Matthew S. 1996. Focus, pragmatic presupposition, and activated propositions. Journal of
Pragmatics 26. 475–523. DOI: 10.1016/0378-2166(95)00059-3
Ekberg, Lena & Carita Paradis. 2009. Editorial: Evidentiality in language and cognition.
Functions of Language 16(1). 5–7. DOI: 10.1075/fol.16.1.02ekb
Eckardt, Regine. 2006. Meaning change in grammaticalization: An enquiry into semantic reanaly-
sis. Oxford: Oxford University Press.
Evans, Nicholas. 2007. Insubordination and its uses. In Irina Nikolaeva (ed.), Finiteness:
Theoretical and empirical foundations, 366–431. Oxford: Oxford University Press.
Faller, Martina. 2002. Semantics and pragmatics of evidentials in Cuzco Quechua. Stanford:
Stanford University dissertation.
Fox, Barbara A. 2001. Evidentiality: Authority, responsibility, and entitlement in English conver-
sation. Journal of Linguistic Anthropology 11(2). 167–192. DOI: 10.1525/jlin.2001.11.2.167
Gipper, Sonja. 2011. Evidentiality and intersubjectivity in Yurakará: An interactional account.
Nijmegen: Max Planck Institut für Psycholinguistik.
Goldberg, Lewis. R. 1981. Language and individual differences: The search for universals in
personality lexicons. In Alan Wheeler (ed.), Review of personality and social psychology 1,
141–165. Beverly Hills: Sage.
Güldemann, Tom. 2008. Quotative indexes in African languages: A synchronic and diachronic
survey. Berlin: Mouton de Gruyter. DOI: 10.1515/9783110211450
Grunow-Hårsta, Karen. 2011. Innovation in nominalization in Magar, a Tibeto-Burman
language of Nepal. In Foong Ha Yap, Karen Grunow-Hårsta & Janick Wrona (eds.),
Nominalization in Asian languages: Asian perspectives (Typological Studies in Language
96), 215–254. Amsterdam: John Benjamins. DOI: 10.1075/tsl.96.08gru
Güldemann, Tom & Manfred von Roncador. 2002. Reported discourse: A meeting ground for dif-
ferent linguistic domains. Amsterdam: John Benjamins. DOI: 10.1075/tsl.52
Hansen, Maj-Britt Mosegaard & Jacqueline Visconti. 2007. On the diachrony of reinforced ne-
gation in French and Italian. In Corinne Rossari, Claudia Ricci & Adriana Spiridon (eds.),
Grammaticalization and pragmatics: Facts, approaches, theoretical issues, 137–171. Bingley,
UK: Emerald.
Heine, Bernd &Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge
University Press. DOI: 10.1017/CBO9780511613463

All rights reserved
Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. 1991. From cognition to gram-
mar: Evidence from African languages. In Elizabeth C. Traugott & Bernd Heine (eds.),
Approaches to grammaticalization, vol. 1, 149–187. Amsterdam: John Benjamins. DOI:
10.1075/tsl.19.1.09hei
Hill, Jane & Judith Irvine (eds.). 1993. Responsibility and evidence in oral discourse. Cambridge:
Cambridge University Press.
Hopper, Paul J. & Elizabeth C. Traugott. 2003. Grammaticalization, 2nd edn. Cambridge:
Cambridge University Press. DOI: 10.1017/CBO9781139165525
Horie, Kaoru. 2012. The convergence of stance-related functions of nominalization in
Modern Japanese. Paper presented at the Workshop on Stance and Discourse: Discourse
Perspectives, Hong Kong Polytechnic University, May 7-9.
Hsieh, Fuhui. 2012. On the grammaticalization of the Kavalan SAY verb zin. Oceanic Linguistics
51(2). 467–492. DOI: 10.1353/ol.2012.0025
Jeon, Hye-Young. 1996. On pragmatic functions of ‘-tako’-ending repercussive question. Korean
Journal of Linguistics 21(3). 889–911.
Johanson, Lars & Bo Utas. 2000. Evidentials: Turkic, Iranian and neighbouring languages. Berlin:
Mouton de Gruyter. DOI: 10.1515/9783110805284
Kim, Mary Shin. 2006. Evidential strategies in Korean conversation: An analysis of interactional
and conversational narrative functions. Los Angeles: University of California, Los Angeles
dissertation.
Kim, Mary Shin. 2011. Negotiating epistemic rights to information in Korean conversation: An
examination of the Korean evidential marker -tamye. Discourse Studies 13(4). 435–459.
DOI: 10.1177/1461445611403259
Kim, Minju. 2011. Grammaticalization in Korean: The evolution of the existential verb. London:
Saffron Books.
Kim, Su Tae. 2005. Machimpep Ssikkutyuy Yunghapkwa Ku Hankyey [Fusion of sentential endings
and its limit]. Seoul: Pakiceng Publishing.
Konoshima, Masatoshi. 1973. Kokugo Joshi no Kenkyuu [A study of particles in Japanese]. Tokyo:
Oofuusha.
Kwon, Iksoo. 2009. The Korean evidential marker -te- revisited: Its semantic constraints and
distancing effects. Paper presented at the Discourse Cognitive Linguistics Society of Korea
Conference, Hankuk University of Foreign Studies, June 20.
Larrivée, Pierre. 2011. The role of pragmatics in grammatical change: The case of French prever-
bal non. Journal of Pragmatics 43. 1987–1996. DOI: 10.1016/j.pragma.2010.12.012
Lee, Jungmee. 2010. The Korean evidential -te: A modal analysis. Empirical Issues in Syntax and
Semantics 8. 287–311.
Lee, Keum-Hee. 2006. Grammaticalization of quotation constructions: On the process and de-
gree of grammaticalization. Kwukehak [Korean Lingustics] 48. 233–258.
Leung, Wai-mun. 2006. On the synchrony and diachrony of sentence final particles: The case of wo
in Cantonese. Pokfulam, HK: University of Hong Kong dissertation.
Leung, Wai-mun. 2009. A study of the Cantonese hearsay particle wo from a tonal perspective.
International Journal of Linguistics 1(1). 1–14. DOI: 10.5296/ijl.v1i1.204
Lim, Dongsik. 2010. Evidentials and interrogatives: A case study from Korean. Los Angeles:
University of Southern California dissertation.
Matthews, Stephen. 1998. Evidentiality and mirativity in Cantonese: wo3, wo4, wo5!
Proceedings of the International Symposium on Chinese Languages and Linguistics. 325–
334. Academica Sinica.

All rights reserved
Michael, Lev. 2008. Nanti evidential practice: Language, knowledge, and social action in an
Amazonian society. Austin: University of Texas at Austin dissertation.
Miura, Akira. 1974. ‘To’ to ‘tte’ [To and tte]. Nihonggo Kyooiku [Japanese Language Education]
24. 23–28.
Nam, Mi-jeong. 2010. The formation and meaning of an ending “-dago” and its kind.
Hanmalyenku [Studies on Korean Language] 26. 109–131.
Nikolaeva, Irina. 2010. Typology of finiteness. Language and Linguistics Compass 4(12). 1176–
1189. DOI: 10.1111/j.1749-818X.2010.00253.x
Noonan, Michael. 1997. Versatile nominalizations. In Joan Bybee, John Haiman & Sandra A.
Thompson (eds.), Essays on language function and language type: Dedicated to T. Givón,
373–394. Amsterdam: John Benjamins. DOI: 10.1075/z.82.21noo
Nuyts, Jan. 2001. Subjectivity as an evidential dimension in epistemic modal expressions.
Journal of Pragmatics 33. 383–400. DOI: 10.1016/S0378-2166(00)00009-6
Park, Jinho. 2008. ‘tako’ lyu emiey tayhan hwayongloncek cepkun [Sentential ending ‘tako’ from
a pragmatic approach]. Yeysan Hakpo [Yeysan Studies] 24. 171–194.
Rhee, Seongha. 2008a. Ellipsis and functional shifts. Studies in Modern Grammar 52. 169–194.
Rhee, Seongha. 2008b. On the rise and fall of Korean nominalizers. In María José López-Couso
& Elena Seoane (eds.), Rethinking grammaticalization: New perspectives (Typological
Studies in Language 76), 239–264. Amsterdam: John Benjamins. DOI: 10.1075/tsl.76.12rhe
Rhee, Seongha. 2011. Analogy-driven grammaticalization: A case of grammaticalization of sen-
tence-final markers from concomitance-connectives. Paper presented at the 9th Biennial
Conference of the Association for Linguistic Typology, University of Hong Kong, July 22.
Rudolph, Dina. 1993. Getting past politeness: The role of linguistic markers of evidentiality and
informational domains in Japanese student-teacher interactions at the graduate level. Los
Angeles: University of Southern California dissertation.
Saegusa, Reiko. 1997. Syntax of tte. Gengo Bunka 34. 21–34.
Schwenter, Scott A. 2006. Fine-tuning Jespersen’s cycle. In Betty J. Birner & Gregory Ward (eds.),
Drawing the boundaries of meaning: Neo-Gricean studies in pragmatics and semantics in hon-
or of Laurence R. Horn, 327–344. Amsterdam: John Benjamins. DOI: 10.1075/slcs.80.19sch
Shi, Chung-kon & Keon-hee Kim. 2011. The modal inquiry into the construction of bound
noun + -ida. Emwunyenkwu [Korean Sentence Structure Studies] 68. 79–102.
Shibatani, Masayoshi. 1990. The languages of Japan. Cambridge: Cambridge University Press.
Shin, Mi Kyong. 2005. Development of nominalizers in some East Asian languages. Shatin: The
Chinese University of Hong Kong MA thesis.
Sohn, Sung-Ock. 2011. Historical development of quotative constructions in Korean. Japanese/
Korean Linguistics 18. 126–143.
Sohn, Sung-Ock. 2012. Development of stance markers in Korean: Diachronic and discourse
perspectives. Paper presented at the Tutorial on Discourse Analysis (Workshop on
Epistemicity, Evidentiality and Attitude), Hong Kong Polytechnic University, Sept 2.
Su, Lily I-wen. 2004. Subjectification and the use of the complementizer SHUO. Concentric:
Studies in Linguistics 30(1). 19–40.
Sun, Tianshin Jackson. 1993. Evidentials in Amdo Tibetan. Bulletin of the Institute of History and
Philology 63(64). 945–1001.
Suzuki, Ryoko. 2007. (Inter)subjectification in the quotative tte in Japanese conversation: Local
change, utterance-ness and verb-ness. Journal of Historical Pragmatics 8(2). 207–237. DOI:
10.1075/jhp.8.2.04suz

All rights reserved
Suzuki, Satoko. 1999. Marker of unexpected statements: An analysis of the quotative particle
datte. Journal of the Association of Teachers of Japanese 33(1). 44–67. DOI: 10.2307/489630
Tamaji, Mizuho & Foong Ha Yap. Manuscript. On the role of phonological change and ellipsis
in the development of evidential and pragmatic markers: Diachronic evidence from con-
verbal ‘say’ constructions in Japanese. Department of Chinese and Bilingual Studies and
Department of English, Hong Kong Polytechnic University.
Wang, Yu-Fang, Aya Katz & Chih-Hua Chen. 2003. Thinking as saying: Shuo (‘say’) in Taiwan
Mandarin conversation and BBS talk. Language Sciences 25(5). 457‒488.
Watters, David. 2008. Nominalization in Kiranti and the Central Himalayish languages of Nepal.
Linguistics of the Tibeto-Burman Area 31(2). 1–43.
Willett, Thomas. 1988. A cross-linguistic survey of the grammaticization of evidentiality. Studies
in Language 12. 51–97. DOI: 10.1075/sl.12.1.04wil
Yap, Foong Ha & Mikyung Ahn. 2011. Evidentiality and counterexpectation marking strategies
in Cantonese and Korean. Paper presented at the Workshop on Evidentiality and Mood,
Seoul National University, October 19.
Yap, Foong Ha, Karen Grunow-Härsta & Janick Wrona (eds.). 2011. Nominalization in Asian
languages: Diachronic and typological perspectives (Typological Studies in Language 96).
Amsterdam: John Benjamins. DOI: 10.1075/tsl.96
Yap, Foong Ha & Stephen Matthews. 2008. The development of nominalizers in East Asian and
Tibeto-Burman languages. In María José López-Couso & Elena Seoane (eds.), Rethinking
grammaticalization: New perspectives (Typological Studies in Language 76), 309–341.
Amsterdam: John Benjamins. DOI: 10.1075/tsl.76.15yap
Yap, Foong Ha, Ying Yang & Tak-sum Wong. Forthcoming. On the development of sentence
final particles (and utterance tags) in Chinese. In Kate Beeching & Ulrich Detges (eds.),
Asymmetrical functions at the left and right periphery. Leiden: Brill.
Yeung, Ka-Wai. 2006. On the status of the complementizer waa6 in Cantonese. Taiwan Journal
of Linguistics 4(1). 1–48.
Yu, Hyeon Gyeong. 2002. Emi ‘-tako’ uymiwa yongpep [The meaning and usage of particle
tako]. Paytalmal [Ethnic Korean] 31. 99–122.
Yuzawa, Kookichiroo. 1954. Edokotoba no Kenkyuu [A study of language in the Edo period].
Tokyo: Meijishoin.
Résumé
L’article examine le développement des marqueurs évidentiels du coréen qui expriment la no-
tion de ‘ouï-dire’, à savoir tako, tamye, tamyense, tanun et tanta, et détaille leurs fonctions prag-
matiques étendues dans le discours. Ces fonctions sont d’abord identifiées en diachronie, du
coréen moyen au coréen moderne et contemporain, avant que soit analysée quantitativement
leur fréquence d’emploi du 16ème siècle au début du 20ème siècle, à partir du corpus histo-
rique UNICONC. L’analyse synchronique du coréen d’aujourd’hui est basée enfin sur le corpus
de langue à la fois parlée et écrite du Sejong contemporain. L’article étudie, d’un point de vue
pragmatique, comment les locuteurs coréens utilisent ces marqueurs évidentiels de ‘ouï-dire’
pour véhiculer les positions et valeurs interpersonnelles et intersubjectives des interlocuteurs
dans les conversations naturelles. A partir des différents rythmes de grammaticalisation de ces
marqueurs évidentiels et aussi de leur fréquence d’emploi, les relations entre les marqueurs

All rights reserved
d’évidentialité et la finitude sont approfondies. Sont plus particulièrement analysés les méca-
nismes et les étapes du changement par lequel différents types de structures évidentielles non
finies ont développé des constructions évidentielles finies. Les résultats de cette étude ont des
implications théoriques et typologiques importantes pour une meilleure compréhension des
mécanismes d’insubordination, où des structures dépendantes deviennent indépendantes et où
des constructions transparentes d’un point de vue lexical se développent en marqueurs gram-
maticalisés exprimant la position du locuteur.
Zusammenfassung
In diesem Artikel untersuchen wir die diachrone Entwicklung der fünf reportativen
Evidenzmarkierungen tako, tamye, tamyense, tanun und tanta des Koreanischen, sowie
ihre erweiterten pragmatischen Funktionen. Hierbei stellen wir zunächst ihre unterschiedli-
chen Funktionen zu verschiedenen Zeiten fest, und analysieren dann die Häufigkeiten dieser
Funktionen im Sprachgebrauch, sowohl diachron vom 16. bis zum frühen 20. Jh. anhand des his-
torischen UNICONC Korpus, als auch synchron anhand des Sejong Korpus (geschriebene und
gesprochene Sprache). Es wird aufgezeigt, wie in natürlichen Konversationen interpersonelle und
intersubjektive Perspektiven durch die Verwendung der reportativen Evidenzmarkierungen aus-
gedrückt werden. Basierend auf Unterschieden in der Geschwindigkeit der Grammatikalisierung
und in der Häufigkeit im Sprachgebrauch dieser Evidenzmarkierungen untersuchen wir auch
die Beziehung zwischen Evidentialität und Finitheit, besonders im Hinblick auf die Abfolge
und die Mechanismen der Entwicklung von infiniten zu finiten evidentiellen Konstruktionen.
Es zeigen sich hier auch breitere theoretische und cross-linguistische Implikationen für das
Verständnis von Insubordination, bei der abhängige Strukturen unabhängig werden und le-
xikalisch transparente Konstruktionen zu grammatikalisierten Formen des Ausdrucks der
Sprecherhaltung werden.
Corresponding author’s address

Mikyung Ahn
Hankuk University of Foreign Studies
#408 Faculty Building I
107 Imun-ro
Dongdaemun-gu
Seoul 130-791
Korea
[email protected]

All rights reserved

Ahn - Ha Yap - 2013 - On - The - Development - of - Korean - SAY - Evidentials - and - Their - Expected - Grammatical - Functions

Uploaded by

Copyright:

Available Formats

Ahn - Ha Yap - 2013 - On - The - Development - of - Korean - SAY - Evidentials - and - Their - Expected - Grammatical - Functions

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ahn - Ha Yap - 2013 - On - The - Development - of - Korean - SAY - Evidentials - and - Their - Expected - Grammatical - Functions

Uploaded by

Copyright:

Available Formats

John Benjamins Publishing Company

This is a contribution from Diachronica 31:3

On the development of Korean SAY evidentials

Mikyung Ahn and Foong Ha Yap

This paper examines the development of five hearsay evidential markers in

Keywords: hearsay evidential markers, diachronic corpus analysis, epistemic

* We wish to gratefully acknowledge support from Hankuk University of Foreign Studies

Diachronica 31:3 (2014), 299–336. doi 10.1075/dia.31.3.01ahn

© 2014. John Benjamins Publishing Company

(1) a. Tom-i cip-ul naka-ss-tako?

© 2014. John Benjamins Publishing Company

4. ‘Insubordination’ is the process whereby a syntactically dependent structure (e.g., an embed-

© 2014. John Benjamins Publishing Company

2. Previous studies on evidential say constructions

© 2014. John Benjamins Publishing Company

3. On the grammaticalization of evidential say constructions and their

3.1 On the grammaticalization of tako

© 2014. John Benjamins Publishing Company

8. This is from Welinsekpo 1459.

© 2014. John Benjamins Publishing Company

© 2014. John Benjamins Publishing Company

© 2014. John Benjamins Publishing Company

A: ttak ikey nwuwu-si-n-ke-ya kuntey, salam-tul-i kulen mal isscanha,

© 2014. John Benjamins Publishing Company

(5) A:oppa ce yeyppe-ci-n ke

© 2014. John Benjamins Publishing Company

or a counterexpectation-marking function in exclamatory contexts, as well as an

13. a. ‘Na-nun pap-ul mek-ess-ta’ hako malha-yss-ta

© 2014. John Benjamins Publishing Company

appears to be a fairly robust pathway in Korean. Below we identify some parallels

Middle Korean Modern Korean Contemporary Korean

3.2 On the grammaticalization of tamye

© 2014. John Benjamins Publishing Company

© 2014. John Benjamins Publishing Company

In mirative contexts, as in (12) below, tamye often further serves a counterexpec-

© 2014. John Benjamins Publishing Company

(12) [ku-ka manhi aphu]-tamye!?

Middle Korean Modern Korean Contemporary Korean

3.3 On the grammaticalization of tamyense

© 2014. John Benjamins Publishing Company

“When I gave birth to a child, my brother-in-law let me drink a lot of

Converbal quotative hamyense, similar to hamye discussed in §3.2, also came to

Similar to tamye, in Contemporary Korean, quotative tamyense has also devel-

© 2014. John Benjamins Publishing Company

(15) a. phikonha-ci anh-a-yo?

© 2014. John Benjamins Publishing Company

→ B: ku namca-lul tasinun an-manna-n tamyense?

Figure 3 below summarizes the development of tamyense as a hearsay evidential

Middle Korean Modern Korean Contemporary Korean

3.4 On the grammaticalization of tanun

© 2014. John Benjamins Publishing Company

In the 19th century, attributive quotative VP-ta hanun underwent phonologi-

© 2014. John Benjamins Publishing Company

constructions and hence no longer needing to be referential), and as seen in (20),

22. Example (19) is from www.google.co.kr (accessed 28 July 2013).

© 2014. John Benjamins Publishing Company

Middle Korean Modern Korean Contemporary Korean

© 2014. John Benjamins Publishing Company

3.5 On the grammaticalization of tanta