Printer Friendly

Clauses emerging as epistemic adverbs in Estonian conversation.

Grammar comes into being in the everyday use of language. This paper demonstrates how four combinations of 1st person pronoun and present tense epistemic verb emerge as adverbs in contemporary spoken Estonian. The analysis is grounded in the understanding that grammar is constantly "done" in our everyday interaction, it is preserved, negotiated, and changed in encounters between the speakers. In other words--it emerges. Some parts of grammar are more stable, others more fluid, and the linguistic structure is constantly created and recreated via usage. Conversation is a crucial locus of grammar and its emergence.

The suggestion to use the term emergent grammar was first made by Paul Hopper who argued that grammar has to be seen as a real-time social phenomenon, the structure of which is always provisional, always negotiable, and in fact epiphenomenal, as much an effect as a cause (Hopper 1987 : 142). Grammar is very much like culture, which is obviously temporal, emergent, always disputed, and forever changing, and this is also valid for grammar. According to this line of thought, structure (grammar, meaning) can be accepted to be much less rigid than what is often assumed, and to inherently involve fuzzy categories (Bybee 1985). Grammar is more a result of spreading of systematicity via repetition than an overarching set of abstract rules (Hopper 1987 : 143-144). Structure is a product of, rather than a prerequisite for, communication, it is an adaptive response to recurrent habits in the way people talk to each other. Obviously, sociocultural linguistic practices are passed down to us from within the speech community. The linguistic structures are already there when we, as actors, enter the continuity of practice. Nevertheless, they are interactionally generated, renegotiated and redefined (Linell 1998 : 125). The generation and renegotiation is tackled by the notion of emergent grammar.

One major outcome of this kind of understanding of language structure is that diachrony and synchrony fade into each other--synchrony is at the same time diachrony since even the synchronic structure crucially depends on repetition and the experience of individual speakers. Also the changes in languages and grammars happen because over time utterances are repeated. Routine repetition may eventually bring about changes in the structure. Repetition may lead to formal reduction, but it may also drain meaning away. This is what can be observed to be happening with some epistemic verb forms in Estonian. Some of them display phonetic assimilation, they may be used with a specialized meaning and they have emerged in a new syntactic role. Also, they lose in pragmatic significance as they tend to have a subsidiary role in a syntactic unit, adding a stance or functioning as an emphasizer. More precisely, they seem to have undergone a process of subjectification, i.e. they acquire meaning characteristics that pertain less to the world being talked about and more to the speaker s organization of that world in the act of speaking (Traugott 1980 : 47; 1989 : 540; 1995; 1997). In this process, as in any kind of linguistic change, frequency is crucial, as it may affect and, ultimately, bring about new form in language (Bybee, Hopper 2001). Frequent phenomena tend to become routinized and potentially lead to change. Speakers can, for example, create new structure by frequent use of certain word combinations in new functions. High-frequency items are also subject to emancipation, i.e. disassociation from the original motivation). Analyzing emergent phenomena thus necessitates a complementary qualitative and quantitative approach.

Processes whereby a form, a combination, or a group of forms is singled out from what was once a more extensive paradigm, fixing or routinizing a meaning or discourse, has often been considered the domain of grammaticalization theory. When it comes to the combinations of pronoun + verb, often classified as parentheticals or comment clauses, there has been some discussion as to whether their emergence as specific linguistic items should be characterized as grammaticalization or pragmaticalization (Waltereit 2002). Formally and semantically, however, the development of many pragmatic markers is consonant with the notion of grammaticalization, and some studies (most notably Brinton 1996 : 272; 2001) argue that it is plausible to account even for pragmatic categories within a theory of grammaticalization. The difference is that in one case the resulting unit is a grammatical item and in the other case, a pragmatic item. Deverbal items discussed by Brinton, such as I mean and y'know, have discourse structuring functions. In general, it is safe to claim that routinization accounts widely for how language evolves and emerges. Studies on grammaticalization have shown how semantic change involves a gradual loss of propositional meaning in favor of a grammatical or pragmatic meaning, underlying the formation of adverbs such indeed and in fact (Traugott 1995). The current paper addresses four Estonian word combinations ma tean 'I know', ma usun, 'I believe' ma arvan ' I think ' and ma ei tea ' I don ' t know ' that seem to be undergoing the same kind of process, acquiring a novel syntactic function. A crucial facet in the variable syntactic-pragmatic usage of this type of items is their prosody, which may distinguish the different functions (Dehe, Wichmann 2010). The study therefore qualitatively focuses on syntactic and prosodic features of these items.

The data come from two spoken language corpora. The first one consists of naturally occurring telemarketing calls as well as everyday calls between family members, relatives, friends, and colleagues. There are about 103,000 words in the corpus and it was recorded in 1997/1998. The other corpus is publicly available,, and constantly growing, but the version checked for the patterns discussed in this study consisted of about 230,000 words. The data comes from a variety of settings, including face-to-face conversations. Unfortunately, not all the sound files for the corpus are available.

Clausal usage

The epistemic verbs arva- 'think (have the opinion)', usku- 'believe', and tead- 'know' are prototypical transitive verbs that take two arguments, one of which can be a complement clause. The pattern is illustrated in (1-4). The subject + predicate + complementizer are boldfaced. In the negative clause the pronoun and the negation word are regularly assimilated in Estonian (e.g. ma + ei = mai).
(1) E: tere, %ee (*) ma tean     et   ma elistan    valesse   kohta,
       hi            I  know:1SG that I  call:1SG   wrong:ILL place:ILL
       aga   akki    te   teate    et    oiget,     ma tean      et
       but   maybe   you:PL know:2PL that right:PRT I    know:1SG that
       te     ka     laenutate valja kostuume.
       you:PL too    lend:2PL costumes:PRT
       Hi. I know, I m calling the wrong place but maybe you know the
       right one. I know that you lend costumes too.

(2) K: m:::a usun        et   ma tellin        jah,
       I     believe:1SG that I  subscribe:1SG yeah
       'I believe, I'll subscribe (to the newspaper)'

(3) K: no ma    arvan      et   kuu       parast
       NO I     think:1SG  that month:GEN after
       <@ voite    ikka elistada. @>
       may:2PL     IKKA call:INF
       'Well, I think you can call in a month.'

(4) P: ei, m hh mai   tea  kas  sa  mind tana  valja kannatad,
       no       I:NEG know QUES you me   today put.up.with
       ma olen      tana  niukseid  lolluseid teinud et
       I    have:1SG today such:PL:PRT follies:PRT do:PPT that
       aa m .hh mai   tea    mis   mul   viga      on @

I:NEG know what I:ADS problem is 'No, I don't know whether you can put up with me today, I have done so stupid things today, I don't know what's wrong with me.'

In these cases, the verbs are used within full clauses and behave grammatically as predicates. However, the same combinations of ma 'I' + verbs may also be used without the complementizers, which shows that the syntactic patterns are leaking. Examples are provided in (5)-(8), where the focused structures are boldfaced.
(5) P: /--/ .hh a   kuskile      mujale   ma teda   jatta     ei
                but anywhere:ALL else:ALL I  it:PRT leave:INF NEG
       (0.2) ma tean     ma oin     taga   seal  klassis
             I  know:1SG I  was:1SG it:COM there class:INS
        'But I couldn't leave it [the purse] anywhere else. (0.2)
          I know that
        I had it in the classroom.'

(6) A: /--/ .h e mm siis e mm (suunad)    me panime ara,
                    then       directions we put:IMF:1PL ARA
       ja   sellele katte      peale, niet ma usun
       and this:ALL cover:GEN on      so   I  believe:1SG
       see peaks        toimima       koik.
       it   must:COND function:SUP all
       'Then we adjusted the directions and put on a cover so I believe
       that it should all function.'

(7) P: no ma arvan     ma uhe     suure   korvitsa    hiivan   kaa  ara,
       NO I  think:1SG I  one:GEN big:GEN pumpkin:GEN take:1SG too ARA
       aga mai   viitsind tassida.
       but I:NEG bother:PPT carry:INF
       'I think I'll take a big pumpkin too but I couldn't be bothered
       carry (it).'

(8) H: mhmh (1.5) aga: (*) mai   tea
       uhuh       but      I:NEG know
       bussi-bussi:reis laheks  vist     kallimaks. (*)     voi kuidas
       bus-  bus.trip   be:COND probably expensive:COMP:TRA or  how
       'Uhuh. (1.5) but (.) I dunno, a bus trip would probably be more
       expensive (*) or?'

It is questionable whether the clauses following the focused items in these examples function as complement clauses. They are not grammatically tied to ma arvan, ma usun, ma tean, and mai tea. For the same reason, it is not clear that the word combinations itself can be analyzed as main clauses that take a complement. In fact, occasional lack of complementizer was the criterion of choosing patterns for this study. Research into conversation in other languages, such as English and Swedish, have shown that lack of complementizer may be a symptom of the fact that personal pronoun + verb patterns do not function as main clauses (Karlsson 2005; Thompson 2002), or that their syntactic status is indeterminate between main clause and parenthetical (Longman Grammar of Spoken and Written English 1999 : 197, cf. 1076-1077; Kaltenbock 2005 : 43-45; Dehe, Wichmann 2010). In Estonian, word combinations such as ma utlen 'I'm saying' and ma raagin 'I'm talking' have been argued to function as a modal frame and not reporting clauses (Hennoste 2004). In the sections below we will look closer into the emergence of the above four word combinations as different kinds of entities in interaction. They are not all used in exactly the same way and there are considerable differences in their frequency as well as positions in the turn.

The study looks systematically at all the epistemic verbs that displayed syntactic leaking in the corpora. Epistemic words express the speaker's commitment to what she is saying, particularly the certainty of her knowledge state. They reflect the speaker's judgment of, or degree of confidence in the knowledge upon which a proposition is based. Depending on the speaker's claimed certainty of knowledge, the items studied here can be arranged along a continuum from more to less certain.
             ma tean       ma usun       ma arvan      ma ei tea
             'I know'      'I believe'   'I think'     'I don't know'
certainty of [right arrow] [right arrow] [right arrow] [right arrow]
  lack of


The items will be dealt with in this very order, focusing on their syntactic position and prosody of production.

Expressing certainty, ma tean 'I know'

There were altogether 143 cases in the corpus where the words m(in)a1 'I' and tean 'know:1SG' occurred in the same clause. Only in five cases the combination ma tean emerges as a potentially non-clausal unit. In example (5) above we saw that it can be used without a complementizer even though it is attached to another clause that could in principle function as a complement. In that case, ma tean was placed before the clause that it commented on. With it the speaker underlined she is sure that she had the purse in the classroom. In addition to this pre-positioned usage, ma tean can be used internally in an ongoing clause. In example (9) it is used after the predicate saatis 'sent' and before the object messidzi 'message'. The item ma tean cannot be translated idiomatically in this position in English and therefore it is rendered in Estonian even in the third line.
(9) K: aga sulle      saatis   ma tean     messidzi
       but you:SG:ALL sent:3SG I  know:1SG message:GEN
       keegi      Beatrix.
       somebody   NAME
       'But you got ma tean a message from somebody called Beatrix.'

Ma tean can also be added after the clause it comments has come to an end, to underline the certainty of the claim. In example (10) it occurs in a strong agreement and co-occurs with another epistemic device nagunii 'of course, anyway'.
(10) H: .hh aa, no siis ei  saa ega ma ikkagi ei  saa nigunii
            oh  NO then NEG can EGA I  still  NEG can anyway
         ilma   sinuta  seda    tapeeti      osta
         without you:ABE this:PRT walpaper:PRT buy:INF
         ja  parast     sa  ei  ole nigunii rahul ju.
         and afterwards you NEG be anyway satisfied JU
         'Oh, then (I) can't- I cannot buy the wallpaper without you
         anyway and afterwards you won't be satisfied anyway, right?'
     V:  ei  ole nagunii, ma tean.
         NEG be  anyway   I  know:1SG
        'Of course I won't, I know.'

Even though the clause-internal and turn-final positions of ma tean are relatively rare, it is clear that the syntactic patterns have leaked to the extent that the verb tean does not necessarily take two arguments any more. This is also the reason why calling items such as this comment clauses is problematic--they are not really clauses but the word combination seems to have a different syntactic role.

There is no complement clause in examples (9) and (10), because there is no complementizer, a complement cannot occur before the main clause (without a correlate), and a complement clause cannot be split to two and surround the main clause. Instead, ma tean is used as an item that is to some extent mobile within the clause, allowing for variable positioning. As ma tean expresses speaker certainty, it is functionally replaceable with an epistemic modal adverb. This is a crucial factor to consider in languages with free word order, such as Estonian. A lot of argumentation on parentheticals vs. pragmatic markers vs. adverbs is based on the English language with strict word order (for an excellent overview, see Brinton 2008 : 1-18). Thus, one crucial feature of pragmatic markers and parentheticals has been claimed to be that they are movable while sentence adverbials are not. In Estonian, sentence adverbials are also movable. Furthermore, both adverbials and items such as ma tean can occur in identical positions clauseinternally, which suggests that the latter is indeed functionally identical with an adverb. They both have a sentential/clausal scope and indicate speaker attitude. It seems that in Estonian, at least, there are more reasons than in English to consider ma tean an adverb, in case it is used as shown in example (9). Furthermore, at least in the case of English, Dehe (2009) argues that the prosodic integration of comment clauses cannot be explained in syntactic accounts which do not allow for a syntactic relation between parenthetical and host. Parentheticals are part of the emerging syntactic structure and should be accounted for as regular parts of it.

The sound file of example (10) is unfortunately not available but in example (9) ma tean is produced as a regular part of the utterance in the sense that it is not marked by increased speed and lowered pitch and volume, as has often been claimed to be characteristic of English parenthetical comment clauses (Quirk, Greenbaum, Leech, Svartvik 1985 : 1112, 1113). There is no corresponding research on Estonian parentheticals and most probably there would be crucial differences between the two languages, especially in regard to pitch movement. However, general functional claims based on English data, such as that prosodic prominence and separation go together with semantic transparency, at least in internal syntactic positions (Dehe, Wichmann 2010), should also be checked on other languages. According to this view, prosodic integration and deaccentuation accompany semantic bleaching, resulting in the more discourse-based and interpersonal meaning of the items. If that is the case, the Estonian ma tean has not started to bleach semantically, as bothwordscarry acartain amount of stress, theitem is not produced lower or quicker than other parts of the utterance, as shown infigure (1).

The fact that the item is not deaccentuated or completely integrated supports the argument that ma tean plays a regular role in the syntactic structure. It functions as a sentence adverbial that epistemically qualifies theclaim that is being conveyed, expressing certainty.


Expressing less-than-certainty, mausun 'I believe'

The combination of m(in)a 'I' and usum 'believe' is even less frequent in the corpora, occurring 54 times. Only four cases are used without the complementizer and the item ix always pre-positioned in regard to the clause that it expresses less-than-certainty about. In example (11) a pastor at a church answers a question by one of his congregation members concerning an announcement of a Christmas show. In his answer the pastor expresses his belief that the people will show up.
(11) K: utlesin,    ja  kolmapaeval    ka, niet ee
        say:IMF:lSG and Wednesdciy:ADS too so.that
        'I did and (I said it) on Wednesday too, so'
     E: mhmh=
     K: =et ee ma usun hh .h rahvas tuleb,    ja: ja pe- see
        that   I  believe:lSG people come:3SG and and    this
        puhapaev saab     veel       oeldud   ka kindlasti.
        Sunday    FUT:3SGoncS.mPre say:lMS:PPT too Oellmtely
        'I believe that people will come and onSunday (I')ll
        definitely say
        it too.'

The prosody of the focused utterance is shown in figure (2), where it can be seen that ma usun is not produced lower or quicker than what follows. It retains the prosodic nature of a separate clause, apart from lack of a prominence (cf. the steep fall in the following clause). The intonation is level and the terminal pitch projects more to come but it is not compressed.


The above two items ma tean and ma usun both express anepistemic stance toward what is being said. With ma usun the speaker quite literally expresses a mere belief and is thus lees than certain about the claim. The items can display epistemic information by constituting a clause with a complement but they do not always display the syntactic properties of a main clause. They may preface, or be inserted into, an ongoing clause. An example of an inserted ma usun is presented in (12). The sound file is not available.
(12) H: .hh et   uhesonaga niisugust     rahulikku    puhkust, (1.0)
            that so        this.kind:PRT peaceful:PRT vecation:PRT
        ma usun xcarsun nngu tit   looUs,     (1.0) ta ei: ee taju
        I believe:1SG an she hoped:3SG the NEG experience there
        'So there she won't experience the kind of peacful vacation
        ma usun
        that she hoped for.'

However, the patterns described above for ma tean and ma usun are peripheral in their overall usage. We will now turn to he two other items where similar patterns are more frequent and where the prosody is accordingly different.

Expressing some uncertainty, ma arvan 'I think'

The word combination ma arvan 'I think (have the opinion)' is generally quite frequent in conversational Estonian (227 cases in the corpus). In addition to literal usage, there are 25 cases where ma arvan is used in varying syntactic positions that cannot be predicted from its subject + predicate structure, or where the syntactic patterns seem to be leaking. In example (7) above the complementizer et 'that' seems to be simply left out. The speaker expresses some uncertainty about whether she will indeed take the pumpkin. The word combination ma arvan seems to project an object or object complement clause but the upcoming spate of talk does not display any features of a complement clause (i.e. a complementizer). Ma arvan thus functions as an epistemic certainty marker in relation to the upcoming clause. In the English language, Thompson and Mulac (1991a : 244-245) argue that the statistically significant lack of (1991a : 244-245) argue that the statistically significant lack of that with I think and I guess in conversation is related to their epistemic meaning. They call these items epistemic parentheticals. Also Estonian comprehensive grammar treats items such as ma tean and ma arvan as parenthetical items, without which the sentence would still be well-formed (Erelt, Kasik, Metslang, Rajandi, Ross, Saari, Tael, Vare 1993 : 103). Thompson (2002) has later suggested that they could be analyzed as epistemic phrases combined with a clause, instead of seeing the structure as a combination of a main clause and a subordinate complement clause. Interestingly, there seems to be a cross-linguistic tendency that the syntactic patterns of epistemic verbs are leaking and that they emerge as parts of epistemic phrases instead (for the Swedish tycker jag / jag tycker 'I think', see Karlsson 2005). Some researchers take a step further and consider items such as I think, I assume, I don't think so, or I doubt as sentence adverbials (Jackendoff 1972 : 95-100). In Estonian, where at least the focused items share syntactic positions and other characteristics with adverbs, there is a good reason for doing so.

In addition to the pre-positioned usage of ma arvan, the corpora include 18 cases of ma arvan in the middle of syntactic and intonation units. Ma arvan can be placed in a variety of positions within a clause. In example (13), it is used after a time adverbial neljaks 'at four (o'clock)' and before the subject ma 'I'. Here, the speaker is informing her interlocutor about the time when she will start going home from her job. In example (14), ma arvan is used after the subject and the predicate ma saan 'I'll get' but before the object nelja 'four'. The speaker is making a guess of the grade she will be granted for her exam. In these cases, the item ma arvan is incorporated into the syntactic units in a new way and we have thus the best reason for analyzing it as an epistemic adverb.
(13) S: ja: - vv e e no v - no igatahes neljaks  ma arvan
        and          NO     NO anyway four:TRA   I  think:1SG
        ma olen   juba    siit      ammu     lainud.
        I  be:1SG already long.ago go:PPT
        'and any way, at four ma arvan I have been gone for a long time
        from here already.'

(14) L: ma saan    ma arvan     nelja    vist     va. voi kolme. /--/
        I  get:1SG I  think:1SG four:GEN probably VA  or  three:GEN
        I('II) get ma arvan a four maybe or a three [the grades]'

Example (14) is produced very low but (13) provides a nice visible contour. In figure (3), it can be seen that ma arvan is produced quickly and without prominences but so is the talk that follows it. Ma arvan is thus prosodically integrated to the ongoing unit, as is (14) where ma arvan is very short and even phonologically reduced. These prosodic features are quite different from what we say with ma tean, which we also saw in the clause-internal position. Still, the item retains its straightforward epistemic semantics. By using it, the speaker indicates that she is uncertain about the claim.


In comparison, we can look at the prosody of the corresponding adverb arvatavasti 'most probably' within a clause 'Jaan arrives most probably on Thursday' in figure (4). Similarly to example (13), it is produced prosodically at level with the rest of the clause, with the same pace and height, but not carrying the main prominence in the clause. Thus, even prosodically ma arvan, functions as an adverb.


In addition to the pre-positioned and inserted us age, there are l7 cases in the corpora where ma arvan is used after the clause it is commenting. The turn-final position is the appropriate places for stance-taking in regard to what was said but definitely not a prototypic position for a pronoun and a transitive verb. This supports the argument that ma arvan functions as an adverb rather than as a clause. An example is presented in (15) The telemarketer is asking a child when his parent will be at home. The child answers ohtupoole 'later in the evening' and adds ma arvan as a hedge. THere is a complex transition relevance place (as described in Ford, Thompson 1996 : 153-157) after ohtupoole, involving syntactic, prosodic and pragmatic completion. Ma arvan is produced incrementally as a post-completion .
(15) M:=ahaa. ja: millal tulevad. (*) ei  tea  kah.
        AHAA  and when   come:3PL     NEG know too
        'Oh and when are they coming? (*) (you) don't know?'
     K: n ohtupoole.           ma arvan.
          later in the evening I  think:1SG
        'Later in the evening, I think.'

The item ma arvan is recurrently used as an epistemic adverb in spoken Estonian and is thus comparable to the frequent usage of the English I think and I guess as epistemic markers (Aijmer 1996; Karkkainen 2003; Thompson, Mulac 1991b). Ma arvan is completely mobile, occurring in a variety of positions in relation to the clause it comments on. In contrast, ma usun and ma tean primarily preceded the clauses they comment on, displaying the pattern that would be expected in case the upcoming clause was indeed a complement. Ma arvan was also relatively more frequent in functions that are not analyzable as subject + predicate. Entire 20 % of its occurrence was more adverb-like, while in case of ma tean and ma usun the adverb-like occurrence constituted merely 3.5 % and 7.4 % respectively. Thus, as far as we can rely on the frequency effects in the current corpus, ma arvan seems to stand out more clearly as an adverb. It expresses epistemic stance towards the preceding, following, or ongoing claim, functioning syntactically as an adverbial. Ma tean and ma usun display a similar syntactic potential but are generally less frequent and also less frequent in the stance-taking function.

When it comes to the semantic development of first person epistemic parentheticals, Traugott (1995) notes that the subject may lose referential (objective) properties, and become simply the starting point of a perspective. Thus, a following grammaticalization cline is proposed: act of cognition > mode of knowing (evidential) > certainty epistemic (Brinton 1996 : 243). This path of ongoing development could also be suggested for at least ma arvan in Estonian. However, ma arvan is not only used to mark a degree of epistemic uncertainty or guess but also to indicate that what is expressed is a personal opinion. Even though all the items in this study are originally subjective claims, ma arvan can furthermore express the social interpersonal stance that the speaker takes in regard to the claim it is attached to. In these cases it is translatable as 'in my opinion'. Example (16) shows an instance where the speaker is giving advice and her ma arvan does not mark the advice as a guess about a state of affairs (comp. examples (13-15)) but as a personal standpoint, which is uncertain in the sense that it is not an objective truth. Thus, clear subjectification and some degree of semantic bleaching can be observed in this example. The item functions as a mitigator for the directness of advice, orienting to the interpersonal need of not imposing on others.
16) V: aga siis  peate    ikka  varem   olema  Tallinas ma arvan.
       but then  must:2PL still earlier be:SUP NAME:INS I think:1SG
       'But then you have to be in Tallinn earlier, I think.'

The sound file for example (16) is unavailable but in the case of (15) ma arvan is prosodically separated from, and only marginally less stressed than, preceding talk. The amount of prosodic prominence on ma arvan varies depending on its position in the utterance. When preceding or following the commented clause, it can be produced with a separate contour and some prominence but in the middle of an ongoing syntactic unit it is integrated, as are other adverbs. Prosodic integration, however, does not necessarily imply semantic bleaching. Most of the time ma arvan still expresses epistemic stance towards what is being conveyed, and only in rare cases can it be analyzed as a mitigator functioning within the interpersonal domain. In summary, there are syntactic, semantic and prosodic reasons for sometimes considering ma arvan as a sentence adverbial in Estonian conversation. Similarly to adverbs, it can be used at the beginning of a clause, in the middle of the clause, or be added to the turn after the turn construction unit has come to an end.

Expressing uncertainty, mai tea 'I don't know'

In many languages, the phrase meaning approximately 'I don't know' displays routinized usage patterns in which it does not occur in its literal meaning and does not function as an independent clause. For example, I dunno in English can be used as a hedge, carrying a more subjective meaning and fulfilling face-saving functions (Scheibman 2000; Tsui 1991). In Estonian, the corresponding knowledge denial is m(in)a ei tea 'I NEG know', which is an extremely frequent collocation in the spoken usage--in 791 cases the words occurred in the same clause in the corpora. The negation word is regularly assimilated with the pronoun in Estonian, resulting in mai but in a lot of cases all the three words sound more like one maitea, with the single word stress on either syllable, but most often on tea. However, the item mai tea is used in a number of different ways, such as for list termination, approximation, and achieving a disjunction (Keevallik 2003 : 78-100). Most of them are irrelevant for the current argumentation. Here the focus is on instances of mai tea that do not constitute a turn on their own. They modify surrounding talk but are not formally involved in a larger syntactic structure by being followed by a complementizer or any other conjunction or question particle.

Most frequently, mai tea precedes the talk it modifies (91 cases). And most frequently, it precedes a whole clause (72 cases) but it can also preface a phrase. The first option was shown in example (8) above and the pre-phrasal position is demonstrated in (17). None of the above items occurred in pre-phrasal position in the data, which suggests that mai tea may be the most adverb-like item among them.
(17) H: vot (*) kuule          mis  sa  arvad     et   tana  ei
        viitsi va.
        VOT     listen:IMP:2SG what you think:2SG that today NEG
        bother or
        'Okay. Listen, so you think that you won't bother today?'
        [to come
        to a casino]

     V: kuule:         jah, (0.8) maitea     omme     voib-olla.
        listen:IMP:2SG yeah       I:NEG.know tomorrow maybe
        'Well, yeah. (0.8) mai tea tomorrow maybe.'

The prosody of the last utterance is shown in figure (5), where it is clear that the maitea is quite compressed compared to the following word which carries the main prosodic prominence. It is produced lower and quicker and thereby constitutes the least prominent production of all the items in the current study (among the ones that could be measured).

The scope of mai tea in the above two cases is either the upcoming clause (in example 8) or phrase, which constitutes an answer to the question (in example 17). The items are prosodically detached from prior talk and latched to the upcoming talk. As the speaker is in both cases indeed providing information, mai tea cannot literally state lack of knowledge. Rather, it expresses uncertainty, i.e. epistemic modality (a similar analysis for I don't know is presented in Tsui 1991 : 619-620). The usage displays subjectification and semantic bleaching, similarly to what was shown for ma arvan in example (16). in example (16). There are also other markers of uncertainty present in the above examples, vist and voib-olla, both 'maybe'. In example (18), mai tea is the only marker of uncertainty, hedging a proposal for health care.

(18) P: =raudrohuteed    aga motle         noh
         milfoil.tea:PRT but think:IMP:2SG NOH
         kuidas  selle    aalega    voi  kurguga kaituda?
         how     this:GEN voiceCOM or throat:COM deal:INF
         'Milfoil tea. But tell me how to deal with the voice
         or the throat?'

     E: mai   tea  lase        tal    rahulikult lihtsalt olla.
        I:NEG know let:IMP:2PL it:ADS peacefully just     be:INF
        'I dunno, just let it be in peace.'

Even in this case, mai tea is intonationally integrated to the upcoming talk but it is more prominent, starting higher and louder than the item in example (17), and carrying a prosodic prominence of its own. There is thus no straightforward relationship between prosodic prominence and syntactic functions in pre-positions, as in both examples the item expresses uncertainty.

The item can also be inserted in the middle of the ongoing clause and there are altogether 29 instances like that in the current corpora. Similarly to the prepositioned usage, inserted mai tea indicates uncertainty about the claim being made. Example 19) comes from a call to an information line and the speaker indicates trouble with formulating the description or name of the place he is looking for. Mai tea is used after copula and before the subject.
(19) H: tere mind uvitaks e      Kivilinna: (*) kaupluses
     hi      I:PRT interest:COND NAME:GEN       shop:INS
     on      maitea   mingi (0.5) juuksur     voi (*)
     be:3SGl:NEG.know some        hairdresser or
     midagi        Kaa  salong voi mingi asi   on.
     something:PRT NAME salon  or  some  thing be:3SG
     'I am interested in, at Kivilinna shop there is maitea some
     kind of
     a hairdresser or something, salon K or something.'


This clause-internal case is prosodically characterized by integration but not compression. Mai tea carries about the same amount of prominence as other prominent syllables in the clause and contributes to the evan speech rhythm. Similarly to clause-internal ma tean and ma arvan above, mai tea constitutes a fully-fledged member of the prosodic unit with no separating pauses or drastic pitch resets.

Finally, the uncertain marker can be added after the clause hat been terminated. The speaker has stated something and retrospectively marks his or her epistemic stance toward it (39 cases in the data). This can be done in two ways, either by just adding mai tea (15 cases) or more often, by adding mai tea together with a conjunction. The two options are demonstrated in examples (20) and (21). The first one is prosodically completely integrated, continuing the already low preceding word. It is also produced with very little energy, as shown in figure (7).
(20) L: okei ma- ma voibolla raagin    taga    ka  praagu maitea.
        okay I   I  maybe     talk:1SG she:COM too now    I:NEG.know
        'okay, maybe I'll talk to her too now, I don't know.'


Low prosodic prominence is characteristic of adverbs in clause-final positions. An example of the clause ma elistaksin siis omme tagasi voibolta 'I'll call back tomorrow then maybe' is shown in figure(8) for comparison. The word for 'maybe' is voibolla, which is an epistemic adverb. In both cases the last stance-taking item is produced as part of the ongoing unit, latching onto it.


In contrast, the conjunction-prefaced variant ve mai tea 'or I dunno' in (21) displays some independence and a separate prosodic contour (demonstrated in figure 9, see especially intensity curve). Still, it is prosodically latched, as ve mai tea is initiated at almost the same height where the previous word ends.
(21) P: /--/ mis  taendab  see on     mingi e      ta- tapselt
            which mean:3SG it  be:3SG some.kind.of    precisely
        siuke tunne   nagu mingit     heina tombaks.
        such feeling as some:PRT hay:PRT smoke:COND
        ve mai   tea noh
        or I:NEG know NOH
        'which means than it is kind of precisely the same feeling as
        when you smoke hay or I don't know.'


The very frequent item mai tea thus displays several features of an epistemic adverb. It may be used in different positions in the turn, including clause-internally and pre-phrasally, which is not how a subject predicate would normally be used. Compared to the three-word combination, the kind of mai tea discussed here is phonologically assimilated. In prepositioned or postpositioned usage, its prosodic integration varies but in the middle of the clause, it displays integration but not compression. In addition, the semantics of the adverb-like mai tea is different from the main clause ma ei tea, which states lack of knowledge. In the adverb-like cases, it merely indicates uncertainty about what has been said, is being said or will be said, and not literally lack of knowledge. It marks the speaker's stance. There are thus phonological, syntactic, semantic, and prosodic reasons for analyzing mai tea as an epistemic adverb, as an item that modifies what is said and that can be freely moved around in the turn and the


The above four combinations of 1st person pronoun + epistemic verb sometimes emerge as adverbs in Estonian conversation. The process whereby something like that can happen has been described as grammaticalization and in many ways the cases display relevant features. We have seen phonological shortening and assimilation, changes in semantics and syntactic positioning and function, as well as subjectification, all of which occur to varying extent with ma tean, ma usun, ma arvan and mai tea. On the other hand, the patterns of usage of these items need not represent language change, but might as well be an illustration of the fact that linguistic categories, word classes and parts of speech included, emerge in usage and cannot be assumed a priori for a linguistic item. Categories have fuzzy boundaries and every instance of usage has to be analyzed taking into consideration its particulars in the current context. Whether we are observing a change or not, frequency of occurrence is crucial for speakers' understanding of function. The more often a word combination is used similarly to adverbs the more reason there is to reconsider its placement among parts of speech in the language structure. Table (1) summarizes the quantitative data on the occurrence of the above items. It shows that particularly ma arvan and mai tea should deserve serious consideration as adverbs because about one fifth of their overall occurrence is adverb-like.

The clause-internal and post-positioned usage arguably enable most clear-cut distinction from the subject + predicate usage of the same words. Even in regard to this, ma arvan and mai tea stick out as markedly more frequent compared to the other items. Prosodically, ma tean, ma arvan, and mai tea behaved similarly to adverbs in clause-internal and post-positioned usage, while in the latter position there was also an option of more prominence and independent intonation contour. This may indeed be one of the interactional motivations for their usage, as it enables the speaker to achieve an extra emphasis on the epistemic item.

Traditionally, items like this have been called parentheticals or parenthetical expressions in linguistics. These are constructions, words, phrases or sentences, which occupy a syntactically peripheral position in a sentence, are typically separated from their host clause by comma intonation, and which function as a gloss or comment on some aspect of the meaning of that sentence (Rouchota 1998 : 97). We have not been able to observe comma intonation in most of the above cases. Rather, the items have been integrated in the larger prosodic units as we would expect to be the case with any syntactically incorporated item. Similar results have been arrived at in the instrumental measurements of some parentheticals in English (Wichmann 2001 : 185-186). Reminding of adverbs, the items discussed in this study exhibit lack of propositional content, syntactic moveability, and optionality in the clause. In interactional encounters ma tean and ma usun, but especially ma arvan and mai tea emerge as epistemic adverbs with partly different semantics and prosody as compared to the corresponding main clauses.

doi: 10.3176/lu.2010.2.01

Transcription and translation conventions

underlining--emphasis; %--glottal sound; .--pitch fall at the end of an intonation unit;,--level pitch at the end of an intonation unit; (0.5)--pause length in tenths of a second; (*)--micropause;:--lengthening of a sound;--truncation; =--latching or continuation of the same speaker across intervening lines; []--overlaps; ()--unsure transcription; <parenthesis>--marked quality of some kind; <@ smile @>--smiling quality; @--laughter syllable; .hh--breathing in, the estimated relative length corresponds to the number of h-s; hh--breathing out; /--/--some talk has been left out from the same turn.

1, 2, 3--person; ABE--abessive; ADS--adessive; ALL--allative; COM--comitative; COMP--comparative; COND--conditional; FUT--future; GEN--genitive; ILL--illative; IMF--imperfect; IMS--impersonal; IMP--imperative; INF--infinitive; INS--inessive; NEG--negation (particles ei, ara); PL--plural; PPT--past participle; PRT--partitive; QUES--question word; SG--singular; SUP--supine; TRA--translative.

Other capital letters--an untranslatable particle.


Aijmer, K. 1996, I think--An English Modal Particle.--Modality in Germanic Languages, Berlin, 1-47.

Brinton, L. J. 1996, Pragmatic Markers in English. Grammaticalization and Discourse Functions, New York-Berlin.

--2001, From Matrix Clause to Pragmatic Marker. The History of look-Forms. --Journal of Historical Pragmatics 2, 177-199.

Bybee, J. L. 1985, Morphology. A Study of the Relation between Meaning and Form, Amsterdam-Philadelphia (Typological Studies in Language 9).

Bybee, J., Hopper, P. 2001, Introduction to Frequency and the Emergence of Linguistic Structure.--Frequency and the Emergence of Linguistic Structure, Amsterdam-Philadelphia (Typological Studies in Language 45), 1-24.

Dehe, N. 2009, Clausal Parentheticals, Intonational Phrasing, and Prosodic Theory.--Journal of Linguistics 45, 569-615.

Dehe, N., Wichmann, A. [forthcoming], Sentence-Initial I think (that) and I believe (that). Prosodic Evidence for Use as Main Clause, Comment Clause and Discourse Marker.--Studies in Language.

Erelt, M., Kasik, R., Metslang, H., Rajandi, H., Ross, K., Saari, H., Tael, K., Vare, S. 1993, Eesti keele grammatika II. Suntaks. Lisa: kiri, Tallinn.

Ford, C. E., Thompson, S. A. 1996, Interactional Units in Conversation. Syntactic, Intonational, and Pragmatic Resources for the Management of Turns.--Interaction and Grammar, Cambridge (Studies in Interactional Sociolinguistics 13), 134-184.

Hennoste, T. 2004, Et-komplementlause kesksete pohiverbide funktsioonid eestikeelses vestluses.--KK, 504-523, 590-609.

Hopper, P. 1987, Emergent Grammar.--Berkeley Linguistics Society. Papers of the Thirteenth Annual Meeting, Berkeley, 139-157.

Jackendoff, R. S. 1972, Semantic Interpretation in Generative Grammar, Cambridge, MA-London.

Kaltenbock, G. 2005, Charting the Boundaries of Syntax. A Taxonomy of Spoken Parenthetical Clauses.--Vienna English Working Papers 14, 21-53.

Karlsson, S. 2005, Modalitet i interaktion--en studie av jag tycker och tycker jag.--Samtal och Grammatik, Lund, 119-138.

Keevallik, L. 2003, From Interaction to Grammar. Estonian Finite Verb Forms in Conversation, Uppsala (Acta Universitatis Upsaliensis. Studia Uralica Upsaliensia 34).

Karkkainen, E. 2003, Epistemic Stance in English Conversation. A description of its Interactional Functions, with a Focus on I think, Amsterdam-Philadelphia.

Linell, P. 1998, Approaching Dialogue. Talk, Interaction and Contexts in Dialogical Perspectives, Amsterdam-Philadelphia.

Longman Grammar of Spoken and Written English, Harlow 1999.

Quirk, R., G r e e n b a u m, S., Leech, G., Svartvik, J. 1985, A Comprehensive Grammar of the English Language, London-New York.

Rouchota, V. 1998, Procedural Meaning and Parenthetical Discourse Markers. --Discourse Markers. Descriptions and Theory, Amsterdam-Philadelphia (Pragmatics & Beyond NS 57), 97-126.

Scheibman, J. 2000, I dunno... A Usage-Based Account of the Phonological Reduction of don't in American English Conversation.--Journal of Pragmatics 32, 105-124.

Thompson, S. A. 2002, "Object Complements" and Conversation. Towards a Realistic Account.--Studies in Language 26, 125-164.

Thompson, S. A., Mulac, A. 1991a, The Discourse Conditions for the Use of the Complementizer that in Conversational English.--Journal of Pragmatics 15, 237-251.

--1991b, A Quantitative Perspective on the Grammaticalization of Epistemic Parentheticals in English.--Approaches to Grammaticalization, Vol. 2. Focus on Types of Grammatical Markers, Amsterdam-Philadelphia (Typological Studies in Language 19), 313-339.

Traugott, E. C. 1980, Meaning-Change in the Development of Grammatical Markers.--Language Sciences 2, 44-61.

--1989, On the Rise of Epistemic Meaning in English. An Example of Subjectification in Semantic Change.--Language 65, 31-55.

--1995, Subjectification in Grammaticalization.--Subjectivity and Subjectivization. Linguistic Perspectives, Cambridge, 31-54.

--1997, Subjectification and the Development of Epistemic Meaning. The Case of promise and threaten.--Modality in Germanic Languages. Historical and Comparative Perspectives, Berlin--New York (Trends in Linguistics, Studies and Monographs 99), 185-210.

Tsui, A. B. M. 1991, The Pragmatic Functions of I don't know.--Text 11, 607-622.

Waltereit, R. 2002, Imperatives, Interruptions in Conversation, and the Rise of Discourse Markers. a Study of Italian guarda.--Linguistics 40, 987-1010.

Wichmann, A. 2001, Spoken Parentheticals.--A Wealth of English. Studies in Honour of Goran Kjellmer, Goteborg (Acta Universitatis Gothoburgensis. Gothenburg Studies in English 18), 177-193.

(1) There is a long and a short version of the pronoun, the longer one is less frequent and used to achieve prominence.


Leelo Keevallik

Uppsala University

Table 1
The frequencies of adverbial usage of the four items in different
positions in relation to the commented clauses or phrases

                         adverbial usage

            pre-positioned   internal   post-positioned

ma tean            3             1             1
ma usun            3             1             0
ma arvan          11            18            17
mai tea           91            29            39

             Overall     % of adverbial
            occurrence       usage

ma tean        143            3.5%
ma usun         54            7.4%
ma arvan       227           20.3%
mai tea        791           20.1%
COPYRIGHT 2010 Estonian Academy Publishers
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Keevallik, Leelo
Publication:Linguistica Uralica
Article Type:Report
Geographic Code:4EXES
Date:Jun 1, 2010
Previous Article:90 years of the Mother Tongue Society.
Next Article:Grade alternation in Voro South Estonian.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters