Still Mirative After All These Years PDF

Still mirative after all these years
SCOTT DELANCEY
Abstract
This article re-presents the case, first presented in DeLancey (1997), for the
mirative as a crosslinguistic category, and responds to critiques of that work by
Gilbert Lazard and Nathan Hill. The nature of the mirative, a category which
marks a statement as representing information which is new or unexpected,
is exemplified with data from Kham (Tibeto-Burman) and Hare (Athabaskan).
The mirative category is shown to be distinct from the well-known mediative
or indirective evidential category. Finally, the role of mirativity in the complex
verbal systems of Tibetan languages is briefly outlined.
Keywords: evidential, Hare, indirective, inferential, inflection, information
structure, Kham, mediative, mirative, syntax, Tibetan
1. Backdrop
I cannot claim authorship of the term mirative. Neither can I claim to have
been the first to propose it as a crosslinguistic category; I believe that credit
goes to Akatsuka (1985). Fifteen years ago, in the first issue of this journal,
I was the first to link the term to the semantic category, and pointed out its
prevalence as a grammatical category. Over the intervening decade and a half,
a considerable body of research has described constructions of this kind in lan-
guages from Finno-Ugrian (Leinonen 2000) to Tsimshianic (Peterson 2010),
from the Himalayas (D. Watters 2002) to Amazonia (Queixalos 2007), and on
around the world (Dixon 2003, LaPolla 2003, Bashir 2010, Kwon 2010, forth-
coming, inter alia). The category has been recognized in general work on evi-
dentiality (Aikhenvald 2004) and in morphosyntactic surveys of various kinds
(Heine & Kuteva 2002, Creissels 2006, Melcuk 1994, Chelliah & de Reuse
2011). Whatever the theoretical status of mirativity, it is clear that a great many
linguists have found the category as I described it useful in the description and
analysis of a range of languages from around the world.
Over the same time there have been less enthusiastic reactions, express-
Linguistic Typology 16 (2012), 529564 14300532/2012/016-0529

DOI 10.1515/lingty-2012-0020 Walter de Gruyter
Brought to you by | University of Oregon Library

Authenticated | 128.223.99.134
Download Date | 2/4/13 7:52 PM
530 Scott DeLancey
ing doubts about whether such a new category was necessary either on the
grounds that my formulation was not sufficiently precise to be convincing, or
that the category I described was simply a different perspective on some other
phenomenon that was already familiar. Indeed the formulation presented in
DeLancey (1997) was vague and noncommittal on some points, and inconsis-
tent (if not incoherent) in at least one crucial respect, and my subsequent article
(DeLancey 2001) did not resolve all of these issues. Thus a response to this
line of criticism, and particularly to the very reasonable concerns expressed by
Gilbert Lazard, is long overdue.
Now in this issue of Linguistic Typology we also have a rather more ambi-
tious, if less cautious, critique from Nathan Hill which argues that everything
that I have said on the subject is simply wrong (Hill 2012b). While Lazard and
others have questioned the place of mirativity in an overall scheme of verbal
categories, Hill argues that there is no such thing in the first place. Frankly,
as far as I can see most of Hills critique has so little substantive content as to
hardly be worth replying to. Nevertheless I am happy to take this opportunity to
restate the case for the mirative in (I hope) somewhat clearer and more explicit
terms.
In Section 2 I will restate the argument for mirativity as a crosslinguistic
category; in the course of this exposition I will address Hills criticisms of sub-
stance. Section 3 will address the status of mirative as a distinct category, espe-
cially with respect to the mediative; here I will also deal with the plausibility of
Hills counterproposal. Section 4 will provide a very brief discussion of the rel-
evance of mirativity to the analysis of the complex Tibetan egophoric/evidential
systems.
2. Mirativity is real after all

The first, and most energetic, part of Hills article is devoted to arguing that the
Lhasa Tibetan form dug (h.dug) is not mirative, as I have described it, and that
since this is the cornerstone of DeLanceys analysis of mirativity, the entire
edifice of mirativity thus tumbles to the ground. It is perhaps not surprising that
Hill, a Tibetanist, takes a special interest in the Tibetan data in my article. But
to describe the interpretation of one Tibetan form as the cornerstone of my
proposal is quite a stretch, and the suggestion that the typological literature
on mirativity depends on DeLanceys description of the Tibetan morpheme
h.dug can most charitably be interpreted as reflecting a limited familiarity with
the literature in question. It is true that I first became aware of the phenomenon
of mirativity when I was trying to untangle the marking of evidentiality and
volition in the Lhasa Tibetan verb paradigm (DeLancey 2001: 371), but to be
aware of a problem is not to have solved it that was back in my postgraduate
days, and it was not until hearing Aksu-Ko & Slobins presentation a few

Still mirative after all these years 531
years later (1981) of what was published as their 1986 article that I began to
think in terms other than simple evidentiality. The foundation of my concept
of mirativity was laid the following summer when I encountered my first pure
mirative while doing fieldwork on Hare.
Hills critique of the concept of mirativity can be summarized as: (i)
DeLanceys proposal that mirative be recognized as a crosslinguistic category
is fundamentally based on his analysis of a particular morpheme, Tibetan dug,
(ii) he completely misunderstood the nature of dug, which expresses direct
sensory evidence, not mirativity, (iii) therefore everything else that he has
ever said about mirativity or evidentiality is wrong, (iv) and therefore every
other author who has identified a mirative construction in a language which
they study is also wrong. (Since these authors, then, must have failed to prop-
erly understand their own data, Hill goes on to explain what is really going on
in these languages.) Hills entire case against the mirative as a crosslinguistic
category rests on point (i), which is simply asserted, rather than argued for,
and thus literally irrefutable. But it could hardly be argued for. The case for
the mirative category in DeLancey (1997) is based on three two-term systems,
Turkish, Hare (DeLancey 1990b), and Sunwar (DeLancey 1992a); Lhasa Ti-
betan and Korean are presented as examples of more complex systems where
mirativity can be recognized as a component, not as primary evidence for the
category. Later in this article we will see that the emphasis on two-term sys-
tems in that article is at the root of the confusion of mediative and mirative
systems in my proposal. In Section 3 I will try and resolve this confusion, as
Lazard and others have demanded.
2.1. Hills critique

Hills substantial claim is that the true sense of dug is not mirative, but direct
sensory evidence, and since my interpretation of dug is the foundation of my
interpretation of everything which I call mirative, all of those constructions in
other languages must logically also express direct evidence. Then it follows
that every construction which has been called mirative must also in fact be
an exponent of the category of direct sensory evidence. The second half of
Hills article consists of discussing examples which have been presented in
the literature as mirative constructions and saying that he doesnt think they
are. Sometimes he says that all the authors examples could conceivably be
interpreted as sensory evidence, in other cases he cant say much more than
that he suspects if he knew more about the language he could show that the
author of the primary source is wrong and he is right. There is no articulated
critique of most of the work which he treats with such contempt.

532 Scott DeLancey
2.2. Meet the mirative

As Hill notes, a number of authors have described mirative constructions in a
wide range of languages, sometimes in terms obviously inspired by my work. A
typical example is Bashirs description of the Dardic language Kalasha (Bashir
2007: 3):
Past tense verb forms are obligatorily coded for the distinction between direct [. . . ]
and inferential (indirect) meaning. Direct subsumes such meanings as personally
witnessed, or having long standing in ones conceptual repertoire, while inferen-
tial includes inference, new information, and hearsay. Present-tense forms do not
have morphologically expressed inferential forms, but inferential counterparts are
supplied by the addition of huLa, the past participle of hik to become. When
huLa appears in narration of directly experienced events, the meaning is mirative,
i.e. that the speaker has just found out about (i.e. was not aware of before) the
content of the assertion.
Hill runs through an assortment of such descriptions in the literature, mostly

examples cited by me or Aikhenvald, and simply asserts, over and over, that
they are not what the authors say they are. If it is inconvenient for the point
which Hill wants to make that such constructions normally are used in both
direct and hearsay contexts, as Bashir, and I, and author after author, state, then
he simply insists that they arent, regardless of how the facts are described in
the source.
An elementary principle of linguistic description is that one analyzes ones
data directly, not by translating it into ones own or some other language and
analyzing that. So good descriptions include sufficient discussion of the ex-
amples to allow the reader to understand them properly. Hills approach is to
ignore or dismiss the authors discussion, and reanalyze the data to fit his no-
tions based on the English gloss or his own rephrasing of it, or sometimes on
his rough translation of the English gloss into Tibetan. His reinterpretations
are for the most part not convincing, or even plausible; often they are directly
contradicted by data presented in the source.
In his discussion of my, Watterss, and other authors examples of mirativity,
Hill repeatedly calls into question our interpretation of the data, and argues that
our account cannot stand without further crucial data, which we fail to provide.
In the case of my work this is because he did not read the articles where the
data which he deems crucial were published. In the case of Watterss work
the kindest interpretation of Hills treatment is to suppose that he simply quit
reading before he got to the sections which are relevant to his claims, although
he also seems to have skipped over some crucial examples and discussion in the
pages which he does directly refer to. His handling of other sources is similarly
careless (to put it as charitably as possible). In the interests of space I will leave
most of the authors whom Hill misrepresents to defend themselves, and limit

myself here to correcting Hills omissions and outright falsehoods about the
work of David Watters, and going over some of my own examples which Hill
has misunderstood or missed altogether.1
2.3. Mirativity and the mirative

As Lazard reminds us, the question of whether there is such a phenomenon
as mirativity or the mirative is in fact two separate questions: whether there is
a semantic category of new or unassimilated information, which can manifest
itself in one way or another in linguistic expression, and, if so, whether this
semantic category is one which regularly finds expression in languages as a
grammatical category. I will use the term mirativity to refer to the semantic
category in the first question, and mirative to refer to the crosslinguistic gram-
matical phenomenon addressed in the second. (And mirative construction
or form will refer to a form or construction in the particular language under
discussion.) It is the second question which has been emphasized by Lazard
and other critics, and which I must primarily deal with in this article. I find it a
bit odd to imagine serious disagreement on the first question, but if I read Hill
right he may not be convinced even of this, so let us begin at the beginning.
The notion of mirativity depends on the concept of a typical utterance,
where (as Donabdian 2001: 430431 puts it)
[t]he speaker has at his/her disposal a propositional content that already obtains
prior to the utterance [. . . ] what is being communicated is old information that has
already been integrated into the speakers background knowledge, or is stable in
the sense that its contents and truth have already been established.
Mirative marking indicates that the proposition being related is not of this kind.
Mirativity is widely expressed by lexicalized adverbials, conventionalized
constructions (English (It) turns out (that) S), intonation, sentence final eval-
uative particles, and other devices which are often not considered part of the
grammatical structure of a language. We also see it manifested in certain uses
of other grammatical constructions. Akatsuka (1985) examines an interesting
parallelism between conditional constructions in Japanese and English. In both
languages, we find a construction in which an if -clause is used to assert the
truth of a proposition, rather than to express conditional meaning. To take one
of her Japanese examples, consider the following:
1. But, since Hill takes it upon himself not only to take Tatevosov & Maisak (1999) to task for
having the temerity to actually use the term admirative in a title, but also to correct their
interpretation of their own data, let me especially recommend to the reader Tatevosov (2007)
and Maisak & Tatevosov (2007), where the authors have more space to present their analyses,
for their clear and explicit evidence and argument. (No one reading Tatevosovs paper will be
able to take seriously the idea that he is simply uncritically swallowing my proposals, as Hill
implies.)

534 Scott DeLancey
(1) konna ni yorokonde kureru no nara, motto hayaku

thus loc happy give nmz if more early
kite ager-eba yokatta
coming give-if good.pst
[Visiting his friend in the hospital, the speaker says to himself:] If
hes so happy to see me, I should have come earlier. (Akatsuka 1985:
630)
As Akatsuka notes, What the speaker connotes in using this type of antecedent
is not I know S1 , but rather I didnt know S1 until now!. I have not collected
natural examples of this in English, but my intuitive sense is that it most easily
occurs in response to new information obtained from someone else:
(2) a. [Coming in from outdoors:] The rains stopped, its beautiful out!
Lets go work in the garden!
b. Oh, if its not raining, Im gonna go for a run.
c. Since its not raining, Im gonna go for a run.
Of course, (2b) would be perfectly possible without the context of (2a), but
then it would express the speakers lack of knowledge about the current state
of the weather. In the context of (2a), it is consistent only with a reading in
which the speaker believes that it is not raining, and has only just learned that.
Contrast it with (2c), which has no mirative force and can be used whether the
proposition is old or new information.
Mirativity, then, is a robust and familiar phenomenon. The question for ty-
pologists is, to what extent do we find this semantic category expressed in
languages by dedicated grammatical constructions? As we have already noted,
mirativity (whether or not by that name) shows up in many grammatical de-
scriptions, such as Bashir (2010: 3):
In Kalasha and Khowar, the forms called inferential in Bashir (1988) report non-
witnessed events or actions learned of by hearsay, inference from observation of
resultant states, or narrated in traditional tales; however, these forms also encode
meanings which are uniquely mirative.
This point is illustrated by (3), a Khowar sentence which [Bashir has] had ad-
dressed to [her] as a foreigner not expected to know Khowar, on a first meeting
with someone (Bashir 2010: 3).
(3) tu khowr kor-k bir-u

you Khowar do-ag.n become(pst.i)-2sg
(Oh), you speak Khowar. (registering surprise)
She then goes on to characterize (3) as

[. . . ] a surprised reaction to an immediate, first-hand observation of an unexpected

ongoing action, and clearly not reporting hearsay, inference from a resultant state,
or non-witnessed action. Nor can its primary purpose be to inform the addressee
(me, in this case) of a fact, since I can be presumed to already know that I speak
Khowar. Its function is to communicate the fact that the speaker has learned some-
thing new and is surprised by it. The mirative meaning of this utterance is clear
and distinct, regardless of the fact that the Khowar inferential form is also used
for other indirective/mediative meanings.
But Hill is ready to tell us that Bashir is wrong about Kalasha, just as everyone
else who has described a mirative construction is wrong about their data. In
the next two sections I will go over some examples where I am in a position to
demonstrate just who is wrong about what.
2.4. The mirative in Kham

D. Watters (2002) provides a detailed description of a mirative construction in
Kham, a Tibeto-Burman language of the Central Himalayan group, which has
been used as an example by Aikhenvald (2004: 211213), and thus attracted
Hills attention. Hills discussion of the Kham data is a good example of his
style of argument and his attitude toward evidence. Watters (2002: 288296)
provides a thorough discussion of the various uses of the mirative, expressed
by a nominalized verb form in construction with o-le-o, a formally nominalized
but effectively finite2 copula inflected for 3rd person singular (3sg-cop-nmz).
In examples I will simply gloss oleo as mir (for mirative), and ignore its internal
structure:
(4) a. ba-duh-ke-r@
go-prior-pfv-3
They already went.
b. ya-ba-duh-wo oleo
3pl-go-prior-nmz mir
They already went! (unanticipated information)
This construction can be used when the information being related is perceived
at first hand, as in (5), said when the speaker had just seen a leopard which he
and the addressee were looking for:
(5) n@-k@ z@ ci sy:-d@ u-li-zya-o oleo sani
there-at emph cep sleep-nf 3-be-cont-nmz mir conf
Hes right there sleeping, see!
2. As in many Tibeto-Burman languages, several morphologically nominalized constructions in

Kham function as finite.

536 Scott DeLancey
It can also be used with statements based on inference, as in (6), which precedes
(5) in the same text, said when the speaker first discovered traces showing that
the leopard had eaten his dog:
(6) a-k@ z@ o-k@i-wo oleo

here-at emph 3sg-eat-nmz mir
He ate [him] right here!
Note that it is not the case that oleo here expresses the inferential category
per se. It is clear from the discourse context that the statement was based on
inference from secondary evidence; the contribution of the mirative here is
to express the speakers attitude (at the moment being described) toward the
information.
2.4.1. Could oleo mark direct evidence? Hill asserts that the oleo construc-
tion cannot be considered a mirative because Watters does not present exam-
ples inexplicable in terms of sensory evidence. He discusses example (5), but
not (6), but I take it from his treatment of other data that he would still consider
this sensory evidence, because, after all, the speaker did see something. But
in a true evidential language, these two statements, in the context in which they
were made, could not be in the same grammatical form. Since the speaker is
a direct witness to the proposition he states in (5), and is explicitly not in (6),
(5) would be in the unmarked or the direct evidential form, and (6) marked as
inferential. (As we will see in Section 4, this is the case in Tibetan.)
In any case, Watters presents many examples inexplicable in terms of sen-
sory evidence. The second and third sentences from the text from which (56)
are taken are (Watters 2002: 418):
(7) nahm-ni ge-hu-zya-k@ te, kh:bya ekh@

lowland-from 1pl-come-cont-when foc pillar mountain
ya-do-zya-o-k@ ge-basi-k@ te ri:-l@ te
3pl-say-cont-nmz-loc 1pl-stay-when foc night-inessive foc
ge-ka:h la:-e b@i-d@ o-ya-si-u oleo
1pl-dog leopard-erg take-nf 3sg-give-1pl-nmz mir
As we were coming from the lowlands, at a place called Pillar Moun-
tain, our dog was taken away on us by a leopard!
(8) ge: te ch@kalnya te k:-r@ zya-d@ ka:h-r@ k:
we foc morning foc meal-pl eat-nf dog-pl meal
ya-nya le-d@ ge-ra-kih-k@ te ma-l-e
give-inf say-nf 1pl-3pl-call-when foc neg-be-impf
We, in the morning, having eaten our food, and thinking to give food
to the dogs, when we called them they werent there.

The mirative oleo at the end of (7) tells us that the loss of the dog was an
unanticipated event, not that there was or wasnt some kind of evidence for
it. The actual evidence is simply the absence of evidence of dogs, as stated in
the next sentence: what sparked the realization of the mirative-marked event
is the failure of the dogs to appear for breakfast not direct perception of
anything, but, precisely, the absence of anything to perceive. (In any case, even
if one wanted to insist that (8) is a report of a dogless situation known by direct
evidence, note that it is not that sentence which is marked as mirative, but the
previous one, describing an event which the speaker at the time had no actual
knowledge of at all, direct or otherwise.)
2.4.2. The language of dreams. Hill makes much of the use of mirative
constructions in reporting dreams, and seems to consider it evidence that a
construction is mirative (even though there is no such thing), to the point of
including it in his checklist of typical features of mirative constructions (or,
maybe, what would be typical features if mirative constructions existed). On
his checklist he lists Kham with a question mark in the Dreams column, but
Watters does explicitly discuss the use of the mirative in relating a dream. He
presents (9) from the narration of a dream in which the speaker, Jaman Sing,
is captured by a group of women and stuffed into a burlap bag. Later, he makes
a startling discovery:
(9) ao-r@ te zha:h-r@ ci oleo, h@i li-d@ te, Na: te

this-pl foc witch-pl cep mir thus say-nf foc I foc
b@n@iz@ Na-che:-ke
very 1sg-frighten-pfv
These are witches! saying to myself, I was greatly frightened. (D.
Watters 2002: 292; I have added morpheme glosses from 2002: 428)
Watters (2002: 292) then goes on:
Jaman Sings words/thoughts, though reported to no-one at the time, are marked
by the mirative. He conveys to his audience now, perhaps weeks after the event,
that his discovery of his captors identity was new, unassimilated knowledge at the
time.
In context (see Watters 2002: 428) this realization does not appear to be the
result of direct perception, even within the dream, but of suddenly realizing the
significance of the previous course of events. Another example from the same
text is even clearer. After the narrator wakes from his dream, he realizes its
meaning:

538 Scott DeLancey
(10) na:h-kin te ph@ri, ao te p@rmesor-e o-cak@rya

that.much-after foc again this foc God-gen 3sg-devotee
Na:-s@ z@ u-li-zya-o oleo h@i li-d@ b@n@i-d@
I-com emph 3sg-be-cont-nmz mir thus say-nf make-nf
h@i b@ Na-li-ke
thus also 1sg-say-pfv
And again after that, This is Gods angel that is with me!, I began
to say. (D. Watters 2002: 431)
This is the speakers conclusion as he considers the meaning of his dream, not
something that he has seen or otherwise directly perceived, either in the dream
or out of it. The import of the mirative here is that at this point in the narrative
the speaker, who has been sick in bed, suddenly realizes that he has the internal
resources to fight his illness, and from that point begins to recover his health.
2.4.3. Hearsay. Hill further asserts that:

Watters (2002: 296) claims that the mirative is also used when the source of knowl-
edge is hearsay; he does not however provide examples of such a usage and the
language does have a separate morpheme for reported speech.
Hill seems to feel entitled to explicitly doubt a direct statement about a lan-
guage by someone who actually knows something about it (Watters was in fact
a fluent speaker of Kham). It is hard not to read this as a direct assertion that
Watters is wrong about, if not actually misrepresenting, his data. The first and
last clauses of Hills sentence are true; the middle one is an outright falsehood.
It seems extraordinarily careless even for such an enthusiast to have simply
missed the section on Third person narratives (D. Watters 2002: 295296,
also referred to by Aikhenvald), where Watters explains that, while usually
in 3rd person narratives which the narrator is not reporting as an eyewitness,
events are marked with the reported speech particle (RSP) di (D. Watters 2002:
295, emphasis added):
An exception to RSP marking in third person narratives occurs with recent events.
interestingly the report of such events is marked with the mirative
oleo. This occurrence accords with Aksu-Ko and Slobins account of a similar
phenomenon in Turkish there, over time, as the once-new information becomes
a part of general world knowledge the mirative is dropped.
He then presents a 13-line text illustrating the point, where the speaker is relat-
ing events which had happened to someone else, and ends each sentence with
the mirative oleo. So, not only does the language use the mirative construction
to mark a statement as based on hearsay, it does so in a very systematic way,
which is described and carefully exemplified in the very grammar which Hill
insists provides no such examples.

2.5. The Athabaskan mirative

If there is a touchstone to my conception of the mirative, it is the mirative
particle l in the Athabaskan language Hare (Fort Good Hope Dene), which
I encountered during fieldwork in 1981. Hill has many complaints about the
Hare data and the argument from it presented in DeLancey (1997). There is
some validity to some of his questions; if I had anticipated his particular line
of disagreement, I could have included data which might have preemptively
addressed some of his concerns, although the relevant examples and discus-
sion are all available in DeLancey (1990b), which Hill does not seem to have
consulted. Subsequent work has shown that similar categories are to be found
in many Athabaskan languages (de Haan 2008), see de Reuse (2003) for the
strikingly similar use in Western Apache of a form cognate to Hare l.
I discovered the Hare particle l while looking for evidential constructions,
and at first fastened on the inferential sense which I was able to elicit for con-
structed examples like (11) as the essential meaning of the form:
(11) jhye sa knayeda l
hereabout bear walk.around.impf mir
There was a bear walking around here! (inferred from discovery of
bear tracks)
But a significant problem for this analysis was that I was unable to elicit ex-
amples like this simply by constructing an inferential context sentence (11)
without the mirative l is perfectly possible regardless of the nature or source
of evidence. So the contribution of l to the sentence must be something else.
When I discussed the question with my consultants, they were quite clear
about what the something else was. As one speaker put it, l is there because
you didnt know (DeLancey 2001: 377).3 In trying to explain to me how to
use l, one speaker suggested this example:
(12) John deshta dya l
John bush sg.go.3sg.su.pfv mir
John went to the bush.
And he explained it thus: You go to Johns house and see hes gone. You ask
where he went, and they tell you he went to the bush. Then you can say it.
On further discussion the speaker agreed that the sentence could also be used if
the speaker simply notices that Johns snowmobile and trapping equipment are
gone and infers his destination, since he would not have taken his equipment
unless he were going out to the bush to set traps. I expect that for Hill, seeing
3. Nowhere in his article does Hill address explicit statements from native speakers cited by me
and other authors about how they understand the forms and examples in question.

540 Scott DeLancey
that John is gone is visual evidence for (12), but this reduces the notion of direct
evidence to vacuity. Seeing that someone is not present is visual evidence that
he is not present; it is not visual evidence of where he is instead. But it is
unnecessary to debate the point; the first context my consultant gave me, the
one which seemed the most obvious and natural to him, is the one where the
only way I know where John went is that someone who knew the information
told me.
Since Hill wants to insist that l, like all putative miratives, really expresses
direct sensory evidence, it is vital to his argument that it not be usable in a
context where the speakers only evidence for the proposition is hearsay. In
DeLancey (1997) I was concerned with exemplifying the mirative force of the
form, but its use in hearsay contexts is described elsewhere. Consider (13):
(13) John deshta raweya l

John bush sg.go.and.return.3sg.su.pfv mir
John made a trip to the bush. (DeLancey 1990b: 154)
Of which the author states that it could be used if I was not aware that John
was gone, or knew he was gone but didnt know where, but I see him returning
with a load of meat, or if he told me after his return where he had
been (DeLancey 1990b: 154, emphasis added).
I suppose Hill may still want to say that I am just wrong about this, but the
fact is, this sentence, and many others with l, can indeed be used in a con-
text where the information was obtained from someone else, with no sensory
perception on the part of the speaker except hearing the words of his source of
information.
In DeLancey (1997, 2001) I portrayed inference and hearsay as functions
of the mirative construction, which was obviously problematic. I was taken by
the parallels between the Hare construction and the Turkish mediative, as de-
scribed by Aksu-Ko & Slobin, and ended up, as Lazard (1999) was quick to
note, defining the mirative as something very much like the classic mediative.4
Assuming my argument (DeLancey 2001, Aikhenvald 2004) that mirativity is
distinct from evidentiality, to say that the mirative codes either indirect evi-
dence or unanticipated information is contradictory. The correct significance
of the fact that mirative constructions can occur in both direct and indirect evi-
dential contexts is precisely that it proves that they are not evidentials direct
vs. indirect evidence is the fundamental evidential distinction, so a construc-
tion which simply ignores that distinction is not an evidential. The essential
4. I use mediative to refer to the category of indirect evidence which subsumes both inference
and hearsay; this category, also referred to as indirective, has been the subject of an exten-
sive literature (e.g., Lazard 1996, 1999, 2000; Johanson 2000, 2003).

meaning of the mirative is mirativity, regardless of source of information. We

will see further evidence in the next two sections of its essential independence
from the direct/indirect opposition.
2.6. Extended and rhetorical uses of the mirative

Given the difficulty which Hill has with the straightforward uses of the mira-
tive, it is perhaps to be expected that he is confused by what we might call
non-literal uses of mirative or evidential constructions. Perhaps the clue to
his perplexity is seen in his questioning whether the Hare mirative would be
grammatical in one set of circumstances or another. This betrays a funda-
mental misunderstanding of the nature of categories like the mirative, which
express the speakers attitude toward the proposition, not some objectively per-
ceivable aspect of it. When we say that construction X means M, we are not
saying the speakers automatically react to M with X, like Skinnerian pigeons,
or even that M must somehow be objectively true of the situation depicted in
the utterance including X. Rather, what it means is that the speaker is depict-
ing the situation as being characterized by M. When an English speaker uses
the present perfect, it is not because the statement has present relevance, in
some objectively determinable sense, but that the speaker has chosen to present
it in that light. All such categories including, notoriously, both mirative and
mediative constructions are commonly seen in contexts where mirativity or
mediativity are invoked for rhetorical or other purposes, from sarcasm to praise
and beyond. This is a topic deserving of extended treatment (for example, see
Queixalos 2007 for an extended catalogue of uses of the mirative in Sikuani);
here I will only touch on points raised by Hill.
2.6.1. Non-literal mirativity. Hill expresses some skepticism about the use
of mirative forms to convey compliments. Of the Hare sentence (14), I state
explicitly: Something like this might be said to someone who has just demon-
strated more wilderness knowledge than the speaker thought he had
(DeLancey 1997: 40, emphasis added).
(14) deshta yedanyie l

bush be.smart/2sg.su/impf mir
Youre smart for the bush! (i.e., are competent at bushcraft and good
at dealing improvisatorily with situations that come up in the bush)
Hill engages in some speculation about when people might say this, but un-
necessarily it should not be that hard to conclude from my description that
the sentence would not be so polite in a context where the speaker ought to
have known already, and indeed it would not. I cant exactly perform Hills
experiment, nor do I claim to be an expert in Dene etiquette, but I can say for

542 Scott DeLancey
certain that this sentence would be very complimentary coming from an older,
experienced man to a younger person, and bizarre, if not downright insulting,
in the other direction. As I have suggested above, Hill misses the point in spec-
ulating whether l would be grammatical in the context where the speaker is
commenting to a world-renowned musician on his skill. Of course it would be
grammatical, in some narrow sense, but it would be socially quite inappro-
priate.
Maslova (2003) provides an excellent example of the mirative use of the
inferential in Yukaghir which nicely illustrates the point. See (15):
(15) a. qalite o:-lel-dek
best.hunter cop-infer-intr.2sg
You proved to be a real hunter!
b. qalite o:-dek
best.hunter cop-intr.2sg
Youre a real hunter!
Maslova (2003: 228229) discusses this example:
It seems that a morphological distinction is drawn between properties that are
displayed and/or acknowledged by the speaker for the first time (Inferential) and
properties that have been established earlier and are therefore known to be present
(Direct). For example, in a narrative about the speakers very first hunting expe-
rience, whereby he was supervised by his elder brother, the brother makes two
encouraging statements, first [(15a)] (right after the hunting was over) and then
[(15b)]. [. . . ] The Inferential marking in [(15a)] indicates that the elder brother
has inferred, on the basis of the boys behavior, that he has a set of qualities re-
quired of qalite (the best hunter of a tribe). By the time of the second utterance,
this fact has already been established, and the Direct form is appropriate.
Another Hare example which should help lay to rest the idea of a sensory
evidential interpretation of l is (16):
(16) gsh yedarehyie l
really be.smart.1sg.su.impf mir
Im really smart! (DeLancey 1990b)
This is something you would say when you have surprised yourself with how
well and quickly you accomplished a formidable task. It doesnt matter if the
task is one that results in sensory evidence, the point is simply that you didnt
think you could do it, or at least not do it so skillfully, but found that you could.
2.6.2. Mirativity in narrative. Another category of example of the mirative

in Watters grammar of Kham that seems difficult to analyze in terms of sen-
sory evidence illustrates what he calls the hindsight use (D. Watters 2002:

294), where the mirative marks a proposition in a narrative which was not
known to the narrator at that time. Here the mirative serves to mark the in-
formation as new to the speaker now, from the perspective of the speaker as a
participant in the narrative who was at the time unaware of it. For example, in
a narrative recounting a visit from his attorney when he was in jail, a speaker
says:
(17) Kathmandu-ni ji pi adikhari sahibe m@ni ge-lai ciu:na

Kathmandu-from G. P. Adikhari Sahib also we-obj visit
hu-ke. ho-e ciu:na u-hu-k@, h@ldar seros@
come-pfv he-erg visit 3sg.su-come-loc Haldar old.man
ni-d@isi-u oleo ge: te ho: ge-ma-s@y-e.
3du.su-visit-1.obj mir we foc that 1pl-neg-know-impf
From Kathmandu, G. P. Adikhari Sahib also came to visit us. When
he came to visit, he had (already) met with Old Man Sergeant. But we
didnt know that.
After the time of the visit, described in this sentence, the speaker learned of
the meeting between Adikhari Sahib, his attorney, and Old Man Sergeant. But
at the time of the visit, he (the speaker) was not aware of this prior meeting.
The use of the mirative here conveys that the information he had met with
Old Man Sergeant, when he eventually learned of it, was new information,
unknown to him at this point in the narrative. There can be no question of
sensory evidence involved in any way in the use of the mirative construction
here. (The same phenomenon is described for Western Apache by de Reuse
(2003: 83).)
3. Mirativity, evidentiality, mediativity, or other?

The rest of Hills argument against the mirative literature is of a piece with his
discussion of Hare and Kham, simply taking a selected set of examples, and
imagining or inventing some way that each might be interpreted as having to
do with the speaker directly perceiving something or other. Once we have taken
a look at one or two examples of his work, there is hardly any need to deal with
it further. The Kham, Hare, and other mirative constructions described in the
literature do not express direct sensory evidence, and Hill gives us no reason
whatever to think otherwise.
A more serious question is whether mirative constructions like these can be
reinterpreted as variations on the mediative category. Lazard has complained,
quite legitimately, that my earlier papers are vague as to whether I am suggest-
ing mirativity as an alternative analysis to mediativity, or as a component of it,
or as an independent category which is also a component of mediativity. Obvi-
ously the vagueness of the presentation reflected a vagueness of conception, but

544 Scott DeLancey
the substantial discussion of these questions, and more importantly, the flood of
literature on evidential and mirative systems of the last fifteen years has done a
great deal to clarify the issues.
Since Hill effectively adduces no empirical support for his argument, it
should hardly be necessary to refute it further. But there is an obvious theo-
retical refutation, the reason why Hills proposal is not only false in fact but
impossible in principle, which will serve as the basis for my explanation for
the similarities between mediative/indirective and two-term mirative systems.
3.1. There can be no direct evidential without a contrasting indirect

If a language has a construction which expresses inferential, indirective/media-
tive, mirative, or whatever, there is necessarily another contrasting construction
which does not. Following Aikhenvald (2004), we can distinguish two types of
two-term system. In Aikhenvalds type A1, the unmarked term explicitly has
the opposite value from the marked term i.e., if the marked term expresses
indirect evidence, the other explicitly expresses the category of direct evidence.
In type A2, the unmarked term is not specified for evidential value, and can in
principle be used to report information from any direct or indirect source. The
latter is the case for the classic mediative systems, where the non-mediative
form is not an evidential, that is, it does not explicitly mean direct evidence.
It will be important in the next section to note that mediative oppositions, as
well as the mirative constructions of Hare, Sunwar, Kham, Kalasha, and other
well-known examples, are also type A2.
In type A1 systems, as in Jarawara (Dixon 2003), Chechen (Molochieva
2010, see below), and elsewhere (de Haan 2001, Aikhenvald 2004), one term
expresses indirect knowledge (either inference plus hearsay, or inference,
hearsay, and some categories of non-visual evidence, see de Haan 2001), and
the other explicitly asserts that the statement is made on the basis of direct ev-
idence. Obviously, in such a language the existence of a direct evidential form
depends on the contrast with an indirect form, and the two forms divide up
the semantic space between them. Of course we also find direct evidentials in
more complex evidential systems, as we will see, but again the direct eviden-
tial finds its meaning from its place in semantic space, hemmed in on all sides
by forms which occupy neighboring semantic real estate. We will see a rather
straightforward example and another which is far from straightforward below.
What we do not, and could not, find is a system where the single marked
category in an A2 system expresses direct evidence and nothing more. A form
which explicitly indicates direct evidence can only exist in opposition to one
or more which expresses indirect evidence. One cannot coherently describe
a hypothetical language in which there was a marked construction indicating
direct sensory evidence contrasting only with an unmarked construction with

no evidential value at all. Hill is asserting that languages such as Kham, Hare,
Sunwar, and Khowar are exactly that. But what would determine when such a
construction would be used? There is no such thing as a direct evidential that
does not contrast with one or more indirect evidentials:
[L]anguages tend to grammaticalize indirect evidentials before they grammatical-
ize direct evidentials. Languages in general develop grammaticalized quotatives
and inferentials before evidentials that mark direct, sensory, evidence. Although
there are minor exceptions to this generalization, it appears to be fairly robust
crosslinguistically, based on the data reported in de Haan (2005). The explanation
for this grammaticalization pattern is related to markedness: when someone is
speaking it will normally be assumed the speaker has first hand evidentiary infor-
mation, unless explicitly stated otherwise. Thus, indirect evidentiality is marked
with respect to direct evidentiality and according to markedness theory will have
to be marked overtly. (de Haan 2008: 6970)
We will see further implications of this when we get to the discussion of Ti-
betan in Section 4.
I have shown above that as a matter of empirical fact the Kham and Hare
mirative categories do express mirativity, not direct evidence. What we have
seen in this section is that, since in both cases the construction in question is
the marked member of a type A2 two-term system, there is no possible way
that their meaning could be direct evidence. In Section 4 we will apply this
argument to the problem of Tibetan evidentiality. In the next two parts of this
section we will look at the relationship between mirative oppositions like this
and the mediative.
3.2. The mediative

With Lazard (1999, cf. Aikhenvald 2004: 15) we must distinguish the classic
mediative from the more expansive notion of evidentiality which we need to
describe more complex systems. The mediative or indirective is consistently
described as ordinarily used in contexts of inference or hearsay, sometimes
also to report directly perceived new information. The mirative constructions
of Kham, Hare, Kalasha, and many other languages are described as primarily
expressing the category of new information, often used in contexts of infer-
ence or hearsay. The obvious question is, are these descriptions of two distinct
categories, or simply different perspectives on the same basic phenomenon?
Plungian (2010: 47) presents the problem as a challenge:
[M]irativity is not used to indicate the source information of a situation. Further-
more, a mirative situation is, as a rule, accessed by means of a direct observation
by the speakers. However, in spite of this fact, mirativity may in the majority of
languages be expressed by means of markers of indirect access, which constitutes
an important descriptive and theoretical problem. Linguists who defend a rigid

546 Scott DeLancey
separation of mirative and evidential values, due to the legitimate semantic differ-
ences between both, cannot explain why these values are regularly expressed by
means of one and the same marker.
Let us see whether we can or not.

A first approach is to note that the descriptions of these systems in the litera-
ture cluster around these two poles, and infer from the considerable consistency
of the descriptions of mediative/indirective systems, and more recently of mi-
rative constructions, that there are two distinct phenomena being described.
That is, to say that these have been described differently by speakers of the
languages, as well as linguists reporting on them because they are different.
Mediative constructions express the category of indirectivity, and to use a
mediative to report direct evidence is to place the statement in the same cate-
gory as statements based on indirect evidence (Lazard 1996, 1999; Johanson
2000; Friedman 2007). A mirative construction expresses the category of new
information, regardless of how it is acquired. Thus, although a list of the evi-
dential circumstances in which a Turkic mediative and a Kham mirative might
be uttered would look similar, their actual distribution in text and speech might
be quite different. And although they could quite possibly be uttered in exactly
the same objective circumstances, they would not be saying the same thing
about it.
Zeisler (2004: 659) describes the mediative as
a much stronger claim about the status of information [than DeLanceys mira-
tive] [. . . ] comprising not merely the relative novelty of immediate perceptions
meeting an unprepared mind, but also, independent of the means of knowledge,
the speakers devaluation of the information as unbelievable. The speaker thus
keeps distance to second-hand information as well as to his or her own observa-
tions and signals that, in contrast to normal observations, s/he would not be willing
to act as a witness under oath.
This is quite different from the mirative in Hare or Kham, which does not
devalue the information, but presents it as fact, often as important fact.
Still, in our current state of knowledge, it is certainly not inconceivable that
the differences in the descriptions of mediative and mirative systems are some-
how artifactual, reflecting different presumptions or methodological choices on
the part of linguists. For example, Donabdian (2001: 432), trying to reconcile
both categorial profiles under a single definition, suggests that the strong as-
sociation in the literature between the mediative category and the particular
values of hearsay and inference may be an artifact of the elicitation process.
A speaker asked to evaluate an example sentence without context must, con-
sciously or otherwise, imagine a context, and the indirect contexts are the
simplest and thus the easiest to imagine. Other values require a more elabo-

rate scenario and are therefore less likely to occur when native speakers reflect
on their language out of context.
3.3. Mirative and indirect evidence together

But there is a stronger argument for the distinctness of the mirative category,
which cannot be interpreted as simply a difference among linguists, rather than
among languages the existence of languages where the expression of mirativ-
ity is not connected with indirect evidence. Let us hypothesize that the reason
for the superficial resemblance between the mediative and the mirative is that
the constructions being compared are always the marked member of a two-
term type A2 opposition, contrasted with an unmarked, non-evidential form.
The apparent similarities then might be inherent in an A2 system. The marked
category will have some primary value: inferential, indirect, or mirative. But
if there is only one marked term, it is likely to have a broad range of use, giv-
ing us the famed mediative constellation primarily formed around inferential
reanalysis of perfects, see DeLancey (1982), Anderson (1986), Lazard (1996),
Donabdian (1996), Michailovsky (1996), Lindstedt (2000), Tatevosov (2001),
inter alia and the indirect evidential uses of true miratives like Hare l. This
is only a slight extension, since information acquired through indirect means
often is mirative in nature otherwise one would already know it, and not have
to rely on inference.
Thus a language which has a mirative construction as well as an indirect
evidential would constitute direct evidence that the two categories are dis-
tinct, especially if the two categories can co-occur. Several such languages
have been reported. Grunow-Hrsta (2007) describes such a system in Magar,
a Tibeto-Burman language of Nepal, very closely related to Kham (Grunow-
Hrsta 2007: 175):
A non-mirative statement simply conveys information, making no claims as to its
novelty or the speakers psychological reaction to it. A mirative statement conveys
that the information is new and unexpected and is as much about this surprising
newness as it is about the information itself.
The mirative construction is very similar to that of Kham, a nominalized clause

in construction with the copula le:
(18) Magar (Tibeto-Burman)
a. thapa i-laN le
Thapa dem-loc cop
Thapa is here.
b. thapa i-laN le-o le
Thapa dem-loc cop-nmz cop(=mir)
[I realize to my surprise that] Thapa is here!

548 Scott DeLancey
This contrasts directly with distinct inferential and hearsay forms (Grunow-
Hrsta 2007: 186; note that the non-mirative forms have a different nominal-
izer):
(19) a. kumari bHim-o im-aN mu-o le

Kumari BHim-gen house-loc sit-nmz mir
(I realized to my surprise that) Kumari lives at BHims house.
b. kumari bHim-o im-aN mu-m2 le-sa
Kumari BHim-gen house-loc sit-nmz impf-infer
Apparently, Kumari lives at BHims house.
c. kumari bHim-o im-aN mu-m2 le ta
Kumari BHim-gen house-loc sit-nmz impf hearsay
They say that Kumari lives at BHims house.
And the mirative combines freely with the indirect evidentials:
(20) hose-ko-ko das-o le-sa

dem-hon-pl leave-nmz mir-infer
(I realize to my surprise that) Apparently, they left.
Thus in Magar it is impossible to consider mirativity to be simply some

kind of side-effect of indirectivity or mediativity, since a mirative construc-
tion can be formally direct or indirect, and an indirect construction mirative or
not.
Another example of the same kind is Chechen, as described in Molochieva
(2010). Chechen has distinct and independent mirative and indirect construc-
tions, which can even co-occur. This gives us four possibilities (examples
and interpretations from Molochieva 2010: 218222, 248250): (i) unmarked
(21a); (ii) marked for mirative, by the suffix -q on the main verb (21b); (iii)
marked for indirect, by the auxiliary xilla (21c); and (iv) marked for both
(21d):
(21) a. Zaara j-ea-na vaiga

Zara.nom(j) j-come.pfv-prf 1pl.inc.all
Zara has come (I expected her to come).
b. Zaara j-ea-na-q vaiga.
Zara.nom(j) j-come.pfv-prf-mir 1pl.inc.all
(Wow!) Zara has come! (I didnt expect her to come).
c. Zaara j-ea-na xilla
Zara.nom(j) j-come.pfv-cvb.ant be.prf
Zara has come (I can see her shoes in the hall, but I didnt see
her come). (Molochieva 2010: 219)

d. Zaara j-ie-na xilla-q

Zara.nom(j) j-come.pfv-cvb.ant be.prf-mir
Look! Zara was here! (I can see her special cookies in the
kitchen (unwitnessed coming; unexpected/new situation; not con-
current with my expectations)).
Chechen has a type A1 evidential system: a statement which is not marked
with xilla is necessarily interpreted as based on direct perception (Molochieva
2010: 216). Thus a sentence like (21b) cannot be inferential, but must express
direct evidence, but this is has nothing to do with the presence or absence of the
mirative suffix. The mirative is equally possible with an inferential predication
such as (21d).
The fact that Chechen has distinct mirative and indirect constructions is suf-
ficient to show that these are distinct categories. The fact that they are expressed
in independent constructions, which can co-occur, shows that they are not even
the same kind of category, i.e., that mirativity and evidentiality are distinct sys-
tems, not exponents of a single larger category. We will see further examples
of the interaction of distinct evidential and mirative distinctions in Tibetic lan-
guages in Section 4.3.
4. Tibetan
4.1. Evidential and other categories in the Tibetic verb
The Tibetic languages have attracted a great deal of attention for their com-
plex and unique evidential/egophoric systems.5 Grammaticalized mirativity is
widely attested across the geographical and genetic range of Tibeto-Burman
(D. Watters 2002, LaPolla 2003, Grunow-Hrsta 2007, Shirai 2007, Willis
2007, Lidz 2007, Noonan 2008, Andvik 2010, Hyslop 2011a, b), although there
are also languages (e.g., Burmese) and branches (e.g., Bodo-Garo) where we
find no evidential or mirative marking at all, and no case has been made for
either as a deeply-rooted characteristic of the family. But in the Tibetic lan-
guages, and a few of their near neighbors, we find evidential and mirative el-
ements combining with aspect, person, and volitionality in very unusual and
complex systems of evidentiality that are without direct analogues in other ar-
eas (Plungian 2010: 1920).
I cannot attempt a complete explanation of the Tibetan verbal system (for
an introduction and overview see Zeisler 2004). There has been a great deal of
5. There is abundant cultural and linguistic evidence for extensive and early contact between
Tibetan and Altaic speakers (Roerich 1930, Eberhard 1942, Beckwith 2009), so it is conceiv-
able that Himalayan evidentiality originates in mediative constructions ultimately borrowed
from Turkic, the apparent epicenter of the Great Evidential Belt (Johanson 2003). (See now
Binnick 2012 for evidence of both mediative and mirative constructions in Mongol.)

550 Scott DeLancey
work on Tibetic languages, and especially on evidential and related categories,

over the last twenty years (e.g., Hung 1991; Sun 1993; Tournadre 1996, 2001;
Saxena 1997; Denwood 1999; Bielmeier 2000; Volkart 2000; Haller 2000a;
Hsler 2001; Hein 2001, 2007; Garrett 2001; Zeisler 2000, 2004; Huber 2000,
2005; Chirkova 2008), but still only a small portion of the Tibetic languages
have been described at all. But we have ample data to show that there is, in
a sense, a modern Tibetan verb system, that is, there is a degree of con-
sistency across many of the modern languages in the gross categories which
are distinguished (see Tournadre 1996). (The phrase Tibetan verb sometimes
refers specifically to the Classical Tibetan system, which is completely differ-
ent from any of the modern languages.) But the system also shows consider-
able variation across the branch (Sun 1993, Bielmeier 2000), and Balti, the
westernmost Western Tibetic language, apparently does not participate at all in
the evidential-egophoric complex which characterizes most of the Tibetic lan-
guages: the distinction between old and new knowledge does not (yet) play a
role in the epistemic pattern of the Balti auxiliaries (Bielmeier 2000: 86).
4.2. The basic pan-Tibetic system

The analysis of Tibetic verbal conjunct/disjunct or egophoric systems re-
quires reference to aspect, evidentiality and mirativity, volitionality, and per-
son. The literature on Tibetan evidentiality is much concerned with the in-
teractions among these (Jn 1979, 1983; DeLancey 1990a; Hsler 2001; Hein
2001; Bickel 2008; Tournadre 2008). The system is more complex in some
languages than others, but there is a fundamental evidential paradigm distin-
guishing three categories. Almost every author who has discussed the system
has a different set of terms; here I will refer to them as generic, personal, and
immediate knowledge. The categories can be illustrated by the now hackneyed
Lhasa sentences (22):6
(22) a. bod-la g.yag yog-red

Tibet-loc yak exist-generic
b. bod-la g.yag yod
Tibet-loc yak exist-personal
c. bod-la g.yag dug
Tibet-loc yak exist-immediate
There are yaks in Tibet.
6. All Written and Central Tibetan examples are presented in a transliteration of standard orthog-
raphy. This is essentially the Wylie (1959) system with the addition of hyphens and equals
signs to represent two phonologically different types of morpheme boundary. (Transliteration
seems to be another topic on which Hill has energetic opinions; see Hill 2012a.)

Hill seems to miss the point of these examples; undoubtedly my earlier discus-
sion of them could have been more precise. In the real world (22b) and (22c)
are fairly unlikely sentences since anyone who knows what a yak is knows
that they are associated with Tibet, the ordinary way for anyone to state that fact
would be with the generic form (22a), which expresses knowledge for which
no evidential source needs be given, as because it is common knowledge.
If the speaker has such warrant for a statement, that is all that is necessary,
and the generic is the appropriate form. The use of either the personal or the
direct form implies that the statement is not being presented as generic knowl-
edge. So (22b) explicitly emphasizes the personal basis for the statement, that
the speaker is making the statement on the basis of personal knowledge rather
than common knowledge. The spirit is better captured in English by some-
thing like In my country, Tibet, we have yaks. The point of this example is
not, as many readers reasonably inferred from my earlier description, that this
is what any Tibetan would be expected to say, but rather that it is something
that only a Tibetan could ever say.
But the main point of these data for the problem of mirativity, which Hill ig-
nores, is stated quite plainly: that no Tibetan could ever say (22c). This means,
quite simply, that however useful direct evidence may be as a description of
this category, it is not the case that what is being expressed here is simply that
the speaker has direct perceptual evidence for the statement. If that were the
case, anyone who has seen a yak in Tibet which would include a great many
Tibetans could, and, one would expect, normally would, use this construc-
tion to report this fact. The reason this is impossible is because any Tibetan
has better basis than that for the statement. If one considers a fact to be generic
knowledge, then that is the strongest basis one can have for the statement, and
one will report the fact using that form. Failing that, personal knowledge is
the next strongest warrant. The essential fact about the direct or immediate
form is that it can be used only when neither of the stronger bases is avail-
able:
The use of this what may be termed the immediate evidential indicates that the
speakers basis for his assertion comes solely from perceptible evidence directly
present in the immediate speech-act situation. what is crucial here is that
the speaker implicitly denies having any information regarding the sit-
uation prior to the current perceptual experience; in other words, this
knowledge is entirely novel for the speaker. (Sun 1993: 996997, empha-
sis added)
(Sun is discussing an Eastern Tibetic language which does not use dug in
evidential forms, but this is the same pan-Tibetic category which Hill calls
direct.) It is for this reason that I adopt Suns term immediate rather than
direct for this category the meaning of the category is not that the speaker

552 Scott DeLancey
has direct evidence, but that immediate direct evidence is the sole basis for the
statement.
4.3. The immediate evidential: Mirativity right in front of your eyes

4.3.1. The meaning of dug. Hill orients his discussion of my Tibetan anal-
ysis around the form dug (h.dug), originally a verb meaning sit, be located
(Hill 2010: 146), which sense it still retains in the archaic Western and Eastern
languages. In many, but not all, Tibetic languages, this has grammaticalized
into a marker of imperfective or durative aspect (Hahn 1974: 154155, Sax-
ena 1997),7 and further developed into the exponent of the direct/immediate
evidential category as a locative/existential copula and in the imperfective
paradigm. But we cannot simply identify the form and the function. When dug
is involved in the evidential system in a Tibetic language, it expresses the im-
mediate evidential value, but the converse is not true: the immediate evidential
is expressed by a range of different forms and constructions across the branch.
Hill is quite correct that both foreign scholars and native grammarians have
identified the primary sense of dug as indicating direct perceptual knowledge
of the event or state on the part of the speaker, and that I have asserted on the
contrary that eyewitness knowledge is not the relevant criterion (DeLancey
1986: 205), rather, the primary sense of the form is mirative. But Hill is mag-
nifying a small, though not inconsequential, issue here. The association of di-
rect evidence forms in Tibetan with novelty or discovery is well-known and
generally acknowledged. I am hardly alone in this view:
Studies of the epistemic categories expressed in Tibetan auxiliaries and copulas

have mostly compared the phenomena with mirativity marking, and this is no
doubt the correct comparandum in diachronic research. (Bickel 2008: 1)
And it is hardly my own invention:
The choice of the auxiliaries red, h.dug, and LT son indicates that the knowledge
of the speaker is based solely or predominantly on his or her immediate or new
visual experience, uncontaminated by knowledge of different sources. (Zeisler
2004: 300, emphasis added)
In addition to the system of viewpoint [i.e., egophoricity], the choice between

yod. and dug. very often relates to what might be called generality. yod. often
implies that the state of affairs in question is generally the case or has been so
7. Resemblant, but not provably cognate, forms with similar existential/locative and aspectual
functions occur in several nearby languages, including Newar in Nepal, the Tani languages in
Arunachal Pradesh, and the Bodo-Garo languages of the Brahmaputra Valley. This local areal
phenomenon presumably originated in Tibetan.

for a considerable period of time. dug. by contrast often refers to a particular,

even momentary state of affairs, and frequently conveys a sense of surprise,
recency or discovery. (Denwood 1999:122, emphasis added)
Denwoods interpretation of the immediate category is explicit in his standard

gloss for immediate verb forms:
(23) kho-la deb dug
he-loc book exist
(I see) he has some books./Ah, he has some books! (Other-centered;
witnessed; discovery.) (Denwood 1999: 122, emphasis added)
We have already seen that Sun characterizes the immediate category in Eastern
Tibetic similarly. In his analysis of the auxiliary systems of Western Tibetic
languages, Bielmeier includes both mirativity and direct perception as parts of
the definition of duk (< dug) but emphasizes the importance of the former
(Bielmeier 2000: 98, emphasis added):
An attempt to interpret duk on the pragmatic-epistemic level as new objective
non-definite knowledge, usually based on recent visual perception faces the
same problem as in Purik. Can the speaker always see that he is a liar? Using
duk in [kh o remba duk he is strong] he may see the particular quality, but in [kh o
sambatSan duk he is intelligent and kh o Sobo duk he is a liar] it seems more
appropriate to interpret the speakers statement as focusing on his new
objective non-definite knowledge.
So although Hill is at great pains to contrast my heretical interpretation of dug

with the orthodox tradition, there is no such radical disagreement; other schol-
ars who have looked at the problem see it similarly. Presumably Hill sees a
contradiction concerning how to characterize the Tibetan immediate evidential
category in terms of our superordinate categories: since mirativity is distinct
from evidentiality, there is a contradiction in calling it as both at the same time.
If this is his point, then it is virtually the only thing he says that I can agree with.
In the next section we will see that, as part of an evidential paradigm, the im-
mediate category should, as Zeisler and Hill insist, be primarily characterized
as an evidential, and why.
4.3.2. Mirativity and the immediate evidential. The argument which Hill
seems to be trying to articulate is made succinctly and coherently by Zeisler
(2004: 301304), who points out that if the Tibetan immediate evidential were
a true mirative, it should be usable in inferential as well as direct evidential
contexts, as long as the information related is new. This is, of course, correct,
and is true of typical mirative constructions such as those of Kham, Kalasha,
and Hare. Zeisler, like Hill, finds the notion of mirativity otiose for the descrip-
tion of Tibetan. But unlike him, she recognizes that this is because the direct

554 Scott DeLancey
evidential category in Tibetan has a strong mirative component (Zeisler 2004:

658):
This distinction in terms of what should rather be called novelty corresponds
more or less to the distinction in terms of evidentiality: immediate perception,
not contaminated by prior (i.e. old or assimilated) knowledge is new and in a
way not yet integrated into the overall knowledge.
For Zeisler, the defining semantic feature of the Tibetan category is uncon-
taminated immediate perception, i.e., information known through direct per-
ception but no other source. That is, Tibetic immediate evidential forms based
on dug do not simply express that the speakers basis for the statement is direct
evidence. Rather, they call explicit attention to that fact, in order to express that
that direct evidence is the sole basis which the speaker has for the statement.
The same point is made (but not sufficiently developed) in DeLancey (2001),
that eyewitness knowledge does not forbid the use of the personal or generic
forms, but that, as Zeisler notes, the direct construction indicates that the
speaker has only direct perceptual evidence, and had no knowledge of the fact
prior to perceiving it. The strictly limited use of the immediate evidential form
compared to miratives such as those of Kham or Hare is an inevitable conse-
quence of its position in a paradigm where it contrasts with other epistemic
categories, the personal and generic. As pointed out above, a language will not
have a construction which is used always and only for statements based on
direct perception, unless that construction contrasts with one or more construc-
tions which express other evidential categories. Typically we find a type A1
two-term contrast between direct and indirect evidence, as in Chechen, but the
Tibetic languages show us the more complex, and quite strange, three-term sys-
tem which we have briefly surveyed. The sense of direct sensory knowledge,
rather than the commoner sense of definite first-hand knowledge (de Haan
1999, 2001), arises through contrast with the generic and egophoric forms. So
the very specific sense associated with dug in most of the constructions in
which it occurs in Tibetic languages is very much an effect of the paradigm of
which it is part. Needless to say, one cannot simply wander off into the wilder-
ness of linguistic variety expecting to find close matches with such a form in
languages which lack the rest of the paradigm. But this is precisely what Hill
has done, and the results are what one would expect.
Still Zeisler and Hill are correct in pointing out that despite its strong mi-
rative connotations, the immediate evidence category in Tibetic languages is,
strictly speaking, an evidential category, and thus by definition not a pure mira-
tive. Note that the immediate category contrasts with the personal and inferen-
tial categories; it cannot co-occur with evidential forms as we see in Chechen
and Magar. In the next subsection I will speculate on the historical connection
between the immediate evidential and mirativity sensu strictu. In Section 4.4

we will see that, in addition to the immediate evidential category, some Tibetic
languages also have a true mirative.
4.3.3. On the origin of Tibetic egophoric systems. As we can see even from
this very sketchy glimpse, the simple concept of mirativity is not a magi-
cal key to the amazing verbal systems of modern Tibetic languages, whose
conjunct/disjunct or egophoric systems are very different from what we
are used to in the world of evidentials. Tibetic, and probably specifically
Central Tibetic, seems to be the center from which egophoric systems have
spread around the Himalayas. In the Tibetan cultural area, true egophoricity
has spread to a few non-Tibetic languages, including not only Bodic but non-
Tibetic Dakpa (DeLancey 1992b: 4648, where the language is called Cuona
Monpa), Bodish but non-Bodic Newar (Hargreaves 2005) and Kaike (D. Wat-
ters 2006), Tibeto-Burman but non-Bodish Akha (Egerod & Hansson 1974,
Thurgood 1986) and Sangkong (see DeLancey 2010: 45), and Tibetospheric
but non-Tibeto-Burman Mongguor (Mongolic; Chinggeltai 1989, Slater 2003)
and Xibe (Tungusic; L 1984, Jang et. al. 2011).
Curnow (2001), and now Hill, criticize my description of Tibetic conjunct-
disjunct systems as grammaticalized mirativity, and that may not have been
the most illuminating description. A better way of putting it is that the pe-
culiarities of the Tibetan system are a result of the expansion of a simple
two-term mirative system into a larger and more complex (more grammat-
ical) system in which each term is constrained to a relatively narrow se-
mantic space. It is clear that, historically, the conjunct or egophoric forms,
which express the category which I have been calling personal, are origi-
nal, and that the conjunct-disjunct/egophoric-allophoric/personal-direct system
arose through the innovation of the constructions which now express the direct
evidential category (Takeuchi 1990, DeLancey 1992b, Denwood 1999). This
is why, pace Curnow, the disjunct/allophoric forms are still formally marked,
and the conjunct/egophoric forms are unmarked. For Curnow this seems to be
a matter of intuition, but the formal evidence is that the contrast is neutralized
before subordinating constructions, where only the conjunct forms can appear
(Chang & Chang 1984, DeLancey 1990a). The member of an opposition which
consistently occurs in conditions of neutralization is the unmarked member by
definition.
The evidential/egophoric systems of the modern Tibetic languages are not
cognate; although often we may see cognate morphemes occurring in dif-
ferent systems, they have developed semi-independently in the different lan-
guages, and do not always correspond even in morphological structure (Sun
1993, Bielmeier 2000). There is only the most tentative indication of evidential
marking in Classical Tibetan (Takeuchi 1990, Denwood 1999: 246), and none
at all of an egophoric/allophoric distinction or of the relevance of volitional-

556 Scott DeLancey
ity (Denwood 1999: 249). In early vernacular writing we see the beginnings
of constructions which have evolved into the Central Tibetic and other verbal
systems (Saxena 1997, Zeisler 2004). Thus the elaborate egophoric systems
of the modern languages are a recent development (Takeuchi 1990, DeLancey
1992, Tournadre 1996, Saxena 1997, Hongladarom 2007). Although they are
similar in the set of semantic categories expressed, they differ considerably
in the constructions and specific morphological material through which they
are expressed. In Central (dBus-gTsang or -tsang) Tibetan, the development
of the system involved the innovation of new locational/existential and equa-
tional copulas to contrast with original yod and yin. This seems to have begun
with the initial grammaticalization of dug, originally sit, stay, into a loca-
tional/existential copula contrasting with yod. The spread of this contrast to the
equational system is much later, so that even the very closely related Lhasa and
Shigatse dialects of Central Tibetan have different innovative forms Lhasa
red corresponds to Shigatse sbas.8
I suggest that the original innovative category represented by the first gram-
maticalization of dug was mirative, on the grounds that a two-term system
will most likely be either unmarked-mirative or direct-indirect, and such sys-
tems typically originate from the innovation of a mirative or indirect category.
Note that where we find two-way oppositions in neighboring languages, such
as Kham, Sunwar, and Qiang, it tends to be the familiar mirative. Since dug is
the innovative form, and nothing in its subsequent history suggests any asso-
ciation with indirectivity, the most likely inference is that it began as a simple
mirative. As we have seen, its semantic range has become more constrained as
the evidential paradigm in which it participates has become more complex and
precise, but it still retains a strong mirative sense, recognized by most authori-
ties on the subject. Eventually (apparently over not more than a few centuries)
the distinction came to be adopted into the verbal system through finitization
of nominalized clause constructions (DeLancey 2011).
Conjunct/disjunct or egophoric systems are not confined to the Tibetospheric
world. Outside of the Tibetan area, conjunct/disjunct-like phenomena have
been reported from the Andes (Dickinson 2000, Curnow 2002, Bruil 2009),
and more recently from the Caucasus (Bickel 2008; Creissels 2008a, b) and
Papua New Guinea (Loughnane 2009). Similarities and differences among all
of these systems have yet to be explored; the suggestions above are solely con-
cerned with the origins of the system in Central Tibetan.
8. I take this spelling from Jn (1958); Haller (2000b) gives bad. The form is /pie/.


4.4. Tibetan even has pure miratives

Hill makes two assertions: that Tibetan dug does not express mirativity, and
that no other category in any other language does either. As we have seen, it
is generally recognized that dug does in fact have a strong mirative force, but
there is a good argument that the category with which it is associated is fun-
damentally an evidential rather than a mirative per se. But even in the Tibetic
languages we find numerous examples of mirative constructions which do not
involve the dug morpheme. Besides the four evidential categories which we
have seen, a fifth term is added to the system in some Tibetic languages, as
in the Central Tibetic language Dzongkha (Bhutan), where van Driem (1992:
169) describes a suffix of acquired knowledge:
The suffix -pas/-bas -b/-w is attached to the regular stem of a verb denoting
a state or condition and indicates that the information expressed in the sentence
is newly acquired knowledge. Conversely, when the suffix is not used [. . . ] this
implies that the situation expressed forms part of the ingrained knowledge of the
speaker, something the speaker has known all along or which, at least, is not a
recently acquired insight or not an only recently observed phenomenon.
The form is a nominalizer inflected for ablative or instrumental case; Noonan

(2008) notes that a common function of a nominalized clause use as finite in
Bodic languages is to express mirativity. We do not at present have a suffi-
ciently detailed description of a system which includes this category9 to allow
a serious investigation of how it relates to the immediate evidential category.
Nevertheless, even if Hill were correct in denying the mirative nature of the
direct evidential category, it turns out that mirativity is still attested as a gram-
matical category in Tibetic.
Similar developments have occurred in other Tibetic languages (see Tour-
nadre 1996, who uses the term rvlatif). A clear description of the interac-
tion and independence of mirativity and evidentiality in a dialect of the western
Spiti language is Hein (2007). (It may be that the Tabo mirative form has some
etymological relation to the Dzongkha acquired knowledge form (Hein 2007:
201202).)
5. Mirative forever
We have seen that even the original case for the mirative can easily weather an
attack such as Hills. I hope that in this article I have buttressed the original
case sufficiently to meet Lazards and Plungians challenges. In any case, it is
a fact, as even Hill acknowledges, that the concept has proved very useful to
descriptive linguists, and has contributed to insightful descriptions and analyses
9. See S. Watters (2007) for a start.

558 Scott DeLancey
of a wide range of languages around the world. The case for the mirative no
longer rests simply on my argumentation or examples, but on a broad base of
data and analysis. To refute the case one would need to demonstrate how these
languages can be better and more insightfully described in some other terms.
(This is what Hill pretends to do, but we have seen the results of that.)
Recent research opens up new vistas in the study of mirativity. There are
many empirical questions concerning the interaction of mirativity with person
(Hein 2001, 2007), with volitionality, with politeness, and many other factors.
Several authors have recently reported 2nd person miratives (Enfield 2007;
Bickel 2008; Molochieva 2010; Hyslop 2011a, b). It is clear, and widely ac-
knowledged, that Tibetic systems involve the interaction of mirativity and evi-
dentiality with person, see, e.g., Bickel (2008: 2):
What differentiates Tibetan epistemic morphology from standard average mira-
tives can best be captured by different responses to two key variables: (i) whose
knowledge is at issue? (ii) what is the knowledge about? I refer to the first variable
as the PERSON variable, and to the second variable as the SCOPE variable.
Crosslinguistic study of conjunct-disjunct systems is only beginning, but al-

ready there is reason to doubt whether non-Tibetic systems of this sort neces-
sarily involve evidentiality or mirativity at all. Neither seems to be relevant in
the Newar system, where intention/volitionality is the fundamental category
involved (Hargreaves 2005). On the other hand, the opposition in Kaike is
strongly mirative, and volitionality plays no evident role (D. Watters 2006).
An egophoric opposition in Galo, a Tibeto-Burman language of Arunachal
Pradesh, is concerned only with the personal/non-personal distinction in cat-
egories of knowledge, and seems to be unconcerned with either volition or
mirativity (Post no date). As we come to better understand these and other
newly-described systems, we will be able to see further into the nature of these
component categories and their interrelations.
Another potential line of research involves the interaction of mirativity with
the expression of emotional attitude. Unexpected information often provokes
some kind of emotional reaction, and it is inevitable that some emotional value
will sometimes attach to a mirative construction. Several available descriptions
refer to implications of disapproval or other negative attitudes associated with
mirative constructions (e.g., Hein 2007, So-Hartmann 2009: 293294). We are
seeing studies such as Queixalos (2007), which present finer-grained analyses
of the various semantic subcategories of a mirative construction; detailed stud-
ies of this sort will be critical to developing our understanding of the place of
mirativity in the world of semantic categories.
It requires a peculiar conception of language description and linguistic ty-
pology to imagine that an imaginary descriptive category, as Hill deems the
mirative to be, could capture the attention of a faddish public and spawn the

amount of specious description, historical reconstruction, and typological anal-

ysis that the mirative has been the occasion of over the past fifteen years. The
efforts of mine to which Hill attributes the popularity of mirativity have, to
tell the truth, not been very engaged with the topic since 2001. So the energy
which has powered the bubble must have come from somewhere else. It is
evident to me that the explanation lies in the fact that many descriptivists en-
counter phenomena which are more easily described in terms of mirativity than
as anything else. Hill never deals with the question of what it is to be a valid
category. For me it is quite simple. The mirative has proven itself to be useful
in the description of a wide range of languages, and descriptive utility is the
only validity there is.
Received: 11 April 2012 University of Oregon

Revised: 9 May 2012
Correspondence address: Department of Linguistics, University of Oregon 1290, Eugene, OR

97403-1290, U.S.A.; e-mail: [email protected]
Abbreviations: 1/2/3 1st/2nd/3rd person; ag agent; all allative; ant anterior; cep counter-
expectancy particle; com comitative; conf confirmative; cont continuous; cop copula; cvb con-
verb; dem demonstrative; du dual; emph emphatic; erg ergative; foc focus; gen genitive; hon
honorific; impf imperfective; inc inclusive; iness inessive; infer inferential; intr intransitive; j
J gender; loc locative; mir mirative; neg negative; nf nonfinal; nmz nominalizer; nom nomi-
native; obj object; pfv perfective; pl plural; prf perfect; pst past; sg singular; su subject.
References
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. & R. M. W. Dixon (eds.). 2003. Studies in evidentiality. Amsterdam:
Benjamins.
Akatsuka, Noriko. 1985. Conditionals and the epistemic scale. Language 61. 625639.
Aksu-Ko, Ayhan A. & Dan I. Slobin. 1986. A psychological account of the development and use
of evidentials in Turkish. In Chafe & Nichols (eds.) 1986, 159167.
Anderson, Lloyd. 1986. Evidentials, paths of change, and mental maps: Typologically regular
asymmetries. In Chafe & Nichols (eds.) 1986, 273312.
Andvik, Erik. 2010. A grammar of Tshangla. Leiden: Brill.
Bashir, Elena. 1988. Inferentiality in Kalasha and Khowar. Chicago Linguistic Society 24(1). 15
30.
Bashir, Elena. 2007. Evidentiality in South Asian languages. In Miriam Butt & Tracy Hol-
loway King (eds.), Proceedings of the LFG06 Conference, 3050. Stanford: CSLI. http://
cslipublications.stanford.edu/LFG/11/lfg06.pdf
Bashir, Elena. 2010. Traces of mirativity in Shina. Himalayan Linguistics 9(2). 155.
Beckwith, Christopher. 2009. Empires of the Silk Road: A history of Central Asia from the Bronze
Age to the present. Princeton, NJ: Princeton University Press.
Bickel, Balthasar. 2008. Verb agreement and epistemic marking: A typological journey from the
Himalayas to the Caucasus. In Huber et. al. (eds.) 2008, 114.
Bielmeier, Roland. 2000. Syntactic, semantic, and pragmatic-epistemic functions of auxiliaries in
Western Tibetan. Linguistics of the Tibeto-Burman Area 23(2). 79125.

560 Scott DeLancey
Binnick, Robert. 2012. The past tenses of the Mongolian verb. Leiden: Brill.
Bruil, Martine. 2009. Evidentiality and mirativity in Barbacoa languages: Was there such a thing as
grammaticalized evidentiality in Proto-Barbacoa? Paper read at the Workshop on the Struc-
ture and Constituency in Languages of the Americas 14, Purdue University.
Chafe, Wallace & Johanna Nichols (eds.). 1986. Evidentiality: The linguistic coding of epistemol-
ogy. Norwood, NJ: Ablex.
Chang, Betty Shefts & Kun Chang. 1984. The certainty hierarchy among Spoken Tibetan verbs of
being. Bulletin of the Institute of History and Philology, Academia Sinica 55. 603635.
Chelliah, Shobhana, & Willem de Reuse. 2011. Handbook of descriptive linguistic fieldwork. Dor-
drecht: Springer.
Chinggeltai. 1989. On subjective mood and objective mood in the Monguor language. In Walter
Heissig & Klaus Sagaster (eds.), Gedanke und Wirkung: Festschrift zum 90. Geburstag von
Nikolaus Poppe, 6775. Wiesbaden: Harrassowitz.
Chirkova, Ekaterina. 2008. Bima yu yu Zng yu fangyn de shzhng fnchu. [The evidential
category in Baima and Tibetan dialects.] Mnz Yuwn 2008(3). 3643.
Creissels, Denis. 2006. Syntaxe gnrale: Une introduction typologique, Vol. 1: Catgories et con-
structions. Paris: Lavoisier.
Creissels, Denis. 2008a. Remarks on so-called conjunct/disjunct systems. Paper read at the Syn-
tax of the Worlds Languages III conference, Berlin.
Creissels, Denis. 2008b. Person variations in Akhvakh verb morphology: Functional motivation
and origin of an uncommon pattern. Sprachtypologie und Universalienforschung 61. 309
325.
Curnow, Timothy. 2001. Why first/non-first person is not grammaticalized mirativity. In Keith
Allan & John Henderson (eds.), Proceedings of ALS2k, the 2000 Conference of the Australian
Linguistic Society, 110. http://www.als.asn.au/proceedings/als2000/curnow.pdf
Curnow, Timothy. 2002. Conjunct/disjunct marking in Awa Pit. Linguistics 40. 611627.
de Haan, Ferdinand. 1999. Evidentiality and epistemic modality: Setting boundaries. Southwest
Journal of Linguistics 18. 83101.
de Haan, Ferdinand. 2001. The place of inference within the evidential system. International Jour-
nal of American Linguistics 67. 193219.
de Haan, Ferdinand. 2008. Evidentiality in Athabaskan. Coyote Papers: Working Papers in Lin-
guistics 16. 6781. http://coyotepapers.sbs.arizona.edu/CPXVI/6.deHaan.pdf
DeLancey, Scott. 1982. Aspect, transitivity, and viewpoint. In Paul Hopper (ed.), Tense-aspect:
Between semantics and pragmatics, 167183. Amsterdam: Benjamins.
DeLancey, Scott. 1986. Evidentiality and volitionality in Tibetan. In Chafe & Nichols (eds.) 1986,
203213.
DeLancey, Scott. 1990a. Ergativity and the cognitive model of event structure in Lhasa Tibetan.
Cognitive Linguistics 1. 289321.
DeLancey, Scott. 1990b. Notes on evidentiality in Hare. International Journal of American Lin-
guistics 56. 152158.
DeLancey, Scott. 1992a. Sunwar copulas. Linguistics of the Tibeto-Burman Area 15(1). 3138.
DeLancey, Scott. 1992b. The historical status of the conjunct/disjunct pattern in Tibeto-Burman.
Acta Linguistica Hafniensia 25. 3962.
DeLancey, Scott. 1997. Mirativity: The grammatical marking of unexpected information. Linguis-
tic Typology 1. 3352.
DeLancey, Scott. 2001. The mirative and evidentiality. Journal of Pragmatics 3. 371384.
DeLancey, Scott. 2010. Towards a history of verb agreement in Tibeto-Burman. Himalayan Lin-
guistics 9(1). 138.
DeLancey, Scott. 2011. Finite structures from clausal nominalization in Tibeto-Burman languages.
In Foong Ha Yap, Karen Grunow-Hrsta & Janick Wrona (eds.), Nominalization in Asian
languages: Diachronic and typological perspectives, 343362. Amsterdam: Benjamins.
Denwood, Philip. 1999. Tibetan. Amsterdam: Benjamins.

de Reuse, William. 2003. Evidentiality in Western Apache (Athabaskan). In Aikhenvald & Dixon
(eds.) 2003, 79100.
Dickinson, Connie. 2000. Mirativity in Tsafiki. Studies in Language 24. 379422.
Dixon, R. M. W. 2003. Evidentiality in Jarawara. In Aikhenvald & Dixon (eds.) 2003, 165187.
Donabdian, Anad. 1996. Pour une interpretation des diffrentes valeurs du mdiatif en armnien
occidental. In Guentchva (ed.) 1996, 87108.
Donabdian, Anad. 2001. Towards a semasiological account of evidentials: An enunciative ap-
proach of -er in Modern Western Armenian. Journal of Pragmatics 33. 421442.
Driem, George van. 1992. The grammar of Dzongkha. Thimphu: Dzongkha Development Com-
mission, Royal Government of Bhutan.
Eberhard, Wolfram. 1942. Kultur und Siedlung der Randvlker Chinas (Supplement to Toung Pao
36). Leiden: Brill.
Egerod, Sren & Inga-Lill Hansson. 1974. An Akha conversation on death and funeral. Acta Ori-
entalia (Hafnensia) 36. 225284.
Enfield, N. J. 2007. A grammar of Lao. Berlin: Mouton de Gruyter.
Friedman, Victor. 2007. The expression of speaker subjectivity in Lak (Daghestan). In Guentchva
& Landaburu (eds.) 2007, 351376.
Garrett, Edward. 2001. Evidentiality and assertion in Tibetan. Los Angeles: University of Califor-
nia at Los Angeles doctoral dissertation.
Grunow-Hrsta, Karen. 2007. Evidentiality and mirativity in Magar. Linguistics of the Tibeto-
Burman Area 30(2). 151194.
Guentchva, Zlatka (ed.) 1996. Lnonciation mdiatise. Leuven: Peeters.
Guentchva, Zlatka & Jon Landaburu (eds.). 2007. Lnonciation mdiatise II: Le traitement
pistmologique de linformation: Illustrations amrindiennes et caucasiennes. Leuven:
Peeters.
Hahn, Michael. 1974. Lehrbuch der Klassischen Tibetischen Schriftsprache. Bonn: Hahn.
Reprinted, Bonn: Indica et Tibetica Verlag, 1985.
Haller, Felix. 2000a.Verbal categories of Shigatse Tibetan and Themchen Tibetan. Linguistics of
the Tibeto-Burman Area 23(2). 175191.
Haller, Felix. 2000b. Dialekt und Erzhlungen von Shigatse. Bonn: VGH.
Hargreaves, David. 2005. Agency and intentional action in Kathmandu Newar. Himalayan Lin-
guistics 5. 148.
Hsler, Katrin. 2001. An empathy-based approach to the description of the verb system of the Dege
dialect of Tibetan. Linguistics of the Tibeto-Burman Area 24(1). 134.
Hein, Veronika. 2001. The role of the speaker in the verbal system of the Tibetan dialect of
Tabo/Spiti. Linguistics of the Tibeto-Burman Area 24(1). 3548.
Hein, Veronika. 2007. The mirative and its interplay with evidentiality in the Tibetan dialect of
Tabo (Spiti). Linguistics of the Tibeto-Burman Area 30(2). 195214.
Heine, Bernd & Tania Kuteva. 2002. World lexicon of grammaticalization. Cambridge: Cambridge
University Press.
Hill, Nathan W. 2010. A lexicon of Tibetan verb stems as reported by the grammatical tradi-
tion. Mnchen: Kommission fr Zentral-und Ostasiatische Studien, Bayerische Akademie
der Wissenschaften.
Hill, Nathan W. 2012a. A note on the history and future of the Wylie system. Review dtudes
Tibtaines 23. 103105.
Hill, Nathan W. 2012b. Mirativity does not exist: h.dug in Lhasa Tibetan and other suspects.
Linguistic Typology 16. 389433.
Hongladarom, Krisadawan. 2007. Evidentiality in Rgyalthang Tibetan. Linguistics of the Tibeto-
Burman Area 30(2). 1744.
Hung, Bfn. 1991. Zng-Mian yu de qngti fnchu. [The category of modality in Tibeto-
Burman.] Mnz Yuwn 1991(2). 2230.

562 Scott DeLancey
Huber, Brigitte. 2000. Preliminary report on evidential categories in Lende Tibetan (Kyirong).
Linguistics of the Tibeto-Burman Area 23(2). 155174.
Huber, Brigitte. 2005. The Tibetan dialect of Lende (Kyirong). Bonn: VGH.
Huber, Brigitte, Marianne Volkart & Paul Widmer (eds.). 2008. Chomolangma, Demawend und
Kasbek: Festschrift fr Roland Bielmeier zu seinem 65. Geburtstag. Halle (Saale): Interna-
tional Institute for Tibetan and Buddhist Studies.
Hyslop, Gwendolyn. 2011a. A grammar of Kurtp. Eugene, OR: University of Oregon doctoral
dissertation.
Hyslop, Gwendolyn. 2011b. Mirativity in Kurtp. Journal of South Asian Linguistics 4. 4360.
Jang, Taeho, Kyungsook Jang & Tom Payne. 2011. Conjunct and disjunct marking in Xibe
(Tungusic-Manchu). Presented at the Thammasat University Linguistics Symposium,
Bangkok, September 2011.
Jn, Png. 1958. Zng yu Las Rkaz Changdou hu de bjio ynjiu. [A comparative study of
Lhasa, Shigatse and Chamdo Tibetan.] Beijng: Kexu Chubansh.
Jn, Png. 1979. Ln Zng yu Las kouyu dngc de tdian yu yufa jigu de guanx. [On the
relations between the characteristics of the verb and the syntactic structure in Spoken Tibetan
(Lhasa dialect).] Mnz Yuwn 1979(3). 173181.
Jn, Png. 1983. Zng yu jianzh. [Outline of Tibetan.] Beijng: Mnz Chubansh.
Johanson, Lars. 2000. Turkic indirectives. In Johanson & Utas (eds.) 2000, 6187.
Johanson, Lars. 2003. Evidentiality in Turkic. In Aikhenvald & Dixon (eds.) 2003, 273290.
Johanson, Lars & Bo Utas (eds.). 2000. Evidentials: Turkic, Iranian and neighbouring languages.
Berlin: Mouton de Gruyter.
Kwon, Iksoo. 2010. I guess Korean has some more mirative markers: -Napo- and -Nmoyang-.
Paper read at the 84th Annual Meeting of the Linguistic Society of America, Baltimore.
Kwon, Iksoo (forthcoming). Please confirm what I inferred: On the Korean inferential-evidential
marker Napo-. Journal of Pragmatics 44.
LaPolla, Randy. 2003. Evidentiality in Qiang. In Aikhenvald & Dixon (eds.) 2003, 6378.
Lazard, Gilbert. 1996. Le mdiatif en persan. Guentchva (ed.) 1996, 2130.
Lazard, Gilbert. 1999. Mirativity, evidentiality, mediativity, or other? Linguistic Typology 3. 91
109.
Lazard, Gilbert. 2000. Le mdiatif: Considerations thoriques et application liranien. In Johan-
son & Utas (eds.) 2000, 209228.
Leinonen, Marja. 2000. Evidentiality in Komi Zyryan. In Johanson & Utas (eds.) 2000, 419440.
L, Shln. 1984. Xb yu dngc chnsh sh de qnzh kouq h fei qnzh kouq. [The personally
knowing and non-personal knowing aspects of the verbal statement in Xibo language.] Mnz
Yuwn 1984(6). 2632.
Lidz, Liberty. 2007. Evidentiality in Yongning Na (Mosuo). Linguistics of the Tibeto-Burman Area
30(2). 4587.
Lindstedt, Jouko. 2000. The perfect aspectual, temporal, and inferential. In sten Dahl (ed.),
Tense and aspect in the languages of Europe, 365383. Berlin: Mouton de Gruyter.
Loughnane, Robyn. 2009. A grammar of Oksapmin. Melbourne: University of Melbourne doctoral
dissertation.
Maisak, Timur, & Sergei Tatesovov. 2007. Beyond evidentiality and mirativity: Evidence from
Tsakhur. In Guentchva & Landaburu (eds.) 2007, 477405.
Maslova, Elena. 2003. Evidentiality in Yukaghir. In Aikhenvald & Dixon (eds.) 2003, 219235.
Melcuk, Igor. 1994. Cours de morphologie gnrale, Vol. 2: Significations morphologiques. Mon-
tral: Les Presses de lUniversit de Montral.
Michailovsky, Boyd. 1996. Linferntel du npali. In Guentchva (ed.) 1996, 109123.
Molochieva, Zarina. 2010. Tense, aspect and mood in Chechen. Leipzig: Universitt Leipzig doc-
toral dissertation.

Noonan, Michael. 2008. Nominalizations in Bodic languages. In Mari Lpes-Couso & Elena
Seoane (eds.), Rethinking grammaticalization: New perspectives, 219273. Amsterdam: Ben-
jamins.
Peterson, Tyler. 2010. Epistemic modality and evidentiality in Gitksan at the semantics-pragmatics
interface. Vancouver: University of British Columbia doctoral dissertation.
Plungian, Vladimir. 2010. Types of verbal evidential marking: An overview. In Gabriele Diewald &
Elena Smirnova (eds.), Linguistic realization of evidentiality in European languages, 1558.
Berlin: De Gruyter Mouton.
Post, Mark (no date). Person-sensitive TAME marking in Galo: Historical origins and functional
motivation. Unpublished ms., Universitt Bern.
Queixalos, Francesc. 2007. Le sikuani et la catgorie de source de linformation. In Guentchva &
Landaburu (eds.) 2007, 129150.
Roerich, J. N. 1930. The animal style among the nomad tribes of northern Tibet. Praha: Semina-
rium Kondakovianum.
Saxena, Anju. 1997. Aspect and evidential morphology in Standard Lhasa Tibetan: A diachronic
study. Cahiers de linguistique Asie orientale 26(2). 281306.
Shirai, Satoko. 2007. Evidentials and evidential-like categories in nDrapa. Linguistics of the Tibeto-
Burman Area 30(2). 125150.
Slater, Keith W. 2003. A grammar of Mangghuer: A language of Chinas Qinghai-Gansu Sprach-
bund. London: RoutledgeCurzon.
So-Hartmann, Helga. 2009. A descriptive grammar of Daai Chin (STEDT Monograph 7). Berke-
ley, CA: Sino-Tibetan Etymological Dictionary and Thesaurus Project.
Sun, Jackson T.-S. 1993. Evidentials in Amdo Tibetan. Bulletin of the Institute of History and
Philology 1993. 9451001.
Takeuchi, Tsugihito. 1990. Chibettogo no jutsubu ni okeru judoshi no kino to sono hattatsu
kutei. [The semantic functions of auxiliaries in Tibetan and their historical development.]
In Sakiyama Osamu & Sato Akihiro (eds.), Ajia no shogengo to ippan gengogaku, 616.
Tokyo: Sanseido.
Tatevosov, Sergei. 2001. From resultatives to evidentials: Multiple uses of the perfect in Nakh-
Daghestanian languages. Journal of Pragmatics 33(3). 443464.
Tatevosov, Sergei. 2007. Evidentiality and mirativity in the Mishar dialect of Tatar. In Guentchva
& Landaburu (eds.) 2007, 407433.
Tatevosov, Sergei & Timur Maisak. 1999. Formy admirativnoj semantiki. [Forms of admira-
tive semantics.] In Andrej Kibrik & Jakov Testelec (eds.), Elementy caxurskogo jazyka v
tipologiceskom osvecenii, 289292. Moskva: Nasledie.
Thurgood, Graham.1986. The nature and origins of the Akha evidentials system. In Chafe &
Nichols (eds.) 1986, 214222.
Tournadre, Nicolas. 1996. Comparaison des systmes mdiatifs de quatre dialects tibtains (tib-
tain central, ladakhi, dzongkha et amdo). In Guentchva (ed.) 1996, 195211.
Tournadre, Nicolas. 2001. Final auxiliary verbs in Literary Tibetan and in the dialects. Linguistics
of the Tibeto-Burman Area 24(1). 49111.
Tournadre, Nicolas. 2008. Arguments against the concept of conjunct/disjunct in Tibetan. In
Huber et. al. (eds.) 2008, 281308.
Volkart, Marianne. 2000. The meaning of the auxiliary morpheme dug in the aspect systems of
some Central Tibetan dialects. Linguistics of the Tibeto-Burman Area 23(2). 127153.
Watters, David. 2002. A grammar of Kham. Cambridge: Cambridge University Press.
Watters, David. 2006. The conjunct-disjunct distinction in Kaike. Nepalese Linguistics 22. 300
319.
Watters, Stephen. 2007. The nature of narrative text in Dzongkha: Evidence from deixis, eviden-
tiality, and mirativity. In Roland Bielmeier & Felix Haller (eds.), Linguistics of the Himalayas
and beyond, 381397. Berlin: Mouton de Gruyter.

564 Scott DeLancey
Willis, Christina. 2007. Evidentiality in Darma (Tibeto-Burman). Linguistics of the Tibeto-Burman

Area 30(2). 89124.
Wylie, Turrell. 1959. A standard system of Tibetan transcription. Harvard Journal of Asiatic Stud-
ies 22. 261267.
Zeisler, Bettina. 2000. Narrative conventions in Tibetan languages: The issue of mirativity. Lin-
guistics of the Tibeto-Burman Area 23(2). 3977.
Zeisler, Bettina. 2004. Relative tense and aspectual values in Tibetan languages. Berlin: Mouton
de Gruyter.


Still Mirative After All These Years PDF

Uploaded by

Copyright:

Available Formats

Still Mirative After All These Years PDF

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

Still Mirative After All These Years PDF

Uploaded by

Copyright:

Available Formats

Still mirative after all these years

Linguistic Typology 16 (2012), 529564 14300532/2012/016-0529

Brought to you by | University of Oregon Library

2. Mirativity is real after all

Brought to you by | University of Oregon Library

2.1. Hills critique

Brought to you by | University of Oregon Library

2.2. Meet the mirative

Hill runs through an assortment of such descriptions in the literature, mostly

Brought to you by | University of Oregon Library

2.3. Mirativity and the mirative

Brought to you by | University of Oregon Library

(1) konna ni yorokonde kureru no nara, motto hayaku

(3) tu khowr kor-k bir-u

She then goes on to characterize (3) as

Brought to you by | University of Oregon Library

[. . . ] a surprised reaction to an immediate, first-hand observation of an unexpected

2.4. The mirative in Kham

2. As in many Tibeto-Burman languages, several morphologically nominalized constructions in

Brought to you by | University of Oregon Library

(6) a-k@ z@ o-k@i-wo oleo

(7) nahm-ni ge-hu-zya-k@ te, kh:bya ekh@

Brought to you by | University of Oregon Library

(9) ao-r@ te zha:h-r@ ci oleo, h@i li-d@ te, Na: te

Watters (2002: 292) then goes on:

Brought to you by | University of Oregon Library

(10) na:h-kin te ph@ri, ao te p@rmesor-e o-cak@rya

2.4.3. Hearsay. Hill further asserts that:

Brought to you by | University of Oregon Library

2.5. The Athabaskan mirative

Brought to you by | University of Oregon Library

(13) John deshta raweya l

Brought to you by | University of Oregon Library

meaning of the mirative is mirativity, regardless of source of information. We

2.6. Extended and rhetorical uses of the mirative

(14) deshta yedanyie l

Brought to you by | University of Oregon Library

2.6.2. Mirativity in narrative. Another category of example of the mirative

Brought to you by | University of Oregon Library

(17) Kathmandu-ni ji pi adikhari sahibe m@ni ge-lai ciu:na

3. Mirativity, evidentiality, mediativity, or other?

Brought to you by | University of Oregon Library

3.1. There can be no direct evidential without a contrasting indirect

Brought to you by | University of Oregon Library

3.2. The mediative

Brought to you by | University of Oregon Library

Let us see whether we can or not.

Brought to you by | University of Oregon Library

3.3. Mirative and indirect evidence together

The mirative construction is very similar to that of Kham, a nominalized clause

Brought to you by | University of Oregon Library

(19) a. kumari bHim-o im-aN mu-o le

And the mirative combines freely with the indirect evidentials:

(20) hose-ko-ko das-o le-sa

Thus in Magar it is impossible to consider mirativity to be simply some

(21) a. Zaara j-ea-na vaiga

Brought to you by | University of Oregon Library

d. Zaara j-ie-na xilla-q

Brought to you by | University of Oregon Library