Aspect and epistemic notions in the system of the present tense forms in Khalkha Mongolian Текст научной статьи по специальности «Языкознание и литературоведение»

Benjamin Brosig

Stockholm University, Stockholm

ASPECT AND EPISTEMIC NOTIONS IN THE PRESENT TENSE SYSTEM OF KHALKHA MONGOLIAN1

This paper deals with the present tense system of spoken Khalkha Mongolian. More specifically, it treats the finitely usable Potential, Habitual, "Resultative" and "Future" markers in their uses on their own and in interaction with several non-finite aspectual markers. As attributive, nominalized and negated predications constitute a separate (albeit related) system, this discussion is restricted to finitely used positive aspect markers in declarative and interrogative sentences. While most modal notions belong to distinct systems, non-past evidentiality, epistemic modality and other modal notions will be touched upon to some degree.

The paper is structured as follows: section 1 addresses the data used for this study. In section 2, the inventory of finite forms in Khalkha Mongolian is first introduced and then put into a semantic context. Section 3 treats the Potential -n and "Resultative" -aa, both as they are used on their own and as markers of evidentiality in combination with the non-finite Progressive, Continuative-Resultative,

1 This paper would not have come about without the input of about 50 informants, most of all Saraa, Soloo, Odko and Galaa. The immense corpus was almost single-handedly transcribed by B. Zoljargal. Thanks also go to D. Guntsetseg for discussing the translations of all example sentences, to Östen Dahl, John Street and Maria Koptjevskaja-Tamm for suggestions that helped to improve both structure and wording, to Stefan Georg, Hans-Jürgen Sasse, Leila Behrens and Dagmar Jung for their inspiration earlier on, to J. Bayansan, M. Bayarsaihan and Enhmaa for their organisational help, and to Gantulga, Hishigjargal, Orulamjab, Miaomiao, Sechenbaatar and Zoljargal for their emotional support while in Mongolia. The research was financed by Stockholm University as my employer and the Sven och Dagmar Salens stiftelse who funded a 7-month stay in Ulaanbaatar with 53270 SEK.

Perfect, evidentially marked Habitual and Prospective2. Section 4 discusses the Habitual on its own and in combination with other complex aspectual forms. Section 5 presents the "Future" participle and its functions in different constructions, while section 6 discusses finite uses of the converb -aad. In section 7, combinations of the particle =l with non-finite aspectual constructions are discussed. Section 8 concludes.

1. Corpus, informants and evaluation

The basic data for this paper are taken from a 589 minutes corpus of spoken Khalkha. In addition, materials from the Internet as well as overheard utterances were occasionally included. The corpus mainly consists of TV programs made up of unscripted, relatively free conversation such as Tanaid honoyo "I want to stay the night at your place" during which the reporter visits families without previous notice, or Derik, a report on crime in Ulaanbaatar3. Almost all materials were recorded in Khalkha-speaking Ulaanbaatar, but discourse participants include a few Khalkha-speaking Oirats. Most speakers seem to be between 35 and 55.

All finite aspectual forms (n = 4244) found within the corpus including the sentence-final particles that follow them were semi-automatically annotated4. I then surveyed them and selected a large

2 In this paper, I follow the convention of writing language-specific categories with capital letters, while general semantic notions or cross-linguistic categories are written with small letters.

3 In extenso: Derik (Benzin: 19 minutes: Do gas stations shortchange petrol; Gazar: 17: a man gets maltreated in a dispute about land ownership; Narantuul: 13: pickpockets at Narantuul market; Yanhan: 24: prostitution in western Ulaanbaatar), Tanaid honoyo (Ganbaatar, bank director: 47, transcribed by Oyuuka; Ariunbold: 41; Badar-Uugan, boxer: 49; Parik Jagaa, show moderator: 56; Cergiin angi, at barracks: 38), Hun ta: Ganbat (20, at the home of a poor cobbler), 16+: Shwees (30, youth-related topics including tattoos), Hoolondoo: Mongold (23, cooking show), Hurdan mor' (32, horse racing sports report), record from a birthday party (171), retelling of a short narrative (2, transcribed by O. Monh-Gerel, Brosig), interview with Enhzul from Tov aimag, conducted by Chuluunceceg for a youth radio station (7).

4 The verbs teg- 'to do like that', ing- 'to do like this' and yaa- 'to do how' were sometimes taken to be lexicalized adverbials and were only annotated

subset of relevant examples for elicitation. Elicitation in Ulaanbaatar was done during overall 267 hours between April and November 2013 with 45 informants with an average age of 25 (17-21: 23, 22-26: 11, 29-40: 9, 55-66: 2). 30 were students and 37 female. Informants were presented with transcribed sentences in their overall context and asked about the semantic difference between similar TAE5 forms in the given environment. The analysis was then done by generalizing from my notes on their answers, both across examples and across markers.

2. Inventory

In this section, the morphemes used in finite predications in Khalkha Mongolian are introduced. Section 2.1 introduces the form inventory from a morpho-syntactic point of view. As formal and semantic classes do not coincide, the latter are discussed separately in section 2.2.

2.1. Inventory of forms

The aspectual system of Khalkha is based on the following finitely usable morphemes: the finite verbal suffixes -n, -laa (-l), -jee (-j) and -v, the participle suffixes -sang, -dag, -aa and -h, and the converbal suffixes -aad, -j and, in writing, -saar and the extremely literary -n. Due to reasons detailed in the respective sections, the use of -v, simple -aa and -h is rather restricted6.

in those cases in which their verbal characteristics were most salient. The highly grammaticalized verb bol- 'to become' was mostly annotated, but inconsistencies related to its use with nouns and adjectives occurred during the initial phase of annotation. The verb bai-, often used as an auxiliary to participles and converbs, was usually not annotated when used with nominals or as locational/possessive verb and was not statistically evaluated. The complementizer verb ge-, originally 'to say', was only included into the statistics when used within the Prospective form.

5 TAME — tense, aspect, modality, evidentiality, thus TAE excludes modality.

6 Cyrillic Mongolian is converted to Latin as follows:

Cyrillic a 6 B r g e 3 H H K M H o e n P c

Latin a b v g d ye j z i i k l m n o ö p r s

Cyrillic T y Y $ x ц H m b H b э M a

Latin t u ü f h c ch sh (not transcribed) ii ' e yu ya

The finite suffixes are used together with several non-finite morpheme combinations. Most consist of the auxiliary bai- preceded by one of the converbs in -j (mostly contracted: -j bai- > -jai- > -jii-), -aad and, in writing, -saar, or by one of the participles in -dag, -sang and, rarely, -h. The auxiliary bol- 'to become' combines with the same participles yielding aspectual and actional meanings, while its combination with the converb -j expresses root possibility or permission. Non-negated -aa is only used finally and then almost only attached to the copula verb bai-. The suffix -ch- (—chih-; -c- before -sang) can in some cases change the actional interpretation of predications. It is mutually exclusive with -jii- within the morphology of one verb and with the copula/locative verb bai- as such, but combines with other aspectual forms.

Additionally, several clitics play a certain role. The clitic =aa can combine with -n, -v and -dag and tends to be mutually exclusive with illocutionary particles. Due to a partial analogy, -jee becomes -j when followed by an illocutionary particle, and past-referring -laa is reduced to -l if followed by the interrogative (c illocutonary) clitic =uu. The lengthenings of -sang and -v to -saang (rare, but more common than -sn=aa) and -aav (more restricted than -v=aa) mirror this use. The modal clitic =iin and its variants can attach to all four aspectual participles. While neither device tends to affect the proposition, their

Most of this reflects common Internet usage, but my transcription diverges by e > o and y > u (instead of e > o/u resp. y > u, avoiding underdifferentiation), ц > c (instead of ts, for convenience) and b > ' (instead of i or zero, preserving both palatalization and syllable structure). Unproblematic spelling mistakes or unconventional spellings in the source (especially in Latin Mongolian examples from the Internet) were usually standardized without note, unless the TAME markers themselves were concerned. Suffixes are cited in their unrounded form without advanced tongue root, e. g. -laa, but are understood to refer to their vocal-harmonic variants (here -loo, -lee, -loo) as well. Most material cited in this paper is drawn from spoken data which is written down as Latin in a very similar fashion. However, historical, synchronically lost segments are not written, so /n/ <na> is transcribed as n, /q/ <n> as ng (or sometimes inconsistently as n, with N being used in unclear cases) and /g/ <ga> as G (though the transcriber was somewhat inconsistent for the latter two). On the other hand, arguably monophthongized diphthongs (but see [Svantesson et al. 2005: 9-11]) are written as diphthongs according to orthographic convention.

distribution and frequency still indicate that they are among the flectional categories of Khalkha verbs. The clitic =l can attach to converbs and non-finite particles, occasionally interacting with aspect. Even interrogative devices (=uu, interrogatively used =iin, the more formal ve) can in some cases change the interpretation of aspect markers. -h is almost restricted to interrogative or modal contexts.

Non-interrogative illocutionary clitics or particles may be compatible or incompatible with the meanings of particular suffixes, but as they do not seem to interfere with the aspectual system in any systematic way, they are not dealt with here7.

Markers of modality such as finite mood forms like -ii for first person intention and participle suffixes like -maar for first person wishes exist on the same level as TAE markers. Some of them combine with aspectual forms, i. e. manai-d ir-j baig-aarai 'be coming to us!' oryav-sang bai-0 'be gone!' However, both adverbials such as zaaval 'certainly' and adjective-based constructions such as -h yos-toi/-gui FUT.P custom-COM/EX.NEG expressing an obligation to do or not to do something are a central part of the Khalkha modality system which might thus be considered by and large distinct from the closely interacting system of means for expressing tense, aspect and evidentiality8.

2.2. Forms and their meanings

Simple -n tends to express potential meaning and simple -aa is restricted to a few peripheral constructions. In complex forms, they are widely used to express diverse present tense meanings with a contrast in evidentiality. The suffix -dag is mostly used to express habituality. The suffix -sang is a factual past marker and -laa and -jee express firsthand and non-firsthand evidential past meaning. -v in its basic past use is restricted to written language or interrogative sentences, while its other spoken language uses are modal. Complex aspectual forms retain evidential distinctions, but implicatures about remoteness are specific to simple forms.

7 An exception is =lee, a shortened form of bilee, that yields past meanings with simple -n. The particles bilee and aj(ee) express evidentiality-related meanings.

8 For a monograph-length treatment of modality in Mongolian (organized according to degree), see [Monh-Amgalan 1998].

Temporal semantics are less important than usage patterns. Certain contexts will result in past interpretations of -dag and -n. Future reference is normal for simple -laa, one of the modal uses of -v is future, and both simple -jee and -sang can exhibit future reference in exceedingly rare contexts.

-aad expresses non-evidential resultative notions. -h is used in contexts that relativize factuality, most often in future questions.

Non-finite complex forms designate aspectual notions such as perfect (-sang bai-), prospectivity (-h ge-j bai-), habituality in the scope of some other notion (-dag bai-), progressivity (-jii-), continuative progressivity (-aad bai-) and potential, sometimes deontic developments (-h bai-). PTCP bol- is used for the attainment of a state where the event indicated by the participle is either intended to be accomplished (-h), a habit (-dag) or already accomplished (-sang), -ch-, while apparently capable of expressing notions such as dissatisfaction, also has more systematic functions such as pinning the future-leaning -laa to the past and turning actionally determined continuative -aad bai- forms into resultatives. Conversely, =l can help avoiding actionally induced resultative interpretations.

Where applicable, =aa can be used as a device that renders utterances more agreeable to the addressee. The particles -h and -aa can connect with =aa via intermediate =iin. The meaning of =iin is subject to ongoing investigation.

3. The evidential forms -n and -aa

The meaning of -n and -aa on their own is quite distinct from their direct and indirect evidentiality function in complex forms. Simple -n (n = 635, including 25 tokens with -ch-n, 8 tokens with -n=uu and one with both) refers to a potential event with occasional realization, be it in the present, past or future, and simple -aa (n = 35) expresses resultative-stative semantics. In complex forms, the semantic contrast is rather between a directly perceived event and an event for which this does not hold true.

3.1. Simple -n

Previous research on -n is not extensive, but mainly focuses on -n as a non-past marker that exhibits some aspectual restrictions when applied to the present, and only Kalchofner [1999] mentions a

limited applicability to past contexts (see [Brosig 2009a: 27-30] for a concise discussion). The most nuanced discussion so far is my own paper from 2009, which I will scrutinize before discussing my new evidence.

In Brosig [2009a], assuming future reference as the basic use, I tried to draw a distinction of present-tense uses that depend on aktionsart and that do not. Independent uses were assumed to be generic uses (such as the movement options of pieces on a chessboard), instructions, in complement clauses when referring to actions that are not temporally located such as in (1), or in citation forms for verbs. Some uses were said to rely on textual conventions such as a historical present used for a sequence of events and a past imperfective use in narratives (novels, jokes, history books). On actionality, I observed that written language sources sometimes use -na9 with stative verbs such as chad- 'to can' and sanagd- 'to seem' in contexts such as (2) where spoken language sources, represented by newspaper interviews with four musicians and one athlete, had -dag instead. With stative predicates such as uzengyad- 'to hate' or med- 'to recognize+know', a timeless use is possible, while the use of -dag would report what the speaker knows about the past and present. Predications of lesser time stability, i. e. accomplishments, activities and those containing position verbs, always receive a future interpretation with the single exception of the irregular verb yav- 'to go'.

(1) dor-dor hel-bel mongol hun dotood yertdnc=dd other-INS say-COND.C Mongolian person inside world=RP busd-aas nuu-na ge-dg-iig med-deg=gui

other-ABL hide-DIR COMP-HAB.P-ACC know-HAB .P=EX .NEG '[It's easy to guess what a Mongolian thinks when you see him. All his emotions stand inscribed into his face.] In other words, a

9 In written language, -sang is written as <san> (word-final /n/ > /q/) and -n as <na> (loss of short vowel phonemes in non-first syllables and insertion of epenthetic vowels only between consonants). It is in principle possible to write <na aa>, but this is rarely done. Instead, readers read <na> as either [n] or [na:] depending on contextual or performative factors. In other words, the written language underdifferentiates.

Mongolian doesn't know how to hide his inner world from others'. [Brosig 2009a: 33]10

(2) surgaal_nomlol-iin chiglel-eer=n' 1. buddiin,

teaching-GEN direction-INS=3POSS Buddhist

2. hristiin, 3. islamiin, 4. busad buyuu "shine

Christian Islamic other or new shashin-ii" chig_barimjaa-tai sum hiid-uud religion-GEN alignment-COM temple monastery-PL ge-j angil-j bolo-h sanagda-na

COMP-C classify-C be.possible-FUT.P seem-DIR 'It seems that from the perspective of their teachings, [Mongolian] temples and monasteries can be classified as 1. Buddhist, 2. Christian, 3. Islamic, and 4. others and those belonging to the "New Religions"'. [Brosig 2009a: 38]

The interpretation of some of this evidence has to be rejected, though. Movement options on a chessboard are a potential development much closer linked to future meaning than a truly generic statement about a class of entities. Secondly, the example given for a chain of subsequent past events most likely referred to an iterated process, so whereas some kind of past imperfective usage cannot be denied, the label "historical present" is not justified.

Apparently, while it is inadequate to ascribe to -n a general potential to express habituality or genericity, restricting its meaning to future and providing a huge list of exceptions does not work either. It is probably futile to try to deduce all subtleties of -n from some overgeneralized Gesamtbedeutung, but a large number of its uses can be characterized as referring to a potential event, be it in the future, present or past. In contrast to other stative markers of potential such as the rare -huic or -maar 'likely to' as used in other dialects, -n does suggest that the potential gets instantiated.

For the future domain, one can make a heuristic division into an intentional future for events under the control of the speaker, a predictive future for events where such control is lacking, a scheduled future ([According to the timetable] the train leaves at noon) and a preparative

10 Most interlinearization glosses are the same as in the Leipzig glossing rules. Abbreviations that differ are given at the end of the paper. Note that 1pl.gen with the stem man- is exclusive.

future [Dahl 2000: 309-318] that can also be grammaticalized into prospective aspect. In the corpus, numerous examples for events that the speaker intends to undertake such as (3) and a fair number of scheduled events such as (4) can be found. There are some arguable instances of predictions, but (5), drawn from a written source yet perfectly possible spoken usage, is more illustrative. As Khalkha has a specialized Prospective -h ge-j bai-, simple -n does not have this usage. Although future events are potential and understood as such, none of the examples below introduces an overt degree of epistemic incertainty on the part of the speaker (which could be added through modal particles or adverbials).

(3) odoo bugd-eer=ee nege-n ail-d zor '-j now all-INS=RP one-AT family-DAT aim.for-C ochi-n

go.to-DIR

'[Here on TV 25, the TV program Tanaid honii is beginning.] We will now all go to one family'.

(4) manai huug-iin naiz german-aas ocho-N 1PL.GEN boy-GEN friend Germany-ABL go.to-DIR

'The friend of our son will come here from Germany. [Let that student stay at your home! ]'

(5) OHU-iin ediin_zasg-iin dsolt 2014-2015 Russian.Federation-GEN economy-GEN growth 2014-2015 on-d buur-na11

year-DAT decrease-DIR

'The economic growth of the Russian Federation will decrease in 2014-2015'.

Instances of present potential with dynamic predications may closely resemble habitual predications. However, -n turns out to refer not to a habitual situation as such, but to an inherent consequence of the existing situation:

11 mongol24.mn/400, 2013-04-15, retrieved 2013-12-18. Headline of an article.

(6) bas manai mongol-chuud yaag-aad bid nar engeed also 1PL.GEN Mongolian-PL.H do.what-C 1PL PL so ovl=n idesh id-deg vee ge-heer bid nar=chin winter=AT food eat-HAB.P Q say-CNS.C 1PL PL=STC huiten-d ih daar-dag energe alda-N manai cold-DAT much feel.cold-HAB.P energy lose-DIR 1PL.GEN mongol oron=chin huiten oron-d

Mongolian country=STC cold country-DAT

tooco-gdo-N hun=chin ald-sang energe kloro

count.as-PASS-DIR person=STC lose-PRF.P energy kalorie

nogoo yug-aar=aa ooh tos-oor=oo noho-i

other what-INS=RP solid.fat fat-INS=RP mend-C

av-dag

take-HAB.P

'If one wants to say why we Mongolians... why we eat winter food stuff like this: we suffer a lot from cold. (Thereby,) we lose energy. Our Mongolia is counted among the cold countries. People regain their lost calories... through... what was it... fat'.

(7) haayaa eng-eed choloo-t cag=aa onGruule-n=dee sometimes do.so-C free-COM time=RP spend-DIR=IP dur-tai yum=aa uz-N, sonin hevell=ee liking-COM thing=RP see-DIR newspaper publication=RP unshi-N, hoorond=oo shatar togloom=oo toglo-N, read-DIR between=RP chess game=RP play-DIR bai-jii-dag _ gazar=daa

COP-PROG-HAB.P place=IP

'[Well, now this is our hall for leisure and cultural activity. Our soldiers spend their free time here.] They will from time to time spend their free time like this. Watch what they like, read the press, play chess with each other. A place to hang out'.

(8) teg-ee=l odoo... ge-e=l ing-ee=l aimaar yum do.so-C=FOC now12 say-C=FOC do.so-C=FOC awful thing

12 While odoo is always glossed as 'now', it is important to note that it is not only used as a time adverbial, but also as a discourse structuring device or even as a device for the speaker to keep her turn and come up with a proper wording.

yav-jai-had=chin=l_ bii ydron odoo zov=l

go-PROG-C.when=STC=FOC 1SG in.general now right=FOC yum hii-jaiG-aa baih=daa ge-j bod-n thing do -PROG-—DIR MP=IP COMP-C think-DIR 'Then, now, [when those police folks] keep on saying, [what, things such that Ganbaatar is a criminal], I feel awful and wonder whether I am really doing the right thing. [But when I then go forth and come to the Suhbaatar square, and the people and the elderly kiss me and then encourage me, then I feel happy (bayarl-aa=lyav-jii-n feel.happy-C=FOC go-PROG-DIR) again]'.

In (6), the event marked with -n is a consequence of the first event marked with -dag. Informants say that if ald-dag were used instead, there would not be any causal connection between the two events13. The events enumerated in (7) are exemplary options rather than frequently recurring events that could be marked with -dag. In (8), the conditions are not specified in the main clause as in (6), but in the subordinate clauses.

A number of stative, sometimes individual-level predications can be conceived of as referring to a present potential:

(9) naidvar-tai, bid nar ter-iig batalgaa-tai hel-j reliability-COM 1PL PL D.DIST-ACC guarantee-COM say-C chada-n

can-DIR

[Being in the course of controlling whether the gas station at which B works tanks up as much petrol as they say and having explained this process to the listeners, the reporter A asks B a tag question (with ..., te?) whether their gas station is reliable. B answers:] 'It is reliable, we can say this with confidence'.

(10) bii yuu ge-hleer huushuur hii-N _ huushuur huushuur 1SG what say-CNS.C huushuur make-DIR

huushuur huushuur ge-heer med-N.

huushuur say-CNS.C come.to.know+know-DIR 'What will I prepare? I'll prepare huushuur. Huushuur, huushuur, huushuur. If you say huushuur, everybody'll recognize it'.

13 Example (7) in [Brosig 2009a: 32] can be reinterpreted along similar

lines.

In (9), chad(a)-n indicates a potential at the time of speaking that, due to the current ongoing investigation, does not exclude the past. The form -dag that generalizes over several points in time would therefore be infelicitous, as would be -jii-n which requires the results of the investigation to be available already. In (10), even the verb med-'to recognize+know' is interpreted in an inceptive way due to the conditional sentence it is used in. -dag would avoid this and suggest a permanent awareness instead.

The past potential use resembles the "imperfective" use that I assumed before. The corpus contains three passages with 3 to 5 past potential uses of -n each. Two of these and a literary example for comparison are cited in a slightly shortened form below:

(11) eej buruu hereg hii-vel zod-ii=shd (...) mother wrong action do-COND.C hit-DIR=IP yamarsaindaa manai aav ih omooro-N (...) at.least 1PL.GEN father much defend-DIR manai aav teg-deg bai-sang 1PL.GEN father do.like.that-HAB.P COP-PRF.P

'Mother would hit me if I did something wrong. At least, my father would protect me a lot. (...) My father used to do so'.

(12) tegeed minii aav arih ih uu-dag bai-sang, then 1SG.GEN father vodka much drink-HAB.P COP-PRF.P tegeed aav=maan' ter hecuu niigm-iing hund then father=1PL.POSS D.DIST difficult society-GEN heavy berh temcel dund bas zarimdaa uu-maar

tough fight amid also sometimes drink-WISH.P sana-gd-dag bai-s=iim bailguu esvel hung think-PASS-HAB.P COP-PRF.P=MC IP or person

am 'tan-d ing-eed zarGald-aad yal-uul-sn-ii daraa

being-DAT do.so-C litigate-C win-CAUS-PRF.P-GEN after nogoo hung=chin bayarlaa ge-ed arih ogo-n ih other person=STC thank.you COMP-C vodka give-DIR much telviiz araadio zas-dag bai-sang, bayarlaa ge-ed TV radio repair-HAB.P COP-PRF.P thank.you COMP-C

arih dgd-n vodka give-DIR

'Then my father used to drink vodka a lot. Within all that difficult social struggles, our father probably sometimes felt the urge to drink. Or, after having helped somebody to win a trial, that person would give vodka to say thank you. He used to fix lots of TV or radio sets, they would give vodka to say thank you'.

(13) ...dur_tavi-laa ted heseg hugacaan-d chimee=gui

orgasm-1H.PST D.DIST.PL part period-DAT sound-EX.NEG bol-j biye biye-s-iig chagna-h met am'sgaa become-C body body-PL-ACC listen-FUT.P like breath dara-n hevte-hed zoloog-iin doog-uur noit press-C lie-C.when NAME-GEN under-PROS humidity orgi-n aarcag=n' zadar-san met aimshigtai gush.forth-C pelvis=3POSS break.up-PRF.P like awful hdnduurle-ne ache-DIR

'And he came. When they fell silent for a while and lay there normalizing their breath as if listening to one another, Zoloo felt how wet she was beneath, and her pelvis was hurting as if broken'. [Narantuyaa 2006: 10]

Many of the example clauses including the one containing zod-ii=shd (-n regularly taking the form -ii before shd) and both instances of dgd-n overtly specify a condition, and for others such a condition is inferred by the informants. While the context sometimes indicates that the condition cannot hold any longer, e. g. due to the death of the participant, informants would not a priori exclude zodiishd from the present: the mother might still hit her son or any child if it behaved naughty, but such behavior on the part of an adult son is unlikely, so the preconditions for the consequence have by and large ceased to exist. It is not clear whether the event actually occurred often or not. -dag bai-sang, in contrast, would indicate that the event used to occur frequently, but does not anymore. Actual past imperfective uses such as in (13) with reference to a past temporary state do not seem to be present in the spoken corpus.

There are a few uses hard to account for that tend to be noticed by younger speakers. For example, toocogdon in (6) is given a potential

interpretation (e. g. the speaker was not sure of her words), but such uses are at the same time associated with the speech of elderly people. (14), uttered by a rural man aged 55, probably does not have a reasonable potential interpretation. In such cases, informants often resorted to explanations based on a reduced frequency, which would harmonize with the adverbial of (14). But as -n(=aa) in other dialects is used in habitual contexts more actively and the population of Ulaanbaatar comes from diverse dialect areas, some speakers probably retain a habitual use not shared by young speakers born and raised in this city.

(14) aa jims av-N jims bol mongol-d baG INTERJ fruit buy-DIR fruit TOP Mongolia-DAT small shiigua jaahan tari-N

water.melon a.bit plant-DIR

'[BB: Then you have to import fruits, don't you? A:] 'Well, we buy fruits. There aren't many fruits in Mongolia. We do plant some water melons'.

-n is also attested with another kind of use that might be characterized as atemporal and abstract. An example where any future or present potential meaning can be ruled out is (15) below. Here, little is said about the period of knowing. -dag, on the other hand, would present the knowledge as a fact observable over a certain period. Two other verbs capable of referring to the present with -n are the locational/ possessive/copula verb bai- 'to be at' and the morpho-semantically irregular verb yav- 'to go' exemplified in (16).

(15) yoron ah=n duug-iin-h=ee in.general elder.brother=3POSS youger.sibling-GEN-NOM=RP nevtruulg-iig med-N

program-ACC come.to .know+know-DIR 'In general, I know your TV show'.

(16) chii haan yava-n? (overheard) 2SG where go-DIR

'Where are you going?'/'Where are you?'

There are a few examples where -n in the scope of quotative ge-almost seems to be used with abstract, non-localized actions akin to (1). However, elicitation indicates that the uses are not entirely abstract

and that e. g. duusa-n in (17) refers to a concrete, foreseeable future during which the house will not finish. The more abstract -h could express that the house will actually never get finished. In this example, informants would only accept -h if a transitive verb form was used instead. In (18), speakers tend to associate the act of birth with the experience of a concrete speaker; -h would indicate a general statement without any reference to personal experience.

(17) ydrddsdd baishing duusa-n ge-j bai-d=guu=m

overall building end-DIR COMP-C COP-HAB .P=EX .NEG=MC shig bai-n shdee like COP-DIR IP

'[We moved into this house in 2000 and made repairs. Up to now, it hasn't ended. Now this...] it almost seems like there is no such thing as for the building to get finished'.

(18) tegeed bas emegtei hung tdrd-n

then also female person give.birth/be.born-DIR

ge-deg bol hamg-iin hecuu zuil

COMP-HAB.P TOP all-GEN difficult matter

'[And my mother also went to Heaven (ddd bol-jii-song uphill

become-PROG-PRF.P) because of giving birth.] So a woman

giving birth is also a most difficult thing'.

A particular clear case of abstract language usage is metalinguistic reference to verbs as words. In dictionaries, the lexical form of verbs is conventionally displayed as -h. An utterance from a newspaper interview and an overheard utterance show the use of -n instead, but both examples are problematic. First, the speaker of (19), yokozuna Dagvadorj, is an Oirat, and the speaker of (20), the linguist Sechenbaatar, is a native of Chakhar, which is very close to Khalkha, but still not identical. More importantly, some informants assumed actual reference for (19) by which an actual action (azargalsan shdee) is taken up again. Such an interpretation is not available for (20), thus 4 of 5 informants preferred -h instead, indicating that -n would be deictic:

(19) R: Yaagaad. Azargal-san shdee?

why lord-PRF.P IP

D: Ugui=ee, azargal-na ge-deg=chin' muuhai ug14 EX.NEG=POL lord-DIR COMP-HAB.P=STC bad word [D: 'Into this much-admired Sumo world, the son of a certain Dolgorsuren from the remote Mongolian steppes entered and wrestled sumo and...' R: '... and lorded15 it?' D: 'You cannot say so.'] R: 'Why, you did lord it!' D: 'No, "to lord" is a bad word'.

(20) nudra-n ge-deg=chin uil ug! (overheard) nudge-DIR COMP-HAB.P=STC action word

'The verb[al equivalent of nudraG 'fist'] is nudran'16.

Next to -n, there are the two forms -n=aa and -n=uu. =aa is probably to be understood as a device that adds politeness, but does not have any aspectual meaning of its own17.Yet, -n=aa or -n paired with other illocutionary particles does seem to be somewhat more common in strictly present habitual contexts. Still, the use in (21) below can be deduced from the potential semantics of -n, as the main difference to the more common form -dag lies in the reluctance of the speaker to generalize over all Mongolians. Similarly, the statements in (22) and (23) are perceived as somewhat tentative in contrast to -dag, which could e. g. be used in (23) to remind the addressee of a regular process:

(21) en-iig manai mongol-chuud bas ih D.PROX-ACC 1PL.GEN Mongolian-PL.H also much uu-n=aa bid nar bas ih hergel-deg suul-iin drink-DIR=POL 1PL PL also much use-HAB.P recent-GEN uy-d... hudaldaan-d gar-aad baig-aa

time-DAT sale-DAT exit-C COP-—DIR 'Our Mongolians drink this a lot as well. We [ourselves] use it a lot, too. It is recently entering the sale [as something that is good against tiredness]'.

14 Asashyooryuu D. Dagvadoij (interviewer B. Ganchimeg): Eej aav, nutag us, eh oron eleg zurhend min' l egshij baidag yum shuu dee. Zuunii shuudan, 2008-10-14. Even online at http://sport.news.mn/content/418.shtml and elsewhere, retrieved 2014-01-15.

15 azargal-, more literally 'to act as or as if one was the herd stallion'.

16 Note that out of context, this example would be interpreted as ' [The word] nudran is a verb!' with focus on this particular word class in contrast to others.

17 Proper eliciation was done for -jii-n=aa discussed in 3.3.1, but its results appear to be valid for simple -n=aa as well.

(22) A: taa heden hel-eer yar'-dg=iin?

2SG.HON how.many language-INS speak-HAB.P=MC J: mm oros angl'-aar bol _ sain yar '-n=aa.

Russian English-INS TOP good speak-DIR=POL A: 'How many languages do you speak?' J: 'Hmm, Russian and English I speak quite well'.

(23) zaa sar-d=aa bol yôrôn neg tav-aas _ naiman well month-DAT=RP TOP in.general one five-ABL eight udaa bol hon-ii=shuu

time TOP spend.night-DIR=IP

[A: 'How many times do you stay the night here?']

B: 'Well, I stay overnight here about five to eight times in a

month'.

Habitual nuances in a modally different context arise in examples where the same verb stem is used twice in a row and the first form receives a long vowel. Here, a confirmative prediction is made due to fundamental readiness/intention/probability/naturalness of a course of events that, while pointing to a future event, can in fitting circumstances also implicate a habit:

(24) A: taa yôrôn her uu-h=uu?

2SG.HON in.general how.much drink-FUT.P=Q J: uu-n=aa uu-n... zuu tat-ii=shuu

drink-DIR=POL drink-DIR 100 pull-DIR=IP A: 'How much do you drink, in general?' J: 'Well, I drink alright. I empty 100 gram at a draught'.

The statement in (24) is interpreted by all five informants as pertaining to habit or ability, while uun uun would rather point to the readiness of the speaker to do so in general and the reversed uun uunaa would imply a wish to do so now (and thus would not fit the context). Similarly, bolnoo bolon indicates a friendly, positive answer to a request for permission, whereas bolon bolon would rather actively invite the person to do so.

The combination of -n with the polar interrogative clitic =uu is used for polite, official imperatives. Its modal nuances vis-à-vis mood forms such as the somewhat similar -aarai (which can presume a positive answer, too, but is also used for advertising) cannot be treated

in any detail here. For the discussion at hand, it is sufficient to note that -n=uu cannot be used for informational questions about the future as the semantics of its two morphemes would suggest. Informants occasionally noted that such a usage would sound like Southern Mongolian. This contrasts with the use of -n in complex aspectual forms where regular information-seeking questions are attested for the Progressive, Continuative and Prospective (n = 49, 10 and 1, respectively; cf. (60) below).

(25) bichig_barimt=aa shalg-uul-n=uu?

documents=RP examine-CAUS-DIR=Q

'[Well, greetings. I am lance corporal (of the police) Enkhbold

of Bayangol district.] Would you let me examine your papers?'

3.2. Simple -aa

-aa is described by Song [1997] as marking either an imperfective or a resultative situation. For the former, Song cites the verb forms based on the stems bai-, yav- 'to go' and hus- 'to wish' [Song 1997: 245-247], and this is basically confirmed by my data on bai-, yav- and taar-(cf. (27) and (29) below). For example, he introduces the following sentence:

(26) nad neg huseh yum baig-aa yumsan18

1SG.DAT one wish-FUT.P thing COP-—DIR MP

'I have a hope, but...'. [Song: 1997: 246]

He then correctly explains that baigaa indicates that "the speaker's hope is not sudden, but has rather been harboured in the speaker's mind for a while and is still cherished in his mind"19. Song [1997: 255-257, 264, 270] assumes that such a present tense continuative interpretation is said to be more likely with stative verbs or with those

18 Note that even though the consonant -g- in baigaa is epenthetic, it is written here as part of the stem. This is in contrast to the "floating" /n of some stems that is historically and in a modified form even synchronically conditioned lexically.

19 The example is unnecessarily complex, though, due to the presence of a modal particle expressing wishing and the sequence -h yum baigaa which could occur within a single predicate (with possible contraction to yav-h=iim bai-n) if the dative nad did not indicate a possessive construction with a proper noun yum as the possessum instead.

few verbs that can be used with -aa in absolute-final position such as bai-, yav- and suu- 'to sit' and that both past and future time adverbials cannot combine with these. I will analyze baigaa as indirect present tense evidential marker in section 3.3, and suu- as an inchoative verb (in the sense of [Breu 2005: 48], i. e. a progressive that can either refer to the phase before and after a terminal point) should probably yield a resultative meaning. In (27), the verb yav- with its opaque aktionsart seems to pattern with baigaa: the speaker has a sleeping bag with him and had it with him even earlier (3 of 4 informants), so no result is referred:

(27) yag en Gazar meshok=oo delg-ee=l

exact D.PROX place sleeping.bag=RP spread.out-C=FOC unt-ch-ii, meshok-toiG=oo yav-aa

sleep-BOU-VOL sleeping.bag-COM=RP go-—DIR '[Today, I won't cause you much trouble,] I can spread out my sleeping bag right here and sleep, I've set out taking my sleeping bag with me'.

However, the forms baigaa and yavaa made up 83% + 10% of the overall usage of -aa in Song's [1997: 274] textbook corpus, and in the spoken corpus (n = 557) they even account for 93% + 4% of all finitely used -aa. In contrast, the stem bol- 'to become' accounts for 2% and all other stems for merely 1%. It is thus conceivable that baigaa and yavaa might require a special analysis. Song [1997: 264-268, 278] observed that -aa often co-occurs with modal particles that express probability (e. g. biz, baih, boluu) including interrogative markers, and noted that some of his informants even accepted bare -aa on stems such as bich- 'to write' with a slightly confident probability interpretation. On this basis, he linked -aa itself to a probability interpretation and also predicted that it could not combine with modal particles that denote speaker confidence such as shuu and yum. If baigaa and yavaa are excluded, an analysis along the lines of weak probability or reduced factuality might indeed be feasible, although this does not allow for making generalizations about the distribution of non-modal, illocutionary devices such as shuu. The aspectual meaning seems to depend on aktionsart, with telic predications referring to resultant states:

(28) joohon davs hii-N_ davs hiig-eed jaahan tyamyaa a.little salt put-DIR salt do-C a.little thyme hiig-eed... davs temyaan ge-e=l bol-oo put-C salt tympan COMP-C=FOC become-—DIR

'... and then one adds a little salt. One adds salt and a bit of tympan. Salt and tympan, then it'd be done'.

(29) chii bid hoyor en horvooG-oos tav' yumuu jarang 2SG 1PL two D.PROX world-ABL 50 or 60 jil-iin daraa yav-j=l taar-aa

year-GEN after go-C=FOC fit-—DIR

'The two of us will most probably depart from this world after fifty or sixty years' .

(30) chii... ugaasaa chii bid hoyor turuun hel-ee biz=dee, 2SG basically 2SG 1PL two initially say-—DIR IP=IP bii hon-j chad-kuu gej!

1SG stay.night-C can-FUT.P.EX.NEG COMP

[Y: 'It's not my boyfriend, just a friend of ours. We just planned

to go for a drink together.' O: 'Are you waiting for him?' Y: 'Yeah,

I told him. I'll be waiting for him.' O: 'If somebody is waiting

for you, can you stay the night?'] Y: 'You... right from the start,

you and I have talked about that, haven't we. That I cannot stay

overnight'.

(31) gehdee joohon yuman-d=l setgle=n gomdo-n

but small thing-DAT=FOC mind=3POSS take.offence-DIR

shuu dee teren-d ner hele-h=guu

IP IP D.DIST-DAT name say-FUT.P=EX.NEG

onGr-hod=l duus-aa shuu dee

pass.by-C.when=FOC end-—DIR IP IP

'But she gets offended by very small things. If her name isn't

mentioned, everything's already over for her'.

In (28), bol-oo refers to a general result that has not yet come about (in contrast to certain future -n) and thus is subject to partial guesswork. In (29), -aa seems to convey that the event in question would be fitting if it should occur (with taar- receiving a deontic or teleological [Narrog 2012: 8] interpretation and -aa expressing a lowered degree of epistemic modality). -n would rather presume the future event as certain and taar- would then refer to a speaker-external

standard of evaluation. (30) is taken by 3 of 4 informants as more aggressive or angry than the more neutral, factual -sang biz dee. Conceivably, a rhetorical question might be sharper if the presumed answer is kept more irreal and not stated as factual (by using -sang) outrightly. (31) focuses on the result of a possible future event, while -n would refer to the entire event.

Song [1997: 279-281], based on his modal analysis, argues that -aa cannot normally co-occur with the first and second person, yet his examples all contain the illocutionary particle biz which might cause incompabilities of its own. For questions with their switch of perspective, first and second person uses can easily enough be found on the Internet, possibly signalling disbelief on the part of the speaker20:

(32) chi ir-ee=yuu?21 2SG come-—DIR=Q

[Mother:] 'You have come?' [I: Why, are you angry?]

(33) yuu? manai angi dandaa eregtei huuhed-tei ge-j what 1PL.GEN class always male child-COM COMP-C chamd bi hel-ee=vuu?22

2SG.DAT 1SG say-—DIR=Q

'What! "Have "I" told you that there are only male students in our class? [You're kidding]'.

Given the rareness of modal -aa in texts and speech, providing a proper analysis would require a substantially larger corpus.

3.3. -n and -aa at the end of complex forms In complex forms, -n and -aa differ with respect to whether the evidence for an event is based on what the speaker currently perceives or whether it is based on anything else, including what has been perceived earlier, is presumed from general reasoning or is held as general knowledge. The evidentiality systems listed by Aikhenvald

20 It seemed to me that regular verb stems in -aa followed by interrogative or other particles are rather frequent in translated Korean soap operas. I did not analyze such examples, but the notion of disbelief is not exactly foreign to this kind of TV program.

21 http://teenblaqkiss.wordpress.com/2013/04/22/nothing-lasts-forever -napT5, retrieved 2013-12-20. From a possibly fictive dialogue posted on a blog.

22 http://news.gogo.mn/joke/?j=533501&p=18, retrieved 2013-12-20.

[2004: 65] do not allow for distinguishing current and previous perception. However, de Haan [2001] suggested that evidentiality systems can make two different basic distinctions: either between firsthand and non-firsthand, rendering all information not perceived directly as non-firsthand, or between direct vs. indirect, grouping directly perceived evidence after the fact with visual and auditory evidence in contrast to indirect hearsay. This category of direct evidence (visual, auditory, evidence after the fact) corresponds very well to the function of -n, while indirect evidence as marked by -aa does not only include hearsay, but even what Aikhenvald [2004: 2-3, 52, passim] calls assumption, i. e. reasoning from non-sensory evidence (which de Haan [2001: 201-202] actually includes into inference and thus counts as direct information), common sense or general knowledge, plus earlier perception. In contrast to a system such as Duna spoken in the New Guinean highlands [San Roque, Loughnane 2012: 125-127] where a distinction between current and previous perception can be made both for sensory evidence and the evidence used for inference, the Khalkha Perfect form in -aa, first, does not infer any event, but only specifies how the speaker perceives the (current or resultant) state and, second, does not claim any previously perceived direct evidence at all i. e. if the inference is based on general reasoning.

As far as I can discern, the difference between -n and -aa in complex forms has not been properly discussed in the literature so far. Cinggeltei [1959: 34-35, 40, 43] does not recognize an independent meaning of the forms -j baig-aa, -aad baig-aa and -sang baig-aa, but explains them as rare variants of -j bai-h etc. This is apparently so because -aa does not play a role in Eastern Mongolian dialects such as Kharchin (cf. [Caodaobateer 2007]) that guided Cinggeltei's native intuition. Song [1997: 277-278], relying on informant judgments on an unspecified number of sentences, describes -j baig-aa as signalling a lack of commitment on the part of the speaker, and -j bai-na as neutral in this respect. Onorbayan [2004: 189-190], without any further comment, characterizes the difference between the two Present Progressive forms -j bai-n and -j baig-aa as il 'open, visible' vs. dald 'concealed'. None of the latter two discusses any other related forms. However, there are several other forms (-dag (-diin) will only be discussed in 4.3):

Table 1. Complex aspectual forms with one copula in the corpus

-n -aa (-aan) -dag (-diin)

Progressive -jii- I-jai- 307 53 9

-j bai- 120 53 2

Continuative -aad bai- 77 41 14

Perfect -sang bai- 52 50 5

Prospective -h gejIgeed bai- 21 24 0

Habitual -dag bai- 1 7 n. a.

So while the Progressive is clearly most frequent, both the Continuative and the Perfect readily combine with -n and -aa, as does the somewhat less frequent Prospective. Even the Habitual does occur (though not *-dag bai-dag).

3.3.1. The Present Progressive -jii-n/aa. The Present Progressive takes the forms -jii-n and -jaig-aa and, in more formal contexts, -j bai-n and -j baig-aa. Except for their evidential meanings, it is assumed that both forms exhibit similar properties and can be discussed in parallel. Given the comparatively small number of forms in -aa, this is only a conjecture.

3.3.1.1. The following examples give a glimpse at the direct vs. indirect difference between both kinds of Progressive forms:

(34) lion ter hoyor nadad tusla-N ge-seng son 'giN NAME D.DIST two 1SG.DAT help-DIR COMP-PRF.P onion hirch-üül-jaig-aa tegeed ter hoorond=n bii cut-CAUS-PROG-—DIR do.so-C D.DIST between=3POSS 1SG maijiron=oo yanzal-jii-n

marjoram=RP put.into.order-PROG-DIR

'She and Lion said that they'd help me. They're cutting the onions. Meanwhile, I'm taking care of the marjoram'.

(35) horin tav-duGaar suvag telviiz tanai-d twenty five-ORD channel TV 2PL.GEN-DAT hon-ii nevtrüüleg ehel-jii-n stay.night-VOL program begin-PROG-DIR

'Here on TV 25, the TV program Tanaidhonii is beginning'.

(36) bii ajl-aas-aa gar-chii-n 1SG work-ABL-RP exit-PROG-DIR

'I'm leaving work [just for today/permanently]'.

In (34), the speaker uses -jiin for the event he is just busy with and -jaigaa for a similar activity of his helpers. Informants agree on the fact that he cannot see them at the time of speaking. In (35), the anchor is announcing that his show is starting now, using eheljiin. Informants would only be fine with eheljaigaa in a context where the speaker states that the TV show is beginning while not watching it herself. (36) was presented to informants out of context. It has two different interpretations: it can relate to the process of leaving as such (e. g. having left one's own office, but not yet the building) including its pre-phase (being in the course of putting one's things into order in preparation for leaving), or the speaker can refer to herself as (caught) in the process of quitting her job. In contrast, garchaigaa could only envisage the process of quitting as underway during a longer period and without the speaker's immediate involvement at the time of speaking.

If directly accessible information is coded by -n, one would not expect the use of -aa with first person subjects, and one might expect -aa with psych verbs for non-first-person actors. This only holds true to some extent:

(37) bid buhen bol nuuc durang-aar en durs-iig 1PL all TOP secret telescope-INS D.PROX form-ACC av-chaig-aa

take-PROG-—DIR

'We are recording this video with a hidden cam from afar'.

(38) en bol neleen tom doromjlol ge-j bii D.PROX TOP rather big humiliation COMP-C 1SG oiloG-joig-oo shd comprehend+understand-PROG-—DIR IP

[The speaker narrates how Chinese citizens committed a crime in Mongolia and put it on YouTube. He then argues that they did so to flaunt what they can do unpunished.] 'As I understand it, this is a very great humiliation'.

(39) telviz ge-deg yoroosoo chii ddr=dd TV COMP-HAB.P generally 2SG self-RP med-jaig-aa sh[d] reklam come.to .know+know-PROG-—DIR IP advertising surtchilgag-aar

propaganda-INS

[A: 'For example, you are doing Centauer. But where do you get your money from?' J:] 'As for TV, you know the TV business yourself. From advertising'.

In (37), the anchor describes what is happening on a video, otherwise using past forms. The recorders are actually not visible to him, and he is not actually recording anything at the time of speaking, so -aa is appropriate. In (38), the speaker uses a verb of cognition about herself. -aa indicates that she has already formed her understanding about this matter a while ago. Indeed, the use of -jiin would implicate that she is just evaluating new information or has just made up her mind. Thus, the difference between direct and indirect perception is very similar to a difference between recently acquired and old information, but as (36) indicates, such a distinction is not basic. In (39), the speaker mentions what his addressee knows using medjaigaa. medjiin could have been used if the speaker had "seen" from the addressee's action or countenance that the addressee holds such knowledge, an interpretation that informants reject for this particular context.

In a number of instances, -jaigaa may have been chosen in order to signal that the speaker is not very actively engaging in the event:

(40) ... talaar nevtruulg=ee beld-eed ehel-ceng

concerning broadcast=RP prepare-C begin-BOU.PRF.P baig-aa en=maan bol todorhoi hemjeen-ii

COP-—DIR D.PROX=1PL.POSS TOP certain measure-GEN cag huGcaa shaarda-gd-dag uchraas jaahan udaan time period require-PASS-HAB.P because a.bit long hulee-lge-j baig-aa wait-CAUS-C COP-—DIR

'[For the following broadcasts... we] have started preparing our broadcast on corruption among officials. As this requires a certain amount of time, it will have you wait for some time'.

(41) A: taa yaahaaraa en-iig shoN bol-song

2SG.HON why(RHET) D.PROX-ACC night become-PRF.P hoiN ir-j uze-dg=iin? after come-C see-HAB.P=MC B: gue harm odr-oos hoish sendiich-eed EX.NEG but day-ABL after loosen.soil-C ge-heer=n shoN yamar uls-uud bai-n

say-CNS.C=3POSS night what people-PL COP-DIR ge-ed ir-eed uz-jaig-aa COMP-C come-C watch-PROG-—DIR A: 'How come you've come here to watch it after night has fallen?' B: 'No, but since the afternoon I've been told that somebody is digging here, so I came and am looking at what kind of people are around here during the night.'

One informant commented on (40) that as the material is not ready yet, they have to keep the audience waiting (or, as the impersonal translation suggests, the situation itself will keep them waiting), while -jiin would indicate that they are doing so on purpose/without need. Another informant framed this contrast as between apologetic and nonchalant. (41) features a subject that executes an action for a quite limited period of time. Two informants commented that the speaker has not actually successfully observed anything. Contextually, it makes sense for B, who has just bullied and threatened someone, to pretend that he has not been doing anything at all.

3.3.1.2. On the aspectual level, a Progressive has to contrast with some other kind of imperfective marker. In the case of Khalkha, this is mainly -dag. One respect in which -jii-n/aa and -dag differ is that -dag cannot co-occur with delimiting time adverbials. So (42) would be acceptable with both forms of the Present Progressive, but not with -dag. On the other hand, sentences like (43) mostly based on -aa are perceived as not limited in scope, and even if they are explicitly contrasted with minimal pairs in -dag, some informants will only report lower certainty (due to indirect -aa) and frequency (as -dag always requires repetition) instead of overall longer duration.To some degree, this vindicates Song [1997: 380-381] and Hashimoto [1995a] who propose a habitual (and the latter also an iterative) meaning variant for the Mongolian Progressive on the basis of less convincing examples.

In (44), it seems that -jiin is used instead of the Habitual merely in order to be more illustrative.

(42) en tucen-d=ee neg arvan jil=l D.PROX small.shop-DAT=RP one ten year=FOC am'dar-chii-n

live-PROG-DIR

'Now I've been living in this kiosk for just ten years'.

(43) hudlaa yar'-aad bai-[n] _ german-d dor humuus untruth say-C COP-DIR Germany-DAT other people zonddd arih uu-j baig-aa

a.lot vodka drink-C COP-—DIR

'He's talking nonsense. Other people in Germany drink spirits a lot'.

(44) binyaamiin, en byasalGal Gurvang hesg-ees NAME D.PROX meditation three part-ABL burde-j bai-n (constructed by informant) consist-C COP-DIR

'Benjamin, this meditation consists of three parts'.

Present Progressive forms of state predicates by and large designate temporal states (cf. [Song 1997: 373-374]):

(45) aa, ih Goy bol-j ter huur-chaiG-aad INTERJ very nice become-2H.PST D.DIST roast-PROG-C hii-seng tsuu ih Goy amtagd-jii-n

put-PRF.P vinegar very nice taste-PROG-DIR

'Aa, it has become very nice. That vinegar that you added after

you had roasted it tastes delicious'.

(46) odoo en gerbul saihang=l am'dar-chii-n now D.PROX family nice=FOC live-PROG-DIR '[But anyway,] now this family is living a good life'.

(47) odoo teg-eed nadaar hii-lge-h yum bai-val now do.so-C 1SG.INS do-CAUS-FUT.P thing exist-COND.C bii huvcas ugaa-j bol-jii-n, hool hii-j 1SG clothes wash-C be.possible-PROG-DIR meal make-C

bol-jii-n, ger cevrel-j bol-jii-n.

be.possible-PROG-DIR house clean-C be.possible-PROG-DIR '[Today I am just like a member of your family.] Now if there are things you want me to do, I can wash clothes, I can prepare a meal, I can clean the house'.

In (45), the speaker simply reports his current perception. Using -dag instead would refer to a general property. -jaigaa would dissociate the event from any immediate experience, i. e. the speaker has tried it before knowing how it tastes. You would expect this form in a commercial. (46) is not as simple. With -jiin, informants have a strong tendency to contrast the current pleasant situation of this family with some unpleasant former situation. With -jaigaa, this pleasant life is taken to be the well-established knowledge of the speaker, thus informants no longer take it as delimited by a previously different situation, though the situation does not necessarily have to be taken as permanent as with -dag. The latter would further imply some kind of objective, quantifiable evidence. The verb bol- in (47) expresses root possibility (opportunity, permission, general situation possibility and situation permissibility, cf. [Depraetere 2011]). Usually, it tends to combine with -n, as a future potential use fits the notion of general situation possibility very well. Here, however, the speaker wants to emphasize both his readiness and the imminence of the possibility, thus choosing -jiin instead.

Hashimoto [1995a] also claims a near future and a resultative meaning for -jiin, but both proposals as well as his translations of the examples associated with them must be rejected23:

(48) odoo galt teregn-ii bilyet ava-h-aar yav-j bai-na now train-GEN ticket buy-FUT.P-INS go-C COP-DIR '[I']m going to buy a train ticket now'. [Hashimoto 1995a]

(49) namaig baraan-ii delguur-t ochi-hod eregtei 1SG.ACC goods-GEN shop-DAT go.to-C.when male hun-ii malgai duusa-j bai-laa person-GEN hat end-C COP-1H.PST

'When I went to the department store, men's hats were sold out'.

[Ibid.]

23 As Hashimoto does not specifically discuss -j bai-n/aa, but more in general -j bai-, I am even quoting an example with past reference here.

In (48), the speaker is actually moving towards a place where she can buy a ticket, thus no future meaning is conceivable. In (49), the store was running out of hats, but a few were still left. In contrast, Ono [1999] correctly concludes that -j bai- does not share the resultative-progressive interpretations of Japanese -te i-ru.

3.3.1.3. There are some phenomena related to aktionsart that deserve special mention. First, there are a number of inceptive verbs such as med- 'to come to know+know', zogs- 'to stop+stand still' or gomd- 'to get offended+be offended' where the period highlighted by a Progressive (but not by -dag) as illustrated in (50) is the period after their starting point which in turn can be referred to by any past form such as bii gomdsong 'I took offence', by implication 'I am offended', or 'I was offended'. Inchoative verbs like suu- 'to sit down+sit' (see [Song 1997]) even lexicalizes both periods. On the other hand, it is quite possible for a Progressive to refer to a series of events, and most easily so with a punctual predication as in (51). It must be noted, though, that Progressives can also be taken to refer to the pre-phase of a punctual event as in (52) (cf. [Ono 1999: 124-125]).

(50) bii teren-d gomd-jii-n

1SG D.DIST-DAT take.offence+be.offended-PROG-DIR [People, when they were working in factories, used to become real workers and [could] live this way. Now, to say it frankly, this (system) doesn't exist anymore.] 'I'm taking offence in this'.

(51) ydron oroi-d=oo hed ge-j or-j in.general evening-DAT=RP how.many COMP-C enter-C ir-jii-n?

come-PROG-DIR

'By the way, when do you come home in the evening?'

(52) 13 nas-tai huuhed 1500-iin arhi uug-aad uhe-j 13 year-COM child 1500-GEN vodka drink-C die-C bai-na24

COP-DIR

'A child of 13 is dying after having drunk 1500 ml vodka'.

24 http://twitt.caak.mn/view/post: 1201419, retrieved 2013-12-10. The entire text (by sesseedd, 2011-12-19) reads: "13 nastai huuhed 1500-iin arhi uugaad uhej baina tsaana chin".

3.3.1.4. The main forms in performatives are the Voluntative -ii, the Firsthand -laa and the literary past form -v. Yet, some kinds of informal performative speech acts can be done with -jiin:

(53) sonirh-ooroi, tan-iig ur '-j bai-na!25 be.interested-IMP.HON 2SG.HON-ACC invite-C COP-DIR 'Get interested, I invite you!'

3.3.1.5. While -jaig-aa does not allow for any such contrast, the form -jii-n=aa in contrast to simple -jii-n indicates an overall friendlier stance of the speaker and is variously described by informants as "friendly", "more intimate" and sometimes "softened", at its most negative as "chatty"26. The reasons for using this form vary widely, ranging from a need to be careful about the statement in question to the speaker expressing his emotions more clearly, be they negative or positive.

(54) A: en bair-iig hezee av-s=iin?

D.PROX flat-ACC when buy-PRF.P=MC B: oo manai-h bas neleen

INTERJ 1PL.GEN-AT also considerably ud-jii-n=aa last-PROG-DIR=POL A: 'When did you buy this flat?' B: 'Oh, we are here now for quite some time'.

(55) odoo saihan nam bur-jii-n=aa

now nice INT tell.lie-PROG-DIR=POL [BB: 'If he says he's 40, you might be 36'. A:] 'Now he's just bullshitting'.

In (54), informants suggested nuances such as sadness, happiness or resignation on the part of the speaker, while others rather emphasized the speaker's friendly stance towards the listener. In (55), the word

25 https://www.facebook.com/media/set/?set=a.202672103086229.478 36.181906298496143&type=1, retrieved 2013-12-10. The headline to a post containing no text, but only pictures of people wearing particular clothes.

26 There might be a gender component to the social perception of =aa in -jiinaa, but both the materials selected for my corpus that contain more male speakers and the substantial female bias within my group of informants make it difficult to investigate this. On the face of it, =aa is frequently used by speakers of both genders.

bur- 'to tell lies' is softened by =aa that makes it recognizable as a "joke"27 rather than an actual defamation. It is quite interesting to observe similar forms in context:

(56) H: amr-iig er-ii? haan-aas

peaceful-ACC search-VOL where-ABL yav-jaig-aa=n taa? go-PROG-—DIR=MC 2SG.HON A: horin tav-dguar suvag telviz-ees yav-jii-n=aa

twenty five-ORD channel TV-ABL go-PROG-DIR=POL zaa tanai shtap-iin daraG mdnhdol well 2PL.GEN staff-GEN leader NAME ah-tai uulz-cha-h hereg-tei bai-n

elder.brother-COM meet-INT-FUT.P necessity-COM COP-DIR H: 'How do you do? What organization are you from?' A: 'I'm coming from TV 25. Well, I need to meet your chief of staff Monhdol'.

(57) ... bichilg-uud-iig odoo nuuc durang-aar

recording-PL-ACC now secret telescope-INS av-s=iim=aa teg-eed en nevtruulg-iig taa

take-PRF.P=MC=POL do.so-C D.PROX broadcast-ACC 2PL buhen-d hure-g-y=aa gej bod-jii-n=aa

all-DAT reach-CAUS-VOL=POL COMP-C think-PROG-DIR=POL '...we have recorded [how citizen B. was maltreated and insulted] with a hidden camera. So I think I shall present this broadcast to all of you'.

In (56), the reporter uses a more polite, softer form to introduce himself, but switches to a more matter-of-fact stance when making his actual request. In (57), the reporter is anxious that materials recorded in such a fashion might meet with disgust from his audience and thus tries to placate them and convince them to watch his broadcast by intensive use of =aa.

3.3.2. The Present Continuative-Resultative -aad bai-n/aa. The Present Continuative is used with unbounded predications to express that an

27 Mongolians tend to use the word narg'aa 'joke' to refer to statements that are socially inappropriate in order to oblige others not to take them literally, even if they do mean it.

event continues. With boundary-oriented predications, it refers to recent, apparently tangible results. In the corpus, -aad bai-n is attested 77 times, and in 8 of these it is preceded by -ch-, a marker signalling the attainment of the boundary of an action ([Ebert 1999: 333, 338] on Kalmyk Oirat, [Matsuoka 2008: 51-54] on Khorchin). Simple -aad baig-aa, on the other hand, only occurs 41 times, only one of these featuring -ch-, but 24 instances are marked with =iin. It is possible to coordinate several adjacent main verbs with a single copula, e. g. deerelheed doromjlood zodood bain 'keeps on maltreating, humiliating and hitting him'.

No extensive research on -aad bai- seems to exist for Khalkha. The marker was discussed in some detail by Cinggeltei [1959: 8-12] and Matsuoka [2008: 33-39, 58-63, 71-75] for Eastern Central Mongolian varieties. Cinggeltei claims that -aad bai- can have a resultative or a continuative/iterative meaning, the latter being indicated by a contraction to -aadai- which is not observable in Khalkha. The difference between continuative and iterative depends on the verb stem: he assigns a continuative meaning to "durative" verbs that I would classify as undirected activities (e. g. ajill- 'to work') and inceptive states (zogs- 'to stop+stand'), but an iterative meaning to "non-durative" verbs that I would classify as directed accomplishments (av- 'to take') and directed achievements (ir- 'to come')28. Matsuoka argues that -aad bai- yields a resultative meaning with transformational actional phrases, but may refer to the same phase as the Present Progressive with inceptive predications. Activities yield iterative readings, which can also be forced for all aktionsarten by the clitic =l. -aad bai-n cannot be used for permanent events such as the earth rotating. (For a more extensive presentation

28 For this paper, I basically assume the aktionsart classes of Croft [2012] with some complementary classes from Breu [2005], but I apply them to verbs or actional phrases (cf. [Johanson 2000; Breu 2005]), not clauses. Mostly as a consequence of this, I assume that inceptive states (as for instance proposed by Johanson [1971], cf. [Breu 2005]) are necessary: while Croft [2012: 37-38] argues on the basis of sentences like I suddenly knew the answer and I instantly remembered her that any state can get an inceptive interpretation in the presence of the right context, this would not explain why a state verb like bod-loo '[I] thought' (cf. Progressive bod-jii-n '[I] think') can only refer to a past situation, whereas an inceptive state verb like med-lee ' [I] recognized' (cf. med-jii-n ' [I] know') is difficult to conceptualize as anything but a beginning.

and discussion of these analyses, see [Brosig 2014a].) For Khalkha, Dugarova and Jaxontova [1988: 219] claim that -aad bai- refers to the resultant state or result of an action with terminative verbs such as ir-'to come' or "neutral" (inceptive or inchoative) verbs such as dms- 'to put on+wear', but to an action in progress with non-terminative verbs such as togl- 'to play'. Adverbials also refer to the resultant state.

The small number of examples in my corpus (n = 36) is unsatisfactory in sorting out aktionsart matters, but as far as the data goes, Dugarova and Jaxontova seem to have it basically right: with undirected activities, -aad bai- usually yields a progressive, more precisely continuative reading, e. g. alhaad bain 'keeps on walking' or setgelzuigeeren togloodbaigaamaa 'keeps on playing on her psychology'. The period thus referred to is perceived as comparatively long or sometimes even as unpleasant by the speaker. For instance, am cangaad bain 'I am thirsty' is the normal way to convey this notion in a situation where the speaker has already been thirsting for a short while and wants to resolve the situation. Am cangajiin, in contrast, is more particular, indicating either that the speaker has just noticed or that she has already been in the situation for a while, but has no remedy for it. Similarly, the actionally interesting (58) indicates that an uncomfortable fact continues to be (or be instantiated as) true.

(58) hedii bii hatuu hel-j baig-aa bolovch how.much/although 1SG hard say-C COP-—DIR although yag unen baig-aad baig-aa

exact true COP-C COP-—DIR

'However hard I'm saying it/Although I'm saying it in a hard way, it continuously turns out to be precisely true'.

The continuative meaning also obtains for accomplishment verbs with indefinite plural objects and the like such as huuhded haldaad baigaa shuu 'keep on infecting children', states as in en jijigdeed bain 'it is still too small [I really wanted to wear it]' and inceptive states such as ingej oilgood baigaan 'I understand it like this'. In (59), even an accomplishment manages to get interpreted as being in progress due to the overall context:

(59) unta-h ge-ed or der=ee zas-aad sleep-DIR.P COMP-C bed pillow=RP put.into.order-C

bai-n=uu? COP-DIR=Q

'Are you making your bed in order to sleep [now]?'

Iterative readings are regularly found with reversible or cyclic punctual predications such utas untraad bainaa '[my] telephone keeps switching off' or endhun zodood bain 'is hitting somebody here' which happen on the same occasion. Adverbials such as in yoron huuhdee haraad bain uu 'are you watching the children, in general' and cag tutam shalgaad bain 'keeps on checking them every hour' also introduce an iterative, multi-occasional interpretation. In (60), even evidential cues play a role: -aa indicates that B's utterance is not based on his current direct perception, and one contingent laughing event does not tend to exceed the probable time of perception here. B must therefore even include events into his reference that he saw previously.

(60) A: aayaa inee-jii-n, inee-jii-n=uu?

EXCL laugh-PROG-DIR laugh/smile-PROG-DIR=Q B: aav-d=aa ineeg-eed baiG-aa

father-DAT=RP laugh/smile-C COP-—DIR [A: 'How old is your child now?' B: 'It's 10 months now.'] A: 'Oh, it's smiling! Is it smiling?' B: 'It's smiling at his father (i. e., "at me")'.

For contextualized accomplishments, informants tend to get resultative interpretations, e. g. daraagiin gazraas urigdaad bain 'he has got invited from the next place [and thus is in a hurry now to get things done here]' or medeelel ireed bain 'we have received word'. In one case, an inchoative verb (i. e. a verb that exhibits two lexicalized phases of action: nuujiin 'is going into hiding'/'is keeping hidden') was understood as referring to a resultant state: hudlaa nuugaadbain uu 'are you hiding lies from me'. Notably, informants did not perceive any reference to the inception of this state, and it required the use of -ch- to get the interpretation 'have you hidden lies from me'.

The form -ch-aad bai-n/aa, while rare, usually seems to combine with accomplishments, referring both to a recent change of state (which can thus be perceived as sudden as in (61)) and the resulting state that still holds. (62) can be used to illustrate the differences between semantically similar forms: the Firsthand Past -laa would express a recent change of state, but no lasting result, the Present

Perfect -sang bai-n would both refer to a change of state and the resulting state, but would not imply any recency, and the simple Continuative -aad bain would gloss over the change of state, thus not highlighting the purposeful actions of the subject:

(61) odoo bid nar=chin N ah-iig

now 1PL PL=STC NAME elder.brother-ACC bai-h=gui bol-sn-oos hoi-sh

COP-FUT.P=EX.NEG become-PRF.P-ABL after-PROS tag_bol-ch-ood baig-aa shd disappear-BOU-C COP-—DIR IP

'Now since the death of N., [these relatives] have gotten out of contact'.

(62) te-heer yag odoo bol sport-iin tdv ordn-ii do.so-CNS.C exact now TOP sport-GEN central palace-GEN gadaa ir-ch-eed bai-n

outside come-BOU-C COP-DIR

'So exactly now I have arrived outside the Central Sport Palace. [I will now ask somebody whether I can go to her place and stay the night there]'.

In (63), the non-inceptive stative verb naid- 'to hope' acquired an inceptive interpretation in the presence of -ch-. The connotation here seems to be that the speaker had no choice but to do what he did, that particular untrained hairdresser being the only person working in his barracks. But such examples are few. For (64), most informants were reluctant to accept -ch- with the undirected activity hai- 'to search', arguing that the verb ol- 'to find' would be used for reference to a successful search, so that no resultant state could be conceptualized here. Yet, one single informant suggested an unsuccessfully completed search. While this does not fit well with the overall context in (64), it shows some of the potential of -ch- to both create and emphasize lexical boundaries.

(63) baig-aa ganc-han uschin-d=ee naid-ch-aad COP-—DIR single-DIM coiffeur-DAT=RP hope-BOU-C bai-n=uu?

COP-DIR=Q

'Have you put your trust into your only hairdresser?'

(64) gue gue bii ddr-iin-h=dd herch-seng EX.NEG EX.NEG 1SG self-GEN-AT=RP cut-PRF.P baicaa-g haig-aad bai-n=l=daa cabbage-ACC search-C COP-DIR=FOC=IP

'No, no, I keep on searching for the cabbage that I cut'.

3.3.3. The Present Perfect -sang bai-n/aa. The Present Perfect -sang bai-n/aa is used to refer to past events and their resultant state, which, however, may be as unspecific as present relevance. Due to the evidential properties on -n and -aa, inferential interpretations of -sang bai-n are only possible, but not necessary.

There is no research on -sang baigaa, but some researchers have commented on -sang bain. The first to do so was Cinggeltei [1959: 15-18] who considers it a composite form where -sang refers to a relative past and bain to an absolute present. The form indicates that there is present evidence that this event occurred in the past. No present result is necessarily assumed. The form can thus be used to emphasize that an event actually took place. Dugarova and Jaxontova [1988: 217-219] write that -sang in -sang bain refers to the time when the event took place with non-terminative verbs, while bai(-n) refers to the time when this was found out. With terminative verbs, they can point to a result. Adverbials such as gurvang cagt 'by(/at) three o'clock' refer to the resultant state, adverbials of time duration such as hoyor cagiin tursh 'for two hours' cannot be used. Umetani [2006: 3] only mentions an inferential usage. Monh-Amgalan [1998: 70] holds that -sang bain expresses epistemic uncertainty as to whether the event actually took place, but he might identify inference itself with uncertainty. Saruul-Erdene [1998: 96] argues that the past is relative to an absolute reference point in the future, providing (65) as example:

(65) margaash chamaig ire-he-d bi nom-oo tomorrow 2SG.ACC come-FUT.P-DAT 1SG book-ACC unsh-chih-san bai-na

read-BOU-PRF.P COP-DIR

'When you arrive tomorrow, I will have read my book'.

It is surprising that he simply lists this as the function of -sang bain. One could use a simple Perfect in German here, too, but nobody would consider future reference to be its main function.

In the corpus, -sang bain (n = 52, 29 of these with -ch-) is roughly as frequent as its closest related forms -sang baigaa (n = 50, 18 of these with -ch- and a mere 3 with =iin), -jee (n = 50), and -aad bain (n = 77). It can either be inferential or perfect for predications that include a lexicalized border:

(66) ... piotrajamp ge-ed en kolonk deer binzen

NAME COMP-C D.PROX gas.station on petrol

tugeeguur deer ir-seng bai-n=aa disperser on come-PRF.P COP-DIR=POL '[Well, dear audience,] we now have arrived at the gas station Petro Jump [in Songin Hairhan district] at the fuel dispensers'.

(67) chamaig hulgai... hulgai-n hereg hii-seng bai-n=aa, 2SG.ACC theft theft-GEN act do-PRF.P COP-DIR=POL tiim uchr-aas bii chamaig odoo saatuul-jai-n=aa

so reason-ABL 1SG 2SG.ACC now arrest-PROG-DIR=POL

ge-ed ing-eed avchir-sang bai-n29

COMP-C do.so-C bring-PRF.P COP-DIR

'[Then the local police took that monk away and brought him to

a room. Having brought him there, now, the policeman said],

"You have committed an act of theft. Because of that, I am

arresting you now." He had brought him there like this'.

(68) zaa zaa buruu oiloG-som bai-n well well wrong understand-PRF.P COP-DIR

'[I had been thinking that there is a separate room. So this is not the case, right?] Well, I have misunderstood it'.

(66) is an example of a pure perfect usage of a verb that might either be interpreted as accomplishment or punctual. The speaker himself has consciously gone to that place and is now about to undertake other things. Using the Non-firsthand Past -jee instead would imply deferred realization. In a similar example, a soldier reports that a particular person has arrived (irseng bain) and now asks his officer how to proceed. (67) contains two perfects. The directed accomplishment avchirsang bain 'has brought' merely prepares the setting for what is to come, namely, that the monk starts with a proper religious ceremony to put

29 From a retelling of a criminal case that was narrated to the speaker before. The file name in my data is khalkha0021.

curses on all the local police. On the other hand, no resultant state is readily observable for the accomplishment hiiseng bainaa 'have done'. The point is rather that the policeman takes all his evidence together and infers that the monk committed a crime, also giving him the opportunity to challenge this finding. In (68), the verb is inceptive-stative, and a deferred-realization-inference is made to the onset of this event, i. e. the process of acquiring an understanding. The speaker is correcting his understanding while speaking, a nuance missing if the Non-firsthand Past -jee/-ch-j had been used instead. If the speaker had intended to refer to the state of his wrong understanding, he could even have used -sang baigaa, e. g. bii en asuudliig buruu oilogsong baigaa, chii nadad odoo sain tailbarlaj dgddrei 'I've acquired and hold a wrong understanding about this matter, so please explain it to me properly now' (constructed by informant).

The perfect in -sang baigaa is not capable of expressing any form of inference, as it refers to a situation that the speaker is already familiar with:

(69) ... hand-aach=ee ge-seng zaa hand-sang

beckon-IMP.INT=POL say-PRF.P well beckon-PRF.P baig-aa, burtguul-sen baig-aa ge-seng COP-—DIR register-PRF.P COP-—DIR say-PRF.P '[Then all of us] said: "Contact [the police, the law!]" He said: "Well, I have contacted the police, I have registered"'. [Then he said: "Please come here right now to meet me, and bring the police with you!"]

(70) minii daatGal-iig vol noming daatGal hii-j 1SG.GEN insurance-ACC TOP NAME insurance make-C dg-sdm baiG-aa

give-PRF.P COP-—DIR

'[I'm somebody who gets around a lot.] As for my insurance, Nomin Insurances have made it for me'.

In (69), the point is that the cited speaker has contacted the police, but as they apparently have not taken any concrete measures, he expects them to forsake him if he does not contact the media (i. e. the speaker) as well. (70) exemplifies the typical expository use associated with commercials or with the speech of TV people in general. The speaker

is emphasizing his current state of being insured, else he might have used the more neutral Factual -sang.

In the corpus, -sang baigaa is rarely used with predications that lack a resultant state, but a few such cases can be found:

(71) en jil bol neleed ih shalgalt hii-gd-seng D.PROX year TOP rather many exam do-PASS-PRF.P baig-aa

COP-—DIR

'This year, a lot of inspections have been done'.

(72) altanturuu=maan' olon ail-uud-aar hoN-song NAME=2PL.POSS many family-PL-INS stay.overnight-PRF.P baig-aa. tanai-d hon-ii nevtruug-iig COP-—DIR 2PL.GEN-DAT stay.overnight-VOL broadcast-ACC odoon-oos bii hii-h-eer bol-jaig-aa

now-ABL 1SG do-FUT.P-INS become-PROG-—DIR '[Here on TV 25, the TV program Tanaid honii is beginning. Dear audience, you all might be surprised to see me, Tengis, today instead of Altanturuu.] Our Altanturuu has stayed with many families now. From now on, I am taking over Tanaid honii'.

(73) C: hoyuu-laa yav-jii-sang bai-n neg=n

two-COLL walk-PROG-PRF.P COP-DIR one=3POSS yamar hun bai-s=iing? what.kind person COP-PRF.P=MC H: med-kuu bii hoyuu-laa

come.to .know+know-FUT.P.EX.NEG 1SG two-COLL ganc-aar=aa=l yav-jii-n alone-INS=RP=FOC go-PROG-DIR C: en hoyor hoy-uulaa mon baig-aa biz=dee? D.PROX two two-COLL same COP-—DIR IP=IP taa hoyor enuugeer yav-jii-sang baig-aa 2PL two D.PROX.INS go-PROG-PRF.P COP-—DIR biz=dee? IP=IP

C: 'You were walking together. Who is the other person?' H: 'I don't know. I am on my way here together... alone'. C: 'These two [on the video] are together, aren't they? You two were walking along here, weren't you?'

(71) and (72) are the only case in the corpus where no tangible result can be construed. But in (71), there are general conclusions that one can deduce from the results of these examinations, and in (72), the basic idea must be that Altanturuu is worn out or has lost interest in presenting Tanaid honii30. A similar phenomenon can be observed in

(73) that has a Progressive in the scope of the Perfect. In the first instance, the police officer who had been interrogating a thief about whether he had an accomplice infers from video material that two people were walking as a pair. Then, in the third sentence, having already established this as his knowledge, he switches from -n to -aa, still referring to a past activity in progress with present relevance31.

3.3.4. The evidentially marked Habitual -dag bai-n/aa. The overall corpus data on -dag bai-n/aa consists only of 1 and 7 examples, respectively, so a proper analysis is difficult to conduct. However, there is a contextual cue to the meaning of -dag baigaa: in 4 of 7 cases, it co-occurs with te / tiimee, a device for forming tag questions and eliciting agreement. If these collocations are used for creating a kind of common ground for the speaker and listener, the event in -dag baigaa does not seem to be taken for granted by both interlocutors.

(74) ddr yumn-ii talaar bid buhen bas av-sang other thing-GEN concerning 1PL all also take-PRF.P baig-aa tiimee biye_uNlegchi-d baig-aa gazar bol COP-—DIR so prostitute-PL COP-—DIR place TOP emh zambraa=guu baidal uus-deg baig-aa order discipline=EX.NEG situation emerge-HAB.P COP-—DIR tiimee... tend shugel-deg, uurel-deg, tiim biz=dee so there be.haunted-HAB.P nestle-HAB.P so IP=IP '[But it's not only about the prostitutes], we all are also affected by other issues, right? At a place where prostitutes are, a chaotic,

30 With Tengis presenting, even the general audience quickly lost interest in the program, so that it was cancelled sometime afterwards.

31 The illocutionary particle collocation biz=dee expresses a fairly strong claim, while mildly challenging the addressee to confirm or disprove it. It does not combine with direct perception -n in complex forms, but it combines with the Potential -n used on its own, so the combinatory restriction is not morphologically conditioned. Rather, in order to state a claim and dare somebody else to challenge it, one usually needs to relie on information that one has already established as one's own knowledge.

criminal overall situation arises, right? [Excessive alcohol usage follows, thieves follow, brawls...] All bad things haunt such places and nest there. It's just like this, right?'

(75) borjgon aylguug-aar yar '-dag tov halh-iin NAME dialect-INS speak-HAB.P central Khalkha-GEN aylguug-aar yar'-dag baig-aa, te

dialect-INS speak-HAB.P COP-—DIR TAG [A: 'The inhabitants of the Central Province, they are part of the Borjigin clan, right?' B: 'Yes'. A:] 'They speak the Borjigin dialect. So they speak the Central Khalkha dialect, right? [What are your thoughts about this?]'

Quite notably, as the journalist in (74) is attempting to make a general statement, the temporal state explanation that I had considered earlier [Brosig 2009b] has to be rejected. As the speaker continues with a regular Habitual form, -dag baigaa here probably emphasizes the subjective knowledge of the speaker used for building up a common ground, while the regular habitual form used in the next sentence is already a factual claim, even if the speaker, still unsure whether the TV audience will accept the argument, still tries to obtain confirmation. In (75), the speaker identifies the Borjigin dialect with the Central dialect, using baigaa instead of bain as she is taking recourse to general reasoning. Meanwhile, the only instance of -dag bain (shd) in the corpus presumes an event that occurred once in the past. This kind of usage will be discussed in more detail in 4.3.3.

(76) A yaa-laa, te? B egch "C NAME do.what-1H.PST TAG NAME elder.sister NAME davh'-aach" ??? ge-l=uu? gallop-IMP.INT UNIDENTIFIED .WORD say-1H.PST=Q teg-seng=chin hariu=guu inee-deg bai-n shd do.so-PRF.P=STC answer=EX.NEG laugh-HAB.P COP-DIR IP 'Do you remember what A did? And wasn't it so that B yelled "C, go!" Because of that, A/C burst into laughter'.

3.3.5. The Present Prospective -h ge-j/ed bai-n/aa. The Present Prospective denotes that given a current situation, some future event is about to happen, as illustrated in (77) below:

(77) yds-ddh ail-d=aa hoN-jii-n

nine-ORD family-DAT=RP stay.night-PROG-DIR ge-j=dee32 yds-ddh ail-d=aa

COMP-2H.PST=IP nine-ORD family-DAT=RP hono-hjii-n stay .night-PROS-DIR

[A: 'Which family is it that you stay at now?' B:] 'I'm staying overnight at my ninth family. No well, I'm about to stay overnight with my ninth family. [But will you let me or not, that hasn't become clear yet!]'

The Prospective has the two forms -h gej bai-n/aa (n = 32) and -h geed bai-n/aa (n = 13). It transparently consists of the constructs -h ge- that also forms part of two constructions expressing intention and purpose (see 5.2) on the one hand and the Progressive or Continuative constructions -j bai- or -aad bai- on the other hand. Like in the Progressive, a contraction of -j bai- to -jai- or -jii- (conditioned by the followed suffix) is possible. If it takes place, even the elision of [g] and the genesis of a single phonological word as in bolo-h-joig-oo < bolo-h ge-j baig-aa (become-FUT.P COMP-C COP-—DIR) 'is about to become' can take place. If /g/ is not lost, it can fuse via a regular phonotactic process with the preceding /x ~ %/ to yield [hk] /xg/. Within the Prospective construction, even geed may be subj ect to vowel harmony, e. g. yaa-h=g-aad baig-aa =n (do.what-FUT.P=COMP-C COP-—DIR=MC) 'what is SUBJ about to do'. The Progressive-based and the Continuative-based Prospective are thought to differ in whether they portray the present situation as enduring or as simply in progress, but this has not been investigated in any detail.

The prospective or foreseen situation can either be due to some kind of schedule as in (77) or (78), an inevitable fact as in (79), knowledge derived from a situation frame as in (80), or the speaker's own intention as in (81) and (82). While -n is used for a currently perceptible situation

32 ge-j=dee must originally have had an evidential meaning, e. g. Shuuh ge-j=dee, bulhai bai-j. Shudarga ge-j=dee, hudlaa bai-j which might be rather literally translated as 'Apparently, they said "court of justice". It turned out to be a cheat. Apparently, they said "earnest". It turned out to be a lie'. A free translation might rather be 'You couldn't call this a trial, it was just a cheat...', possibly attacking a presupposition rather than an actual statement. There are even derived constructions such as olo-h=ch ge-j=dee 'it is a pity that they will/would find it' which is mainly used for disappointment.

as in (77), (79), (80) and (81), -aa is used for already established knowledge: the speaker of (78) refers to an event at a distant place, and the speaker of (82) as the cook of a cooking show must have made his plan of preparing the noodles in such a manner well in advance.

(78) manai yapon-d bai-dag band' ber gui-h 1PL.GEN Japan-DAT COP-HAB.P boy bride beg-FUT.P ge-jaig-aa

COMP-PROG-—DIR

'Our son in Japan is about to propose to somebody'.

(79) odoo minii tor-son odor ehle-hchii-n now 1SG.GEN be.born-PRF.P day begin-PROG-DIR 'Now [= i. e. 23:55] my birthday's about to begin'.

(80) bii yoron enG-hed haaN hoN-k-ood

1SG in.general do.so-C.when where stay.night-FUT.P.COMP-C

bai-n=aa?

COP-DIR=POL

'Now, by the way, where [within this house] am I going to sleep?'

(81) naiz-iin-d=aa ocho-j hono-h ge-j friend-GEN-DAT=RP proceed-C stay.night-FUT.P COMP-C bai-n

COP-DIR

'I am about to go to my friend's and stay the night there'.

(82) goimong=oo bii jigne-hjaig-aa noodle=RP 1SG steam-PROS-—DIR

'[Well, now I shall fetch my noodles and prepare them.] As for the noodles, I'm about to steam them'.

4. The Habitual

In addition to being followed by the copula as was discussed briefly in 3.3.4, the Habitual Participle -dag can be used on its own either with reference to habitual, characterizing or generic events discussed in 4.1 or to unexpected mono-occasional events as discussed in 4.2. Together with non-finite aspect markers as discussed in 4.3, it refers to iterative or unexpected mono-occasional events with an internal time structure.

4.1. Habitual use of simple -dag

In general, -dag is described as marking both habituality and undelimited states [Byambasan et al. 1987: 101; Svantesson 1991: 197; Kim 1995: 143-144], the former with dynamic and the latter with non-dynamic or generic predications [Song 1997: 288-297]. The most detailed analysis of its imperfective semantics is provided by Hashimoto [1995b]. He criticizes the term MtMtt 'habitual' and assumes five semantic categories: (1) M *) M'TftM/repeated actions/situations, i. e. events iterated over multiple occasions that may, but need not form a habit. Notably, -dag does not combine with adverbials that indicate a concrete number of repetitions or with the adverb odoo that too concretely focuses on a delimited here and now. All of Hashimoto's examples are dynamic verbs; (2) / ^^ stative

actions/situations, i. e. states as marked by stative verbs and stative senses of verbs like ajilla- 'to work > be employed'; (3) ^ftt/ characteristics/habit, i. e. attitudinal and potential predications in the sense of Bertinetto and Lenci [2012], e. g. Minii naiz can-aar nadaas sain gulda-dag (1SG.GEN friend ski-INS 1 SG.ABL good glide-HAB.P) 'My friend skies better than I' [Hashimoto 1995b: 53]; (4) naming, but this is limited to the complementizer verb ge- and can thus safely be ignored here; (5) ^ffiffi&^fM / ^^ generic actions/situations. Hashimoto subclassifies these into several degrees of genericity, all of which can be marked with -dag33.

In the corpus, simple -dag has a token frequency of 479, precisely 75% as frequent as simple -n. In 59 cases or 12% of cases, it is used together with interrogative clitics, and in 33 instances it was used together with -ch-. Even in the corpus, -dag is used for regularly observable phenomena as in (83), habits as in (84) and (85), and routines as in (86).

(83) minii sugan-d us uraG-ch-laa minii en

1SG.GEN armpit-DAT hair grow-BOU- 1H.PST 1SG.GEN D.PROX boojgoon-ii deeg-uur us urg-aad ehel-lee ge-ed balls-GEN on-PROS hair grow-C begin-1H.PST COMP-C

33 These terms can be romanized (Hepburn-system) as follows:

shukansei, koi, jitai, ff^M^ kuri-kaesu, jotai-teki-na,

tokusei, jusei, nazuke, sojo-teki-na.

eng-eed huuhed ih ai-dag hyamar-dag

do.so-C child much fear-HAB.P worry-HAB.P

'"In my armpits, hair has grown. All over my balls, hair has

started to grow". — Children are very afraid of such things.

They feel worried'.

(84) tern-iig=ee bii ajll-uul-dag teg-eed biznis-iing D.DIST-ACC=RP 1SG work-CAUS-HAB.P do.so-C business-GEN zovolgoo og-dog teg-eed hajuuG-aar=n lekts advice give-HAB.P do.so-C next-INS=3POSS lecture unsh-jii-n, bankn-uuda-d lekts unsh-dag read-PROG-DIR bank-PL-DAT lecture READ-HAB.P meregjl-eer=ee lekts ungsh-dag end tend profession-INS=RP lecture read-HAB.P here there surGuuli-uda-d bagshil-dag

school-PL-DAT teach-HAB.P

'[Now there are two language centers.] Those I run. Then I do business consulting. On the side, I'm giving lectures, I give lectures at banks. Read lectures in my profession. Teach at schools here and there'.

(85) bii ugan nom-iin delguur-eer ih yav-dag 1SG actually book-GEN shop-INS much go-HAB.P 'I actually go a lot through bookshops'.

(86) teg-ee=l naim ge-e=l hoolon-d=oo or-dog do.so-C=FOC eight COMP-C=FOC meal-DAT=RP enter-HAB.P bid nar

1PL PL

'[After the running exercises, we come back, clean up here] and then at eight we have a meal'.

If -n were used instead, (83) would not portray regular observations, but a generally predictable process and would thus become less scientific and more explanatory, (84) would refer to the future, and the exactness of the time and the regularity of the event would decrease in 0(86). For this very reason, some informants are skeptical as to whether -n can be used in (85) together with the adverbial ih 'much'.

-dag is also used with predications that either characterize some fundamental attitude of the subject or represent its potential function

rather than referring to the concrete, regularly reoccurring actualization of an event [Bertinetto, Lenci 2012]. -n would result in a future meaning for all examples below, suggesting an unexperienced speaker in (87) and (88) and a reinterpretation of discourse-structuring odoo into temporal odoo in (89):

(87) bii lavshaa-g bol ih Goy hii-deg 1SG noodle.soup-ACC TOP very nice make-HAB.P 'As for noodle soup, I prepare it very well'.

(88) tohooromjn-ii davuu tal=n bol tohooromj equipment-GEN advantageous side=3POSS TOP equipment ura-gshaa pad ge-ed una-nguut mor'-d joohon front-ALL IDEO COMP-C fall-C.as.soon.as horse-PL a.little zog_tus-aad mor'-d saihan taivan ereg-deg stop.dead-C horse-PL nice peaceful turn-HAB.P

'The advantage of this equipment is that as soon as it thuds off, the horses stop short and they turn nice and smooth'.

(89) odoo zalG-aad buten dorvon cag cenegel-deg now connect-C full four hour charge-HAB.P 'Now plugged in, it charges fully within four hours'.

Generic events, here understood as events that intrinsically characterize a class of entities or its behavior, are not well represented in the corpus, but undelimited states such as a subgroup being part of a larger organization or a people that belong to a region or yar 'dag 'speak' a particular language can be found. The contrast to -n is sometimes rather slim:

(90) cagdaag-iin baiguulG-iin neg butec-d=ee police-GEN organization-GEN one structure-DAT=RP bagt-dag

group-HAB.P

'[Well, our troop is one group of the Interior Soldiers.] It belongs to one part of the police organization'.

(91) Hovguuded bol arvaas aruun gurvan nasand ter huuhdiin tomsgon tomrood tiimee en dotor bol er belgiin esuud er belgiin gormonuud bol bolovsorch eheldgee. Tegenguut bol huuhdiin maan tomsognii hemjeen tomroN. Duu hooloi=n buduurch ehleN. Yagaad gevel tomsgn dotor er belgiin gormon testeron gedeg gormon bolovsorch

eheldeg. Enentei uylduulaad huuhdiin maan duu=n buduureed ehleN. Ternees gadN odoo dnddgnii orchim sugand=n us urgaj ehleN. Tegeed tdvdnhch joohon tomrood engeed odoo duu hooloin buduureed ehelchdeg.

'For boys, between 10 and 13 years, their testicles enlarge and, yes, meanwhile, the gametes and the male hormones begin (-dg=aa) to develop. Immediately afterwards, the testicles will enlarge (-n). Their voice will start (-n) to become deeper. This is because within their testicles, the hormone testosterone begins (-dag) to be developed. In connection with this, the voices of our children will start (-n) to become deeper. Moreover, around their testicles and in their armpits hair will grow (-n). So even their larynx grows a little and then their voice begins (-ch-dag) to become deeper'.

In (90), -dag refers to an official, well-established relationship, whereas -n would render it the opinion of the speaker ('could be grouped'). In (91), the main changes themselves are marked with -dag. As -dag cannot establish any connection between the events, -n is used to indicate that some events follow from others. In the last sentence, -ch-dag which often marks conditioned changes was used instead, possibly motivated by the condition integrated as a subordinate clause into the same sentence.

As observed in the literature, -dag is also used with stative predicates to express temporally undelimited states. In the case of psych verbs such as in (92), (93) and (94), the thoughts in question are expressed as the opinion of the subject. Note that even adverbials such as dnddddr 'today' in (94) and odoo elsewhere can appear next to -dag to some extent, being included into the scope of -dag, but not exhausting it:

(92) ... gej bod-dog Gor'd-dog

COMP think-HAB.P hope.vainly-HAB.P '[I'm very high-spirited towards my own work, so I] hope and pray that [my daughter, having gone to London, is at any rate observing from the newspapers that her father is not sitting around idly]'.

(93) hot-toi oirhon bolhoor joohon hogjol baG jereg city-COM close because little development small grade taaruu yumshig sanagd-dag

mediocre like seem-HAB.P

'It seems to me that as we are so close to the city [of Ulaanbaatar], [our own] development is a bit mediocre'.

(94) teren-d=n bii onoodor bayral-dag D.DIST-DAT=3POSS 1SG today feel.happy-HAB.P

'[My father was always striving for the truth. While the public took him to be a polemic and he was getting onto bad terms with many leaders, he was always defending the common people.] Of this, I am proud today'.

A variant of -dag with attached =aa is attested in the forms bodojyavdgaa 'went thinking [throughout my life that I had lost this great happiness]', hurdgee 'reached [a result to some extent]' and bolovsorch eheldgee '[meanwhile, the gametes and the male hormones] begin to develop', but abstracting from these three examples to a general function is probably not advisable. In a few cases that I observed, this form was used somewhat differently when rearticulating a statement in a clear manner that I had indicated I had not heard properly.

4.2. Mono-occasional use of simple -dag Song [1997: 297-300] claims that -dag on its own does not combine with past (or future) time adverbials as in *Dorj omno /daraa/ongorson jil/ireh jil end ir-deg NAME before/later/last year/next year here come-HAB.P, but requires an auxiliary (bai- for past, bol- for future). While there seems to be widespread implicit agreement on this in the literature, Sechenbaatar [2003: 125-126] pointed out for the closely related Chakhar dialect that -dag can be used in the past (and with a past adverbial) without a copula:

(95) ter jil bid hoyor Shilinhoto-d neg uulz-dag D.DIST year 1PL two PLACE-DAT one meet-HAB.P shdee

IP

'We met once in Shilinhot that year [do you remember]'.

Sechenbaatar did not emphasize it, but it is notable that this event is not only past, but even mono-occasional (without being stative).

In Brosig [2012: 16-17], I listed a number of examples of this type with structures such as (-jii-/-aad bai-) -dag (yum) (bilee) as well as the past form av-dag bai-jee 'I should have taken it' which expresses a regretful irreal deferred realization:

(96) bii chinii erdm-iin ajl-iig haya-ch-dag (overheard) 1SG 2SG.GEN science-GEN work-ACC toss-BOU-HAB.P

'So I lost your research paper'.

(97) taa yaagaad shud=ee ugaa-h-d=aa tol' 2SG.HON why tooth=RP wash-FUT.P-DAT=PROSS mirror muuhai bol-g-ood bai-dag bilee? (overheard) bad become-CAUS-C COP-HAB.P RECOLLECTION

'Why are you dirtying the mirror when brushing your teeth?'

(98) ... och-loo tegeed tend=ee 4 hono-sn-ii

go.to-2H.PST do.so-C there=RP 4 spend.night-PRF.P-GEN

daraa neg drdmdldg-iin ajil-d or-dog_yum

after one drilling-GEN work-DAT enter-HAB.P MC

bai-n=aa34 COP-DIR=POL

'[In early October, I left home and, journeying 4100 km in four days and a half,] came to [a town named Kalgoorlie Boulder in western Australia]. Then, having stayed there for 4 days, I was hired for drilling work'.

Both (96) and (97) were uttered in circumstances where the event in question had occurred recently and for the first time. Extrapolating from informant assessments, (96) appears to express surprise and regret and (97) surprise and annoyance. In (98), informants understood -dag as a device to tell a story in an interesting way. I then speculated whether examples like (96) and (97) can be understood as characterizing the person in question rather than expressing an event that repeats, but did not propose an explanation that covers all examples.

The corpus does not contain many examples where -dag refers to a single event, but it seems to be possible to establish at least two uses. First, -dag can be used as a device to tell past events (usually within a short narrative) in a more interesting, exciting way or, conversely,

34 http://gawaas.blogspot.se/2007_11_01_archive.html, 2007-11-17, retrieved 2014-01-22. Travel report.

as a device to convey one's own surprise or excitement. The only example for this use in the corpus is (99), but examples that informants constructed according to the scheme of (98) and several examples that I had the chance to observe in Mongolia myself render this use quite salient. The second use is together with an interrogative word or =iin in non-past rhetorical/aggressive questions such as (100) and (101) that express surprise and displeasedness, mirroring the similar use of bilee in an interrogatively framed accusation such as (97).

(99) teg-eed erdenbat shar kostyum pijak-tai or-j do.so-C NAME yellow suit suit_jacket-COM enter-C ir-deg bai-h=gui=yuu

come-HAB.P COP-FUT.P=EX.NEG=Q

'Then, imagine, Erdenebat entered wearing a yellow suit jacket!' [directly preceded and followed by quoted direct speech, but overall embedded in a story using past forms]

(100) chamaas=chin yamar aimaar arih unert-t=iin=be? 2SG.ABL=STC what awful vodka smell-HAB.P=MC=Q puuh yostoi hamar setle-h ge-jii-n

INTERJ really nose nick-FUT.P COMP-PROG-DIR 'What awful vodka smell is emanating from you! Ick! It really cuts into my nose'.

(101) A: en tan-ii gazar ge-seng biz=dee? X: tiin

D.PROX 2SG.HON-GEN place say-PRF.P IP=IP so

A: taa yaahaaraa en-iig shoN bol-song

2SG.HON why(RHET) D.PROX-ACC night become-PRF.P hoiN ir-j uze-dg=iin after come-C see-HAB.P=MC [A, who is a journalist, is talking to X, a person who he thinks is fraudulently claiming ownership of a place:] A: 'This place belongs to you, right?' X: 'Yes'. A: 'How come you've come here to watch it after night has fallen?'

How could the change of meaning from a habitual to a mono-occasional, emotional -dag have come about? Especially if undesirable events are concerned, it might be worthwhile to generalize from one event to a habit:

(102) edn-uud C.-d dg-hddr togl-ood

3PL-PL NAME-DAT give-CNS.C play-C hay-ch-d=iin toss-BOU-HAB.P=MC

[A: 'I cannot get the camera to work'. B: 'Why, yes. I had known it all along.] When they gave it/give something to C., she used/uses it carelessly'.

For (102), the larger context provided in square brackets indicates that the speaker refers to one concrete event. Both the converb -haar that indicates that the matrix clause is a logical consequence of the subordinate clause and additional aspecto-actional devices such as -ch-help to delimit the repetitive meaning of -dag to circumstances in which a certain condition obtains. So a characterizing generalization may suffice to explain such examples. A similar delimitaton is very transparently achieved by -aad bai-dag in (97). Such devices are absent in (100) and (101) which have =iin with unclear function and only contain rhetorical, if any generalizations. If this line of reasoning is correct and the use for surprised present observations and interesting past narratives are related, the past narrative use would be a secondary development. The generalization is lost, while the surprise on the part of the speaker is transformed into a narrative device to emphasize surprising, interesting events. This hypothesis seems to be supported by an intermediate example such as (96) which expresses astonishment about a past event. Both uses have in common that they refer to an event that, from the speaker's point of view, diverges from the normal course of events.

4.3. -dag at the end of complex forms

The suffix -dag is not only used on its own, but can also combine with non-finite aspectual forms. These forms are of particular interest as we find an aspectual marker immediately within the scope of another aspectual marker, a phenomenon that has been mentioned in passing in Mongolian studies [Cinggeltei 1959: 32-34, 40-43], but remains underdescribed and poorly if at all integrated into general frameworks of aspect. This does not indicate that it is very rare: Molochieva's description of Chechen contains a form similar to the Mongolian Progressive Habitual:

(103) daada jol hwooqu-sh v-oallu-sh father.NOM(V) hay.NOM rub:IPFV-CVBsim V-be.PROG-CVBsim xylu txo baazara

be.HAB.PRS 1PL/EXCL.NOM(D) market.ADV d-oelxu-sh

D-go:PL:IPFV-CVBsim 35

'Father used to be cutting hay every time when we were going to the market'. [Molochieva 2010: 49]

This form "is used to express situations in progress occurring in a specified time frame and presented in a habitual context", but it is only compatible with accomplishments ("gradually terminative verbs"), thus excluding aktionsarten such as achievements ("totally stative verbs"), inceptive states ("inchoative-stative verbs") [Molochieva 2010: 49, 99-102] and, according to her table, even activities. Note, though, that while the verb hwooqu- in (103) might be an accomplishment, the actional phrase (in the sense of [Johanson 2000]) is atelic due to the unspecific direct object. In Khalkha, there are not only corresponding forms such as a Progressive Habitual (n = 11) and Continuative Habitual (n = 11, none with -ch-), but even a Perfect Habitual (n = 7, five of these with -ch-) and, possibly, a Prospective Habitual. As these contexts imply the repeated presence of some observer, -dag contributes a habitual or characterizing, but not generic meaning.

4.3.1. -jii-dag and -aad bai-dag. The basic idea of -jii dag and -aad baidag is to give additional prominence to the internal structure of any single subevent within a repeated process. This structure easily combines with an incidence scheme, but it need not do so:

(104) ah-iin maash olong saihang saihang elder.brother-GEN very.EMPH many nice nice naiz nar hureel-jii-dag

friend PL circle-PROG-HAB.P

'[We have a lot of nice friends.] The many nice friends of my elder36 brother are dropping by constantly'.

35 "sim" means 'simultaneous', "v" and "d" represent gender classes.

36 Note that while the form ah-iin refers to the elder brother of the speaker in a way that is in principle unambiguous (due to the absence of =n'), some informants took this sentence as referring to the friends of the speaker

(105) engeed hal-aa=l uug-aad bai-dag do.so-C become.hot-C=FOC drink-C COP-HAB.P [Billiard is a very fine game. Then you also get drunk, and then bills...] 'Then I keep on drinking and getting drunk'. [Under such circumstances, there have been times when I had to pawn my mobile phone.]

(106) dandaa or-j ir-hd=n zalg-aatai hung always enter-C come-C.when=3POSS connect-RES person suu-jii-dag

sit.down+sit-PROG-HAB.P

'Each time they entered, somebody was sitting there who was plugged in'.

(107) ydrdn jijuur manaa-tai uy-d=ee ydrdn in.general caretaker sentry-COM time-DAT=RP in.general cerg-uud-teig=ee end=ee hamt bairl-aad soldier-PL-COM=RP here=RP together take.quarter-C bai-dag

COP-HAB.P

'In general, at those times when the sentry is with us, he is taking quarters here together with the soldiers'.

Informants say about (104) that the speaker is emphasizing the kindness of his brother's friends who always come and interact. Using simple -dag instead might rather be understood as characterizing the speaker as somebody who is very often visited by friends and thus as boastful. No incidence scheme is present here. There is a similar example where the children of a family are said to bujginjiidag baihgui yuu 'be swirling around', implicating that the observer delights in watching this process in detail. Simple -dag might rather emphasize the difficulties inherent in such a densely-packed living situation. In (105), if the speaker plays billiard, the process of drinking goes on constantly, without sufficient regard for the bills. (106) is from the perspective of a virus-infested internet cafe plus hairdresser's who were ordered by

himself who in the preceding sentence referred to himself as ah=n', literally 'somebody's elder brother', a common form of self-reference for a male speaker who is older than the addressee. This is of no consequence for the overall analysis of this sentence.

the police to shut down the internet business until the problem has been overcome. But always when the police came by and checked, they found someone sitting there with an internet connection, be it accidental or the result of disobedience. At the time of speaking, the matter is still not settled, and the police have increased their pressure against the owners. In (107), the speaker does not claim that there is always a sentry, but when there is, he usually takes quarters together with the soldiers he is responsible for. The adverbial yoron 'in general' refers to the outer aspectual phase.

Due to its continuative meaning, -aad baidag is frequently applied to rather persistent unpleasant situations that occur on repeated occasions. This is possible with conventional verbs in the presence or absence of =iin and even with the stative verb bai-:

(108) humuus yum id-eed bai-dag oor=oo ide-j people thing eat-C COP-HAB.P self=RP eat-C bol-d=gui

may-HAB.P=EX.NEG

[Yes, we spoke about this before. About how difficult it is for you to lose weight.] 'The other people keep on eating things. And you yourself are not allowed to eat'.

(109) teg-ee=l huuhed yaa-h ge-j omoor-ood do.so-C=FOC child do.what-FUT.P COMP-C defend-C bai-dg=iin?

COP-HAB .P=MC

'So why do you keep on defending the child?'

(110) odoo end baig-aad bai-dag, naiz ohing tend now here COP-C COP-HAB.P friend girl there baig-aad bai-dag, uulz-maar bai-dag aan? COP-C COP-HAB.P meet-WISH.P COP-HAB.P INTERJ

[But that must be stressful.] 'Now you're here, your girlfriend is there, and you want to meet, don't you?'

In (108), using simple -dag would result in a bizarre sentence where somebody unnecessarily states that other people eat as such, without relating it to the following sentence. In (109), the speaker accuses the addressee of interfering with justified pedagogical measures. In (110),

the speaker mentions the separation of a romantic couple because one of them is in the army.

4.3.2. -sang bai-dag. The Habitual Perfect -sang bai-dag refers to a perfect situation that occurs multiple times. Notably, all instances in the corpus that actually fit this definition use either -cang or -chih-sang instead of simple -sang.

(111) manai mongol end tend bas cagaan_hooltn-ii... 1PL.GEN Mongolian here there also vegetarian-GEN mah-guu huushuur ge-heer baicaa tdms meat-EX.NEG huushuur COMP-CNS.C cabbage potatoe en ter hii-ceng bai-dag

D.PROX D.DIS put-BOU.PRF.P COP-HAB.P [Because it's vegetarian, I'll make huushuur with veggies.] 'In our Mongolia, here and there, vegetarian... meatless huushuur are filled with cabbage and potatoes'. [You can prepare such food alright, but I will prepare rather different huushuur.]

(112) teg-eed dddrjingdd davhi-j davhi-j ire-hleer do.so-C every/all.day gallop-C gallop-C come-CNS.C ohi-d-uud=aa=l ih san-cam daughter-PL-PL=RP=FOC much remember-BOU.PRF.P bai-d=iing

COP-HAB .P=MC

'Then, after running around all day, when I arrive at home, I am always in the state of having missed my daughters a lot'.

(113) huuhed=maan' ndgdd=l cheher hool neh goy child=2P.POSS other=FOC sweet meal very nice amt-tai hool neh-seng nyalh balchir ge-ed taste-COM meal chase-PRF.P infant infant COMP-C bod-ood yav-jii-dag getel al' hediinee think-C go-PROG-HAB.P but already huuhde-d=maan' belgiin boijilt-iin uy=n child-DAT=2P.POSS sexual growth-GEN time=3POSS ehel-chih-seng, orognizam-d=n neleen gunzgii begin-BOU-PRF.P organism-DAT=3POSS rather deep

oorchlolt or-cong bai-dag

change enter-BOU.PRF.P COP-HAB.P

[We ourselves don't notice this process.] 'We continue to think of

our children almost as of babies who chase after sweets and

tasty food. But for our children puberty has already begun and

rather profound changes have already taken place in their bodies'.

(111) appears to relate to what can be observed as a result of the process of preparing huushuur when the speaker enters any such restaurant. In (112), an incidence scheme is made explicit by a temporal adverbial clause. In (113), the resultant state obtains at the time specified by the Continuative Habitual in the preceding sentence. It also illustrates a coordinate structure between two Perfect Participles in the scope of a single copula.

4.3.3. Mono-occasional uses. The Progressive, Continuative, Perfect as well as the Processive can all be used with reference to a single past event, signalling that the event takes place in spite of what could be expected given the normal course of events or overall current situation. As in other contexts, =iin seems to harmonize with this use, but its presence is neither required nor semantically neutral:

(114) chii yaahaaraa manai duu-tei buudal-d 2SG why(RHET) 1PL.GEN younger.sibling-COM hotel-DAT hon-jii-dg=iin? chii yaahaaraa manai stay.overnight-PROG-HAB.P=MC 2SG why(RHET) 1PL.GEN duu-g en zam-d uruu younger.sibling-ACC D.PROX way-DAT down tata-dg=iin?

pull-HAB .P=MC

[Now this is my sister!] 'How come you're staying in a hotel with my sister? How dare you lead my sister astray?' [I'll kill you! I'll butcher you! I'll hand you over to the police!]

(115) bii ochigdor mashin-d yav-jai-sang=chin shees 1SG yesterday car-DAT go-PROG-PRF.P=STC urine hur-eed bai-dag (constructed by informant) reach-C COP-HAB.P

'When I was travelling by car yesterday, I desperately needed to pee'. [But the car wasn't going to stop yet. So...]

(116) manai gergii manai mongol-chuud-iin deed 1PL.GEN wife 2PL.GEN Mongolian-PL.H-GEN high surguul'-uuda-d hevlel medeell-iin-hen-d zoriul-j neg school-PL-DAT press news-GEN-AT-DAT dedicate-C one bddn nom yavuul-sang bai-dag bai-h=gui=yuu? bunch book send-PRF.P COP-HAB.P COP-FUT.P=EX.NEG=Q 'You see, my wife has sent a whole lot of books for the people at the journalism departments at our Mongolian colleges'.

(114) is from a fictive conversation used to illustrate a trick that pimps play, acting as if they were the elder brother of their prostitutes to intimidate and blackmail customers37. The use of the Progressive Habitual seems to imply that they had not had sex yet, while simple -dgiin in tatadgiin seems to presume some form of completion. In (115), the speaker both relates the event in an interesting way and portrays it as contradicting an element of the scheduled case of events, i. e. for the car to drive on. (97) is another example of a Continuative Habitual. With regard to (116), informants agree that the event only took place once, even though the overall context does not seem to preclude multiple occasions.

The corpus does not contain any Habitual Prospectives, which are also fairly rare on Google, ranging from zero to two dozen hits for several of the most frequent Khalkha verbs. No analysis with the help of native speakers was undertaken, but these uses seem to be mono-occasional rather than habitual38. Possibly, a need for actual habitual generalizations over imminent events does not arise, whereas -dag as a

37 That is possible because prostitution is criminalized in Mongolia (Sadar samuuntai temceh huul' "Law to fight obscenity", 2012-01-19, §§4, 13.2.1). However, it seems that only prostitutes and pimps are prosecuted (14-30 days in custody, confiscation of money thus earned), but not their customers.

38 Uses that seem multi-occasional tend to be existential constructions without any prospective meaning at all: Yapon-chuud-iin ulamjlal yoson-d (...) vyendyetaa buyuu cusa-n dshdd ava-h gej bai-dag (Japan-PL.H-GEN tradition custom-DAT vendetta or blood-AT revenge take-FUT.p comp cop-hab.p) 'In Japanese tradition, there is [the concept of] vendetta or taking blood vengeance.' http://mongolian.ruvr.ru/2012_04_06/70860116/, retrieved 2013-12-28.

discourse- or stance-related device easily combines with a prospective form39.

(117) udahgui bid hoyor sui tavi-h ge-j bai-dag40 soon 1PL two tax place-FUT.P COMP-C COP-HAB.P 'Soon, the two of us are going to get engaged'. [...But the two of us are quarrelling more and more.]

(118) 400 jil-iin daraa jinhene El_Dorado-g inge-j 400 year-GEN after true PLACE-ACC do.so-C olo-h ge-j bai-dag=aa41

find-FUT.P COMP-C COP-HAB.P=POL

'After 400 years, we [Columbians] are [now] about to find the true El Dorado [through trading flowers]!'

(119) mongol-chuud bid tor-iin terguun=ee unen Mongolian-PL.H 1PL state-GEN head=RP truth hel-sn-ii toloo shoron-d hii-j ala-h ge-j say-PRF.P-GEN for prison-DAT do-C kill-FUT.P COMP-C bai-dag42

COP-HAB.P

[This world is complicated.] 'We Mongolians are about to put our head of state into prison and kill him because he spoke the truth'.

The uses in (117) and (118) refer to future events as indicated by the adverbials. The more complicated (119), similar to (96), alludes to one single past event: as the Democratic Party, next to putting the former president into prison for corruption (and thus preventing his reelection bid), also abolished the death penalty, it is improbable that

39 It is entirely possible that some instances of written -dag for uni-occasional events correspond to spoken -d(g)=iin, as a frequent use of yum in written language appears to be discouraged in school, and indeed yum appears to be much less frequent in written materials.

40 http://khalkhbuu.blog.gogo.mn/mcomments98671?page=45, retrieved 2013-12-28.

41 http://www.tsahimurtuu.mn/index.php/stories/2012-05-10-04-48-56/501-archive-story-294, retrieved 2013-12-28.

42 http://www.shuurhai.mn/?p=34195&replytocom=67832, retrieved 201312-28. The quote is from a comment (only shown on Google) to the letter of former president Enhbayar sent from his hospital bed (de iure from prison) and published on this webpage on 2013-03-18.

the part about killing is to be taken literally. More likely, it characterizes the subject as being inclined to act in this way at any given point in time.

5. -h

The participle in -h has a very particular distribution. It marked future in Middle Mongol and was used in finite position [Brosig 2014b], but is now extremely rare in absolute sentence-final position, accounting for only 12 of its overall 374 finite uses in the corpus. Synchronically most prominent is the use with diverse interrogative devices (n = 192) such as =uu (n = 123) and =iim=uu (n = 22) in polar questions and ve (n = 23) and =iin (n = 16) in content questions. Variants of =iin are also used in declaratives (n = 29) and can combine with non-finite aspectual forms (though n = 0); =iin may also precede the form bai-n and, rarely, baig-aa (n = 15+2). In another set of uses, -h can be used before particles originating from copulas (n = 27), the synchronic copula (n = 8), and before bol- including the fossilized bolvuu (n = 6+6).

In 5.1, I will discuss irreal uses in questions and guesses. 5.2 contains a discussion of -h before ge- and on its own. Finally, 5.3 presents data on -h bai- as well as the diachronically related -h(=iim)=san. -h bol-and -h=iin in positive sentences are not discussed in this paper.

5.1. -h with particles /clitics relativizing factuality

-h is the most neutral device for asking about the future. While most ofthe occurrences of -h in the corpus are intentional futures as in (120), predictive uses as in (121) are perfectly possible. -ch-h=uu, also illustrated in (120), expresses a wish of the speaker on his own part. =iim=uu as in (122) and (123) indicates some kind of presumption. In (122), for instance, the speaker presumes that he should prepare rice on the basis of the overall situation or a previous understanding, and just asks for confirmation.

(120) A: odoo yuu hii-h ge-j baig-aa=n, naraa

now what make-FUT.P COMP-C COP--DIR=MC NAME egch? elder.sister

J: odoo ah=n baicaa-tai huurG

now elder.brother=3POSS cabbage-COM plate hii-ch-h=uu? make -BOU-FUT.P=Q

A: zaza, za, baicaa-tai huurG... naag-uur=chin

well well cabbage-COM plate D.MED-PROS=2POSS hii-h=uu? make-FUT.P=Q A: 'Now what are you going to prepare, Ms. Naraa?' J: 'Shall I prepare a "dish with cabbage"? A: 'Ok, a dish with cabbage. Shall we prepare it around there (where you stand now)?'

(121) hi temtseen-ii ehn-ii shat=n margaash hi contest-GEN beginning-GEN step=3POSS tomorrow duusa-h=uu onoodor=uu? tnx43

end-FUT.P=Q today=Q thanks

'Hi! Will the first part of the contest end tomorrow, or today? Thanks!'

(122) en-iig hii-h=iim=uu? D.PROX-ACC make -FUT .P=MC=Q

[A:] 'Shall I prepare this?' [B: 'Oo, we probably don't need rice'.]

(123) en baicaa-g bugd-ii=n herch-h=iim=uu? D.PROX cabbage-ACC all-ACC=3POSS cut-FUT.P=MC=Q [A:] 'Shall I cut all of the cabbage?' [B: 'Yes, cut it all!']

Polar questions containing -h=iim=uu try to engage the addressee, so informants reject the use of the form in examples like (124) that do not relate to any particular addressee. The same holds for =h=iin in content questions, and as -h ve is considered very literary, speakers use -h=uu under such circumstances even in the presence of an interrogative proword as in (125).

(124) teg-ee=l ing-ee=l

do.so-C=FOC do.so-C=FOC

togt-jai-sn=aa

be.fixed-PROG-PRF.P=3POSS

minii omon togto-j 1SG.GEN before be.fixed-C bii haring ter-iig 1SG but D.DIST-ACC

43 https://m.facebook.com/story.php?story_fbid=10151867089219335 &id=27368964334, retrieved 2013-12-29.

av-h=uu, bai-h=uu ge-j bod-jii-song take-FUT.P=Q stop-FUT.P=Q COMP-C think-PROG-PRF.P 'Then, while [the balloon] was standing still before me, I wondered whether I should take it or leave it'.

(125) yaa-j biznis-iig hii-h=uu, yaa-j bii odoo do.what-C business-ACC do-FUT.P=Q do.what-C 1SG now am'darl=aa saiir-uul-h=uu{=ch' ge-d=iim=uu} life=RP improve-CAUS-FUT.P=Q=FOC COMP-HAB.P=MC=Q iim=l yuman-d odoo bii ih bod-jii-song such=FOC thing-DAT now 1SG much think-PROG-PRF.P 'How can I do business? How can I improve my life? Now about such questions, I was thinking a lot'.

Particles like baih 'probably, maybe' (< bai-h), bolvuu 'id.' (< bol-v=uu) and nee 'probably' (< =n'=ee, thus apparently a nomina-lization) are often used together with -h, as are epistemically stronger particles such as bailgui(=dee) 'certainly', illocutionarily more committing particles such as baihgui(=yuu) 'of course' (originating as a device for tag questions, but also implicating naturalness) and vii which expresses anxiety about a possible future course of events.

(126) taa buhen daraa=n yarilcalg-aas hara-h 2PL all after=3POSS conversation-ABL see-FUT.P baih=aa

IP=POL

'[Now these two are in a pretty drunk state.] You all will probably see it later from the conversation'.

(127) ter tal deer sanaa zovo-h=gui=l=dee,

D.DIST side on mind suffer-FUT.P=EX.NEG=FOC=IP zugeer yanz_bur-iin am 'tan horlo-h vii, esvel just diverse-GEN being harm-FUT.P IP or gemte-j bert-chi-h vii ge-j sanaa

get.injured-C get.hurt-BOU-FUT.P IP COMP-C mind zov-dg=iim=daa suffer-HAB.P=MC=IP

'[A family is peaceful enough,] I'm not worried about that. Just that diverse people might harm us, or that some of us might get injured, that I worry about'.

5.2. -h followed by ge- or in absolute final position -h is used together with the verb ge- 'to say' in several combinations. These have been described in the literature in some detail: the modal -h gesen yum is used to express willingness or desire [Song 2002] as in (128), whereas the more basic, non-finite construction -h gej with other verbs is used to express a purpose [Hashimoto 2004] as in (129). Neither construction is properly within the scope of this paper. -h also forms part of the Prospective as discussed in 3.3.5.

(128) uuchl-aarai oor-oos=chin neg zuil asuu-h excuse-IMP.HON self-ABL=2POSS one matter ask-FUT.P ge-s=iim=aa

say-PRF.P=MC=POL

'Excuse me, I want to ask you one thing'.

(129) uildverchn-ii evlel bol Ganc-hang zasagll-ruu bish worker-GEN union TOP only-DIM authority-ALL ID.NEG zasaglal hyana-h ge-j yav-dag

authority check-FUT.P COMP-C go-HAB.P

bai-h=guu=yuu

COP-FUT.P=EX.NEG=POL

'You see, the workers' unions don't only go to the authorities, they go there to check on them'.

As mentioned in 3.1, -h (in the scope of ge-) can be used to refer to lexical verbs as such, and most dictionaries list verbal lemmas in -h. It can occupy the final position in a sentence that consists of two equated nominal phrases as in (130), but this use is not finite in the proper sense. An actual finite use of -h on its own is attested, though. In (131) and (132), for some unclear reason, the clause with -h indicates the purpose for an adjacent preceding clause; in (133) and (134), unrelated events are enumerated:

(130) taa nar-iin nileen ih hii-deg zuil bol 2PL PL-GEN rather much do-HAB.P act TOP zahn-ii-h=aa daavuu-g induude-h collar-GEN-AT=RP cloth-ACC iron-FUT.P (paraphrase used during elicitation)

'Something that you are doing quite a lot is ironing the cloth of your collar'.

(131) yapon-d ydrdn her bolomjiin ajil bai-n? Japan-DAT in.general how.much acceptable work COP-DIR sur-aad=l magadgui ddr-iin-h=dd surgalt-iin learn-C=FOC maybe self-GEN-AT=RP study-GEN tdlbr-iig ajil hiig-ee=l olo-h

fee-ACC work do-C=FOC find-FUT.P

'How adequate are the jobs in Japan? If one is just studying and maybe wants to earn one's tuition fee from that?'

(132) B uut av-aad ir-0 ge-sii=shdee, NAME bag(plastic.etc) take-C come-IMP say-PRF.P=IP dvdd=n hog hii-h

grantfather=3POSS waste put-FUT.P

'I said to her: "B, bring a plastic bag". So that her grandfather (i. e. 'you') can put the waste there'.

(133) aa yaag-aad ge-vel ter belgiin am'darl-d INTERJ do.what-C say-COND.C D.DIST sexual life-DAT ert or-n ge-deg=chen dsd-j

early enter-DIR COMP-HAB.P=STC grow-C torn'-oo=gui huuhed ter bol-oo=gui

grow.up-—DIR=EX.NEG child D.DIST become-—DIR=EX.NEG asuudal-ruu ert or-sn-oor=oo ih evguu

problem-ALL early enter-PRF.P-INS=RP much unpleasant baidl-aar halira-h, tiim=ee, hicheel

situation-INS drop.back-FUT.P so=POL study surlaG-d=aa muuda-h, eej aav-iin huuhed study-DAT=RP worsen-FUT.P mother father-GEN child bai-h ter saihan orchn-oos=oo hdndiird-h,

COP-FUT.P D.DIST nice surrounding-ABL=RP alienate-FUT.P za teg-eed... hamg-iin ayuul-tai=n belgiin

well do.so-C all-GEN danger-COM=3POSS sexual zam-iin dvchdn-d nerve-gde-j ir-j baig-aa way-GEN disease-DAT affect-PASS-C come-C COP-—DIR 'Well, this is so because, if it enters into sex life early, the child that has not yet grown up, by early having entered into those not yet timely problems, slips off [the right path] in a very awkward manner, gets worse in his studies, alienates himself from the

pleasant environment in which he is his parents' child, and, most dangerously, is affected by sexually transmittable diseases'.

(134) zugtaag-aad eserguucel uzuul-eed, en bol elbeg

flee-C resistance show-C D.PROX TOP abundant

tohoild-dog uzegdl=l=dee en cagdaag-iin

happen-HAB.P phenomenon=FOC=IP D.PROX police-GEN mords-iig huul-j haya-h, cagdaa-g zodo-h, epaulette-ACC strip.off-C throw-FUT.P police-ACC hit-FUT.P tiim=ee so=POL

'They flee or show resistance... this is a very common phenomenon. To rip off and throw away the epaulettes of a policeman, to hit a policeman, right?'

The causal (131) and (132) seem acceptable to most, while not all informants. Both could be paraphrased by changing the order of the two clauses and linking them by the postpositional construction -h-iin tul. So the postposed sentences with -h might actually constitute a kind of explicit afterthought. The enumerations in (133) and (134) differ structurally to some degree. The enumeration in (133) is completed by a different, more clearly finite verbal form. The enumerated events as such are not connected, neither temporally nor by common agents, whereas converbal forms would imply that. 3 of 4 informants reject the use of -h for the last verb nervegd-. In (134) and similar examples, tee/tiimee 'right?' is used after or, in one single case, somewhat before the verb in -h. The enumerated events are conceived of as exemplary, non-exhaustive and thus even as less permanent than would e. g. be implicated by -dag.

5.3. -h in complex forms with the copula

Due to the sheer rarity of most complex forms that feature -h, the treatment in this section will have to be incomplete. -h, if followed by a past copula, is used to express an irreal situation that would have come about in the past if only certain conditions had obtained (cf. [Cinggeltei 1959: 18-20, 43]). This meaning differs from -h san, which originated from the structurally equivalent -qu a-ysan and expresses a wish about a present situation on the part of the speaker. -h can also be followed by the present copula (in the corpus always with interspersed =l, but it can be absent at least in written materials). It is most commonly used with modal verbs such as bol- 'to be possible', chad- 'to be able',

zoh'- 'to be appropriate' and expresses future developments that might come about given a present situation [Cinggeltei 1959: 18]. Song [1997: 339-340] attributes counterfactual past obligation to -h bai- plus past copula.This certainly holds for his own examples, but not for (135). However, it turns out to be applicable also to present obligations as in (137) and (138) below.

(135) B ah tend bai-sang bol bid hed-iig NAME elder.brother there COP-PRF.P if 1SG some-ACC gar-ga-j dg-ddd teg-seng bol ineedtei yum exit-CAUS-C give-C do.so-PRF.P if funny thing bolo-h bai-sang

become-FUT.P COP-PRF.P

'If B had been there, we would have put several [of such beers] on the table, and then it would have been funny'.

(136) tdmdr bol soronz-oor inge-h sen metal TOP magnet-INS do.so-FUT.P MP

'If it were iron, I'd just draw it out with a magnet'.

(137) Goy-oor or-uul-j bol-h=l beautiful-INS enter-CAUS-C be.possible-FUT.P=FOC baig-aa=n

COP-—DIR=MC

[D: 'No, no, one has to insert that very beautifully'. B:] 'One could indeed insert it beautifully [, yes]'.

(138) bi daraa-h heden zdvldmj-uud-iig dgd-h bai-na44 1SG after-AT some advice-PL-ACC give-FUT.P COP-DIR 'I would give the following advice'.

While (135) expresses an apparently unemotional guess about a past possibility, the speaker of (136) expresses his frustration that he cannot proceed with the task at hand as he had hoped. Like (135), the examples (137) and (138) express potential situations, but one that could still be realized. In (137), the people in charge do not do as they could and should, to the speaker's frustration. In (138), the writer

44 B. Taivan. Udaa daraalsan muu medee: DOW JONES-g 360 negjeer

unagalaa. http://orloo.info/readfull/ride/12-news/business-economy/3265-----

dow-jones--360--a, 2008-06-27, retrieved 2014-01-22.

realizes that there is some advice s/he should give, though there is not more s/he can do. It seems that this latter example only represents a number of somewhat fixed expressions in formal contexts. A colloquial speaker, as one informant suggested, might rather use forms like ogmoor bain (wish) or ogii (intention) in such a first-person context.

The corpus does not contain complex forms ending in -h, but -jii-h=uu was found to be viable in (139) if the sentence is interpreted as addressed to a second person, then probably forming a friendly, informal request. In contrast, -jii-n=uu as used in the actual sentence is a question whether the subject is working within schedule. Due to lack of examples, no proper description is attempted here.

(139) manai hun hool hii-j

1PL.GEN person meal make-C amj-jii-h=uu? (constructed)

succeed.in.time -PROG-FUT. P=Q 'Would you try to prepare the meal in time?'

6. -aad

Finite -aad expresses a resulting or pertaining present or past situation that requires or required activity on the part of a participant or discourse-participant. It has non-evidential resultative semantics, with the aspectual interpretation depending on actionality and the temporal situation on overall context.

The suffix -aad is known as a converbal suffix, but like many converbs, it has more than one function. In the terminology of Nedjalkov [1995: 97-98, 105], -aad would be trifunctional, namely as a converb (adjunct, adverbial), infinitive (depending on auxiliary verbs) and finite verbal suffix, while being unable to express nominal attributes or nominal actants. Standard classifications (e. g. [Onorbayan 2004: 272]) account for the first two uses, grouping -aad together with -j, -n and sometimes -saar, i. e. the converbs that can be used before bai- in literary registers. Finite uses of -aad, on the other hand, have to my knowledge not been described for Khalkha. Yet, sentence-final uses of -aad are not uncommon. I annotated around 200 tokens as potentially finite, but as I did not work on all of these with native speakers, some of them would have to be discarded. At any rate, I confirmed 49 tokens as finite: 23 instances of (-ch)-aad and 15 instances of (-ch)=aad=l

in absolute sentence-final position, as well as 6, 2 and 3 instances in which -aad is followed by the illocutionary devices baihgui=yuu for explanatory tags, te for confirmation-seeking tags and =uu for polar questions. In Khalkha, these devices can only be used after sentence-final predicates, so a suffix -aad that directly follows them must be finite.

As a converb, -aad links up sequential events such as in (88) above. Events marked by -aad are syntactically subordinate, but not semantically modifying, a syntactic relationship sometimes called cosubordination [Foley, Van Valin 1984: 241-242; Johanson 1995: 321-323]45. (83) illustrates the use of -aad with the auxiliary ehl- 'to begin'. In contrast, (140) is already intermediate between a converbal and a finite use.

(140) zaa en-iig=ee zuuhan-d hii-hleer,

well DEM.PROX-ACC=RP oven-DAT put-CNS.C sous=maan' dtdgr-ddd, teg-eed

sauce=1PL.POSS thicken-C do.so-C

deer-ees=n, bysalg=aa taraag-aad

on-ABL=3POSS cheese=RP distribute-C

(...) jaahan deer-ees=n odoo talh=aa cac-aad

a.little on-ABL=3POSS now bread=RP sprinkle-C 'Well, through putting this into the oven, the sauce has thickened, now I distribute some cheese and it...' [B: Pepper?] 'Oh yes!' [C: And breadcrumbs on top of it?] 'Yes, yes! Now I sprinkle some bread on it...' [It is really nice to have great cooks as assistants].

45 Bazarragachaa [1999: 98-108] assumes altogether six clause-combining functions of -aad: (1) Several actions succeed each other; (2) The first action causes the second action. These two make up 7,5 pages of his 9 pages of examples; (3) The first action describes the way in which the second action takes place. Very few examples, but I accept some as valid. Those have in common that the manner verb is always directly adjacent to or only one word remote from the main verb; (4) Several events take place simultaneously. The examples are based on inceptive verbs or can be interpreted as sequencies or instantiate the construction -aad time.noun bol-PST; (5) The first event is the opposite event of (i. e. somehow contradicts) the second; (6) The first event is the goal of the second event. These two are only possible with the verb stems bai- and ge-, respectively, and thus represent no general usage pattern of -aad. Thus, I take these first two functions as the only ones that are lexically unrestricted and allow both verbs to have separate argument structures.

In (140), a cook describes how he prepares a meal, connecting everything he does with an additional -aad and thus indicating that he is not done yet. He does not actually end the sentence, as he switches to a distantly related topic in the end. If there is any difference between the finite and the non-finite usage, it only becomes clear when the next sentence starts, and the time of deliberation that the speaker puts in-between his utterances makes it likely enough that he starts with something new. The corpus also contains one sentence uttered by two speakers who take turns ending each clause with -aad, while not actually bringing it to a close. So this use is finite, but not deliberately so: it is rather implicated that additional information is to be added. Among deliberately non-finite uses, those most closely resembling (140) are resultative or perfect uses based on telic predicates:

(141) en[d]=cheen=n yas urg-aad buruu edg-eed here=?STC=3POSS bone grow-C wrong heal-C

[A: 'So you were hurt rather seriously?' O: 'Yes.] Here, a bone has grown and healed in a wrong way'.

(142) gue guu hoNg-oor=n av-aad EX.NEG EX.NEG day&night-INS=3POSS take-C

[A: 'No, we cannot go to anybody's home'. B: 'Why not?' A: 'Then it wouldn't make a difference if we take two hours at a hotel'. B:] 'No, no, I mean if we take it for a night...' [A: 'The two of us only have two hours'.]

(143) ehl-eed ail-d or-j ir-seng hun yum begin-C family-DAT enter-C come-PRF.P person MP bai-n, tavagn-ii idee ams-aad

COP-DIR plate-GEN food taste-C

[A: 'What shall I do? If you say, "Peel the potatoes! Break open the eggs! Clean the floor!", then I'll just clean it'. U: 'Well, anyway.] First of all, you've just come into a family, so first taste from [the food on] the plates'.

The process of healing in (141) is probably complete, but the imperfect result requires further measures, something not directly expressed by a Continuative-Resultative form. In (142), -aad is best translated as a conditional. It differs from the conditional converb -val in that the latter would indicate a mere suggestion, whereas -aad

indicates a rather strong proposal. In (143), the speaker is suggesting that any other activities might only be undertaken in the situation resulting when the addressee has complied with standard customs of welcoming somebody and tried some of the food that the host family provided. So something is to happen after the event marked with -aad, and this is deliberately evoked by closing the sentence with -aad. As in its non-finite use, -aad only denotes aspect, but not tense, and as no tensed matrix verb exists, the time of the event must be identified from context. Still, the few cases of -aad with future time reference attested in the corpus such as this one are all based on accomplishments.

A finite use with -aad is more frequent with undelimited states and with activities. In some cases, reference is made to situations confined within the past or present:

(144) bii=ch gaih-aad 1SG=FOC be.surprised-C 'I was surprised, too'.

(145) B: chii gadaa gar-aad uil-aad bai-sang shd

2SG outside exit-C cry-C COP-PRF.P IP D: ter=chen soht-oo=ch=gui=ee

D .DIST=STC get. drunk-—DIR=FOC=EX .NEG=POL C: jaahan bayarl-aad, te

a.bit be.happy-C TAG B: setgel=n hddl-ddd

mind=3POSS move-C B: ['You were a bit drunk. Really;] you got outside and were crying'. D: 'That wasn't drunk yet!' C: 'He felt a bit happy, right'. B: 'His mind was moved (lit. moving)'.

(146) taa nar odoo yuu yar'-aad baig-aa=n? bii 2PL PL now what speak-C COP-—DIR=MC 1SG gad-uur zar-dag yamar baraa bish=dee outside-PROS sell-HAB.P what.kind good ID.NEG=IP yuu solior-ood?

what go.crazy-C

'What are you talking about? I am not some kind of goods that you sell outside! Why are you talking such nonsense?'

Before (144) was uttered, the speaker ended a narrative with the marker -v that seems to be used for marking surprising events with which the speaker is already familiar. The addressee then expressed his surprise with an intonation-bearing bilabial interjection, upon which the speaker uttered (144). Informants are not inclined to assume that the speaker is still surprised at the time of speaking. Rather, the speaker indicates by using -aad that he did not know then how to act. In (145), the first sentence establishes a past situation. The resultant situations indicated by the two forms in -aad, i. e. that the subject was happy or deeply moved, are to be understood as necessary conditions for the subsequent crying. (146) relates to a present situation. A pimp and a customer had been bargaining over a prostitute's price, and she is objecting to that. The sentence with -aad neatly parallels the Continuative in the same utterance, but is perceived as less insistent and aggressive.

However, some situations marked with -aad on atelic predicates outlast a single occasion, though no claim to habituality or intrinsicality is made:

(147) en-iig haichal-ch-maar baig-aa biz D.PROX-ACC cut.with.scissors-BOU-WISH.P COP-—DIR IP shal teneg hara-gd-aad

INT stupid see-PASS-C

'You'll definitely want to cut this off. It looks extremely stupid'.

(148) teg-eed yoron dotood jurman-d=aa do.so-C in.general inner system-DAT=RP zaa-cang bai-s=iin manai anG indicate-BOU.PRF.P COP-PRF.P=MC 1SG.GEN class garutas horigl-ood

mobile.phone prohibit-C

[There are advantages and disadvantages if young people have mobile phones.] 'But actually it has been decided in our internal system. In our barracks, mobile phones are forbidden'.

In (147), -aad is again used to introduce a reason. Informants indicate that the situation is claimed to have lasted from the time when the speaker spotted it up to the time of the utterance. Evidential nuances inherent in the Continuative e. g. to the effect that the speaker just made up his mind or already holds this opinion for some time are avoided. In (148), several informants perceive no reference to the inception

clearly expressable for this presumably inceptive-stative predication by means of the Continuative-Resultative: the prohibition simply holds. Other informants suggest a recent enactment of a new rule. Be it as it may, the more important issue for the informants was that the soldiers have to relate to this prohibition, e. g. by complying or circumventing it46.

7. =l

The Restrictive Focus clitic =l is sometimes used to prevent a resultative interpretation of -aad. In its simplest function, this clitic attaches to nouns, emphasizing that the reference is restricted to that particular entity and none other. In this function, it is paralleled by =ch which expresses that reference holds even for this entity, as it holds for some others:

(149) bi=ch / =l yav-ya

1SG=FOC/=FOC go-VOL

'Let me go, too'. 'Even I want to go'. 'I'd better go'.

/ ' Only I want to go'. 'Let me go alone'. 'I for my part want to go'.

[Kullmann, Tserenpil 1996: 348]

The restrictive focus clitic =l can interact with the converb -aad by rendering a resultative interpretation improbable, so that the phase referred to by -aad=l will only be a lexicalized phase of the verb, while any reference to a lexicalized border is prevented. On the other hand, =l can also simply express restrictive focus, leaving the resultative meaning of -aad intact. For this, it does not matter whether -aad is followed by bai- (n = 25) or not (n = 73). When =l combines with participles (n = 9), only a restrictive interpretation is possible47, and at least most of the instances of -j=l bai- (n = 12) are restrictive, too.

46 The only complex construct ending in -aad attested in the corpus is Progressive -jaig-aad. All of these examples seem to be interpretable as ellipsis (as they describe a background situation to an event sometimes explicitly mentioned in the context), and there are only 7 tokens, so I refrain from analyzing them here.

47 While Kullmann and Tserenpil [1996: 348-349] do not formulate any kind of analysis, their accurately translated examples basically suggest the same analysis as I am proposing here: ir-eed=l baina 'keeps coming', unsh-aad=l bai-na 'I'm still reading', endir-eed(=l) bai-na=l=daa 'Well yes, he has come here, (but...)', ter end ir-sen=l bai-na 'He came here {indifferent attitude}'.

The clitic combination =l=daa can follow variants of =l bai- (n = 12) and is then harmonic with a restrictive meaning. If -ch- is used in the same predicate as -aad=l, it retains its border-oriented meaning, so that =l must be restrictive48.

In the first examples, -aad prevents a resultative interpretation:

(150) ... zaa en bor halzan

well D.PROX brown with.white.patch.on.forehead uree uragshl-aa=l bai-n

stallion(3to5years) advance-C=FOC COP-DIR 'Well, this brown stallion is still advancing consistently'.

(151) ... ge-e=l gad-uur yanz yanz-aar

COMP-C=FOC outside-PROS manner manner-INS yar'-aa=l bai-n=shdee speak-C=FOC COP-DIR=IP

["Badar-Uugan has stopped associating with normal people. He has become a heavenly being"], 'people outside are talking like this'. [But I don't get this impression.]

(152) manai nutag devsger deer bai-jaig-aa nogoo=l 1PL.GEN country terrain on COP-PROG-—DIR other=FOC hovord-song saihang am't-d-iig al-aa=l become.rare-PRF.P beautiful creature-PL-ACC kill-C=FOC ustg-aa=l, te?

extinguish-C=FOC TAG

'They are killing and exterminating those beautiful threatened creatures that live in our steppes, aren't they'.

In (150), which might turn out to be either a directed activity or an accomplishment, informants emphasize that the race horse slowly keeps making progress. The slowness might either be appreciated or disapproved of by the speaker, and it is not clear whether it can actually improve its overall position. Without =l, the speaker would

48 =ch can in principle attach to participles and even to -aad, most often so in the construction (av)-aad=ch baig-aa yum shig 'it even seems that X is taking continuously' (non-finite due to shig 'like, as if'). It can attach to finite verb forms when followed by the complementizer verb ge-. However, all these functions are related to information structure only and do not interact with aspectuality.

rather comment on the current state, i. e. that the horse has improved its position somewhat, but it is not clear whether it will continue to improve it. In (151), the speaker focuses on the content of the rumour. The absence of =l would rather emphasize the speaker's anger at the unreasonable spreaders of the rumour itself. This latter meaning is most easily derivable if the speech content is taken to be a result rather than an ongoing process. While (152) refers to a continuing, not concretely delimited process, the absence of =l would confront the addressee with accomplished facts.

(153) below is ambiguous between a continuative and a restrictive resultative interpretation. (154) and (155) illustrate the restrictive usage with the participle -sang:

(153) undse=n ajl=aa med-ee=l

basic=AT work=RP execute.as.responsible-C=FOC

teg-ee=l

do.so-C=FOC

[There's no point in drinking until one loses consciousness. Probably one would drink alright. How can one sit around just looking at it?] 'First I get done with my actual work, then I drink. / As long as I can get done with/exert control over my actual work, I drink'.

(154) pizdaa uul=n=ch teg-ii ge-j INTERJ original=3POSS=FOC do.so-VOL COMP-C

bod-oo=gui bolovch teg-seng=l_baig-aa

think-—DIR=EX .NEG although do.so-PRF.P=FOC COP-—DIR yum=daa49

MC=IP

'What the fuck! Originally, I hadn't intended to do so, but (regrettably) I have done just so'.

(155) ter=chin=ch_geseng bod-dog bodl=oo uuran-d=aa D.DIST=STC=FOC think-HAB.P thought=RP anger-DAT=RP

49 http://gawaas.blogspot.de/2G08_04_01_aichive.html, retrieved 2014-01-03.

gar-ga-j hel-seng=l baig-aa

exit-CAUS-C say-PRF.P=FOC COP-—DIR

'Even he, too, has expressed his opinion only in anger'. [If one

considers this, try to talk it over with him. Only if you can't

forgive each other, it'd be better to divorce.]

(153) may either have an iterative interpretation where the speaker claims that she customarily accomplishes her core job while drinking to some modest extent. Alternatively, the restrictive interpretation 'only after I have taken responsibility for my core work, only after having done so' is possible. Native speakers are aware of this ambiguity. In (154), the predicate receives an emphatic restrictive focus. Here, emphasis is more important than restriction (just ~ precisely), but an additive =ch 'even' would still not fit the context. In (155), the speaker tries to downplay the event by inserting =l (i. e. your boyfriend said so only when angry, but you don't know what he would say if he was not).

In the examples below, -ch- enforces a resultative interpretation, rendering =l restrictive:

(156) odoo ehn-ii hoyor hurgen-tei now beginning-GEN two son.in.law-COM bol-ch-oo=l bai-n become-BOU-C=FOC COP-DIR

[In any case, God has blessed me with four daughters. When I was young, I still used to think that God doesn't bless me with any boys, but it's not like this. Somebody who has girls gets sons, right?] 'Now I have got the first two sons-in-law'. [The time for the next two hasn't come yet.]

(157) jaahan biy sergee-h, te, neg dorov dorvo-n a.bit body wake.up-FUT.P TAG one four four-AT

kilmiter gazar gui-j ir-ch-ee=l_teg-ee=l

kilometer space run-C come-BOU-C=FOC do.so-C=FOC '[Because I currently don't have a contest,] in order to activate my body, I have just run four... four kilometers and come back here'.

In (156) and (157), the event has taken place once and is completed, with the result obtaining at the time of speech, while =l somewhat diminishes the appreciation of the result. =l in (156) expresses

a certain impatience to get the next two sons-in-law. The absence of =l would indicate complete satisfaction with the two he already has and the absence of -ch- that even his first two daughters are only in the process of getting married. In (157), the speaker either holds that he exercised rather little, or he indicates that, after having done so, he has nothing more to do.

8. Conclusion

In this paper, I have tried to give an overview of finite present tense forms in Khalkha on their own and in combination with complex aspectual forms. New findings include:

• a basically unified analysis of simple -n as marking an occasionally realized potential action;

• an evidential explanation of the opposition between the Potential -n and the Modal Resultative -aa as Direct vs. Indirect in complex aspectual forms;

• a first analysis of the mono-occasional function of the Habitual -dag;

• a first description of sentence-final -aad.

New details were also added to what was known about the function of =l and -h, and descriptions of the function of complex aspectual forms were given that, elaborating on the pioneering work of Cinggeltei [1959], provide a better overview over the functional inventory of Mongolian, hopefully superseding rationalistic approaches such as those of Saruul-Erdene [1998] and Onorbayan [2004: 289-290] who started with semantic notions and filled them with linguistic material instead of taking the system of forms as the basis. Conversely, a coherent analysis of the formal inventory even enables researchers to address semantic notions adequately. For instance, the analysis provided above suggests that present imperfectivity is structured in the way sketched in Table 2:

Table 2. The present imperfective domain in Khalkha

Continuative Progressive Habitual Potential

neutral -aad -dag -n

direct -aad bai-n -jii-n -dag bai-n

indirect -aad baig-aa -jaig-aa -dag baig-aa

In the system of this table, complex aspectual forms ending in -dag do not form a category of their own, but merely add additional internal structure to an event. In principle, one can form more complex forms such as -j bai-dag bai-n that would add evidentiality even here. Paucity of data prevents any conclusions about simple evidential uses of -dag. Past imperfectivity could presumably be included into the table by adding Factual, Firsthand and Non-firsthand variants on the Continuative, Progressive and Habitual levels. As I did not analyze markers of past imperfectivity, the precise semantic extensions of these forms are hard to determine, though.

Given the relatively limited scope of this paper, it is not yet possible to address the expression of perfectivity (for which a discussion of simple past forms is required in addition to the perfect and resultative forms discussed here) or future (for which simple past forms and the Voluntative mood play a role next to forms such as -n, -h, -h bai-, -h gej bai-).

There are a number of questions that I did not have the space or time to cover here, some of which I hope to address in the near future. The functions of =iin (limited are accounts given by Monh-Amgalan [1998] and Kang [2003]) and -ch- (no serious description available at all) closely relate to the core system of tense, aspect, modality and evidentiality in Mongolian, and their semantic range has so far only been scratched. The construction PTCPyum bai-n might deserve particular attention50. An integrated analysis of -h is still missing, and constructions based on bol- have been left out. The overall instantiation of evidentiality and epistemic stance [Mushin 2001] in Khalkha still has to be clarified. There are pioneering studies on the function of Khalkha participles in negated [Hashimoto 2007] and attributive and nominalized [Song 1997] position, but a unified analysis of the paradigmatic limitations of their TAME meaning is still missing.

On the other hand, a number of other research problems were not addressed due to limitations of the corpus or my methodology. My decision to investigate actual spoken language meant working with rather limited data, precluding investigations into infrequent constructions

50 Xueyan [2013: 183-184] treats it as a very general "indirect" marker (she seems to mean "non-firsthand" as she classifies -sang bai-n as indirect), but it is not frequent enough to fit into a paradigm, and it seems to be semantically more specific, maybe general reasoning.

such as -aad bai-h=uu/-jii-h=uu, limiting access to semantically rare uses of attested morphemes and morpheme combinations such as -dag bai- and -aa (other than in baig-aa) as well as to more complex structures such as those in Table 3:

Table 3. Finite Khalkha TAE forms with two copulas (from Google)

stem c/p+cop p/c+cop finite suffix

V -aad/-j/-sang bai- -sang/-dag/-j bai- -n /-dag /-sang

-dag bai- -sang bai- -n/-dag

The table leaves out -laa and -jee (the distribution of which is thought to be similar to -sang in its finite use) and implies that the same form of the copula could be used several times within complex predicates which is not the case for adjacent positions. Otherwise, all combinations starting with -aad, -j and -sang are acceptable, and -aad (excluding its final use) is restricted to first position. baidag bai- must always be followed by -sang, so Habituals cannot directly fall into the scope of Progressives. Combinations with three copulas are also possible. As the first morpheme is most often a converb and the last element mostly -n, this seems to be close to or even the maximal number of copulas usable. Complex forms are thought to be mostly compositional, but idiosyncrasies can be found at the two-copula-level when two converbs are involved:

(158) bur gutr-aad bai-j bai-san yum51 completely be.depressed-C COP-C COP-PRF.P MC '[Thank you,] I was really being depressed; [because of having read this I am getting better]'.

(159) teg-eed gaih-aad skafn-ii haalga do.so-C be.surprised-C cupboard-GEN door ongoilgo-son=chin' ah=maan' boov-oor=oo open-PRF .P=2POSS elder.brother=1POSS dick-INS=RP

51 Sanaa saihan bol zayaa sain, http://184.107.215.140/news/show/ 4190, 2012-09-26, retrieved 2014-01-22. From one comment.

togl-ood baij bai-san52 play-C COP-C COP-PRF.P

[One day my two brothers and I were playing. Then, my older brother had disappeared.] 'I wondered where he might be, and when I opened the cupboard, he was in there playing with his dick'.

There seems to be an added notion of locality or inactivity in (158) such as 'I was here feeling depressed' (vs. 'I was engaged in feeling depressed'), more visible in (159), but more research is needed for any reliable conclusion.

Aktionsart is another issue. While I took into account verbal semantics as far as the corpus material allowed, it was not possible to properly investigate different actional structures as did for example Molochieva [2010: 182-211] for Chechen or Tatevosov [2002, 2008] for Bagwalal, Mari, Tatar, and Karachay-Balkar. That kind of research would either require risky approaches such as elicitation only or rather large corpora, but it is obviously needed for a full understanding of aspectuality in Khalkha.

A final issue that would require a different methodology is the difference between -n and -dag. The usage of -n is known to vary substantially between different major dialects, and there must be different shades for different regions of one dialect area as well. Moreover, even informants were prone to ascribe some uses of -n to older speakers. Using a corpus with mainly middle-old speakers not properly controlled for dialect and mostly female student informants around 25 from any Khalkha dialect area is no particularly effective approach here. I would predict that a corpus of 20 hours with speakers between 20 and 25 born and raised in Ulaanbaatar would yield significantly clearer results.

Abbreviations

1h — firsthand evidential marker; 2h — non-firsthand evidential marker; at—attributive case (-n) /attributive form for already case-marked words (-h); bou — boundary-emphasizing aktionsart modifier; c — converb; coll — collective (numeral); cns — consequence (for the converb forms -hlaar and -haar); d — demonstrative; dim — diminutive; dir — potential/direct evidence

52 http://betheman.blog.banjig.net/index.php?bdate=2007-01-01& edate= 2007-12-31, Minii ijil huisten bolson n', 2007-10-16, retrieved 2014-01-22

marker; -dir — resultative/indirect marker; ex.neg — existential negation; foc — focus marker; ideo — ideophone; id.neg — identity negation; int — intensifier; interj — interjection; ip — interlocutionary particle or clitic; hab — habitual; med — medial (close to addressee); mc — modal clitic (=iin ~ yum); MP — modal particle; name — proper name; ord — ordinal number; p — participle; place — place name; pl.h — plural restricted to humans; pol — polite; pros — prosecutive case (on nominals), prospective (on verbs); res — derivative resultative suffix; rhet — rhetorical; rp — reflexive-possessive; stc — stance-marking clitic; tag—tag question particle; top — topic; vol — voluntative mood.

Bibliography

Aikhenvald 2004 — A. Aikhenvald. Evidentiality [Oxford Linguistics]. Oxford:

Oxford University Press, 2004. Bazarragchaa 1999—M. Bazarragchaa. Mongol helnii üiliin tiin yalgal [Mongolian

converbs]. Ulaanbaatar: Mongol ulsiin ih surguul', 1999. Bertinetto, Lenci 2012 — P. M. Bertinetto, A. Lenci. Pluractionality, habituality and gnomic imperfectivity // R. I. Binnick (ed.). The Oxford Handbook of Tense and Aspect [Oxford Handbooks in Linguistics]. Oxford: Oxford University Press, 2012. P. 852-880. Breu 2005—W. Breu. Verbalaspekt und Sprachkontakt: ein Vergleich der Systeme zweier slavischer Minderheitensprachen (SWR/MSL) // S. Kempgen (ed.). Slavistische Linguistik 2003: Referate des XXXIX. Konstanzer Slavistischen Arbeitstreffens, Bamberg 15.-19.9.2003 [Slavistische Beiträge 442]. München: Sagner, 2005. S. 37-95. Brosig 2009a—B. Brosig. The function of the suffix -na in Khalkh Mongolian //

Altai Hakpo 19, 2009. P. 87-112. Brosig 2009b — B. Brosig. Die satzfiniten Aspektmarker des modernen Khalkha-

Mongolischen. MA Thesis. University of Bonn, Bonn, 2009. Brosig 2012 — B. Brosig. "Bilee" sul ügiin utga, hereglee [The meaning and usage of the particle "bilee"] // Hel zohiol sudlal 5 (37), 2012. P. 10-18. Brosig 2014a — B. Brosig. The tense-aspect system of Khorchin Mongolian // P. Suihkonen, L. Whaley (eds.). On Diversity and Complexity of Languages Spoken in Europe and North and Central Asia [Studies in Language Companion Series 164]. Amsterdam — Philadelphia: Benjamins, 2014. P. 3-65.

Brosig 2014b — B. Brosig. The aspect-evidentiality system of Middle Mongol // Uralo-Altaiskie issledovaniia / Ural-Altaic Studies 13, 2, 2014. P. 7-38. Byambasan et. al. 1987 — P. Byambasan, C. Önörbayan, B. Pürev-Ochir, J. Sanjaa, C. Janchivdorj. Orchin cagiin mongol helnii ügzüin baiguulalt [The structure

of [verbal] morphology of contemporary Mongolian]. Ulaanbaatar: Shinjleh uhaanii akadyemi, 1987.

Caodaobateer 2007 — Caodaobateer. Kalaqin mengguyu yanjiu [A grammar of Kharchin Mongolian]. Beijing: Minzu chubanshe, 2007.

Cinggeltei 1959 — Cinggeltei. Mongyol kelen-u tusalaqu uile uge-yin tuqai [On auxiliaries in Mongolian] // Obor Mongyol-un yeke suryayuli-yin erdem sinjilegen-u sedkul. Gun uqayan neyigem sinjileku uqayan 1, 1959. P. 1-103.

Croft 2012 — W. Croft. Verbs: Aspect and Causal Structure [Oxford Linguistics]. Oxford: Oxford University Press, 2012.

Dahl 2000 — O. Dahl. The grammar of future time reference in European languages // O. Dahl (ed.). Tense and aspect in the languages of Europe [Empirical approaches to language typology 20: EUROTYP 6]. Berlin — New York: Mouton de Gruyter, 2000. P. 309-328.

Depraetere 2011—I. Depraetere. Towards a more explicit taxonomy of root possibility in English // English Language and Linguistics 15, 1, 2011. P. 1-29.

Dugarova, Jaxontova 1988 — G. S. Dugarova, N. S. Jaxontova. Resultative and perfect in Mongolian // V. P. Nedjalkov (ed.). Typology of resultative constructions [Typological studies in language 12]. Amsterdam: Benjamins, 1988. P. 209-220.

Ebert 1999 — K. Ebert. Degrees of focality in Kalmyk imperfectives // W. Abraham, L. Kulikov (eds.). Tense-aspect, transitivity and causativity: Essays in honour of Vladimir Nedjalkov [Studies in language companion series 50]. Amsterdam: Benjamins, 1999. P. 323-339.

Foley, Van Valin 1984 — W. Foley, R. Van Valin, Jr. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press, 1984.

Haan 2001 — F. de Haan. The place of inference within the evidential system // International Journal of American Linguistics 67, 2001. P. 193-219.

Hashimoto 1995a — K. Hashimoto. Progressive in Mongolian // T. Sakutaro, A. Kojiro, M. Riichiro (eds.). In honor of Nobuyuki Higashi: Papers contributed on the occasion of his sixtieth birthday, September 4, 1995. Tokyo: Kenkyusha, 1995. P. 333-351.

Hashimoto 1995b — K. Hashimoto. Shukanso [The Habitual] // Muroran kogyo daigaku kenkyu hokoku (bunrihen) 45, 1995. P. 35-67.

Hashimoto 2004 — K. Hashimoto. Metarepresentational motivations — the grammaticalization of the verb of saying ge- in Mongolian // Tsukuba English Studies 22, 2004. P. 35-49.

Hashimoto 2007 — K. Hashimoto. Hitei no rensa — Mongorugo no hiteihyogen no kino [The negation linkage — The functions of negative expressions in Mongolian] // Ninchikagaku 5, 2007. P. 49-74.

Johanson 1971 — L. Johanson. Aspekt im Türkischen: Vorstudien zu einer Beschreibung des türkeitürkischen Aspektsystems [Studia Turcica Upsaliensia 1]. Uppsala: Almqvist & Wiksell, 1971.

Johanson 1995 — L. Johanson. On Turkic converb clauses // M. Haspelmath, E. König (eds.). Converbs in a cross-linguistic perspective: structure and meaning of adverbial verb forms, adverbial particles, gerunds [Empirical approaches to language typology 13]. Berlin: Mouton de Gruyter, 1995. P. 313-347.

Johanson 2000 — L. Johanson. Viewpoint operators in European languages // Ö. Dahl (ed.) Tense and aspect in the languages of Europe [Empirical approaches to language typology 20: EUROTYP 6]. Berlin — New York: Mouton de Gruyter, 2000. P. 27-187.

Kalchofner 1999 — P. Kalchofner. Die Leistung der Indikativformen im Khalkha-Mongolischen // Zentralasiatische Studien 29, 1999. S. 163-199.

Kang 2003 — S. Kang. Hyöndae Monggoö-üi yangt'ae ch'ömsae kwanhan yön'gu [A study of modal particles in Modern Mongolian] // Mongolhak 14, 2003. P. 1-52.

Kim 1995 — Kim Gi-Song. Orchin cagiin mongol helnii üil ügiin engiin cag, tüünii togtolcoo [The simple tenses of contemporary Mongolian and their system]. PhD Thesis. Mongol ulsiin ündesnii ih surguul', Ulaanbaatar, 1995 (published 1998).

Kullmann, Tserenpil 1996 — R. Kullmann, D. Tserenpil. Mongolian Grammar. Hong Kong: Jensco, 1996.

Matsuoka 2008 — Y. Matsuoka. Mongorugo-no asupekuto-ni kan-suru kenkyü [A study of aspect in Mongolian]. PhD Thesis. Kyüshü University, Fukuoka, 2008.

Molochiva 2010 — Z. Molochieva. Tense, aspect, and mood in Chechen. Doctoral Thesis. Universität Leipzig, Leipzig, 2010.

Mönh-Amgalan 1998 — Y. Mönh-Amgalan. Orchin cagiin mongol helnii baimjiin ai [Modality in contemporary Mongolian]. Ulaanbaatar: Moncame, 1998.

Mushin 2001 — I. Mushin. Evidentiality and Epistemological Stance: Narrative Retelling [Pragmatics & Beyond New Series 87]. Philadelphia: Benjamins, 2001.

Narantuyaa 2006 — G. Narantuyaa. Ulaan sarnaig geesen n' [Loss of the red rose]. Ulaanbaatar: S. n., 2006.

Narrog 2012 — H. Narrog. Modality, Subjectivity, and Semantic Change: A Cross-Linguistic Perspective [Oxford Linguistics]. Oxford: Oxford University Press, 2012.

Nedjalkov 1995 — V. Nedjalkov. Some typological parameters of converbs // M. Haspelmath, E. König (eds.). Converbs in a cross-linguistic perspective: structure and meaning of adverbial verb forms, adverbial particles,

gerunds [Empirical approaches to language typology 13]. Berlin: Mouton de Gruyter, 1995. P. 97-136.

Ono 1999 — S. Ono. Nihongo no te iru kei to mongorugo no -j/ch baina kei no hikaku [A comparison of Mongolian form -j/ch baina and the Japanese form te iru] // Soyol erdem deed surguul'. Erdem shinjilgeenii bichig 2, 1999. P. 117-140.

Önörbayan 2004 — C. Önörbayan. Orchin cagiin mongol helnii üg züi [The morphology of contemporary Mongolian]. Ulaanbaatar: Mongol ulsiin bolovsroliin ih surguul', 2004.

San Roque, Loughane 2012 — L. San Roque, R. Loughnane. The New Guinea Highlands evidentiality area // Linguistic Typology 16, 2012. P. 111-167.

Saruul-Erdene 1998 — M. Saruul-Erdene. Mongol helnii üil ügiin cag zaah togtolcoo [The verbal tense system of Mongolian] // Mongol hel shinjlel 2, 1998. P. 92-97.

Sechenbaatar 2003 — B. Sechenbaatar. The Chakhar dialect of Mongol: A Morphological Description [Suomalais-Ugrilaisen Seuran Toimituksia / Mémoires de la Société Finno-Ougrienne 243]. Helsinki: Finno-Ugrian Society, 2003.

Song 1997 — Song Jae-mog. Tense, aspect and modality in Khalkha Mongolian. PhD Thesis. University of London, SOAS, London, 1997.

Song 2002 — Song Jae-mog. Grammaticalization of the verb ge- 'to say' in Khalkha Mongolian // Osaka keizai hooka daigaku sogo kagaku kenkyusho nenpo 3, 2002. P. 29-38.

Svantesson 1991 — J.-O. Svantesson. Tense, mood and aspect in Mongolian // Lund University, Department of linguistics, Working papers 38, 1991. P. 189-204.

Svantesson et al. 2005 — J.-O. Svantesson, A. Tsendina, A. Karlsson, V Franzén. The Phonology of Mongolian [The Phonology of the World's Languages]. New York: Oxford University Press, 2005.

Tatevosov 2002 — S. Tatevosov. The parameter of actionality // Linguistic Typology 6, 2002. P. 317-401.

Tatevosov 2008 — S. Tatevosov. Subevental structure and non-culmination // Empirical Issues in Syntax and Semantics 7, 2008. P. 393-422. (http://www. cssp.cnrs.fr/eiss7/tatevosov-eiss7.pdf)

Umetani 2006 — H. Umetani. "Baina"-iin niilmel üil ögüülehüünii tuslah üil ügeer orson heregleeg busad heregleenüüdtei har'cuulan sudlah n' [Studying the use of "baina" as an auxiliary in complex predicates by comparing it to its other uses]. Paper presented at the Olon ulsiin mongolch erdemtdiin IX ih hural. Ulaanbaatar, August 10, 2006.

Xueyan 2013 — Xueyan. Mengguyu chuanju fanchou [Evidentiality in Mongolian] // A. Yakup, L. Tohti, Zhang Dingjing (eds.). Aertai yuxi yuyan chuanju fanchou yanjiu. Beijing: Zhongyang minzu daxue chubanshe, 2013. P. 180-186.