A mirative construction description: observations about an occurrence in Brazilian Portuguese spoken in Belém

Notions of evidentiality and mirativity are related; however there are aspects of each of them which indicate that mirativity is a distinct conceptual category. One issue is how languages encode these categories. The semantic domain of evidentiality refers to the source of knowledge behind assertions. In Brazilian Portuguese, for example, there is no specific category for evidentiality or mirativity, but there is a syntactic construction in one dialect of Brazilian Portuguese that exhibits mirative overtones. The main goal of this paper is to describe that construction, based on a functional semantic approach.


Introduction
Earlier studies have discussed the related notions of evidentiality and mirativity (CHAFE; NICHOLS, 1986;DENDALE;TASMOWSKI, 2001).The semantic domain of evidentiality covers the information source of knowledge behind assertions.However, as pointed out by Willett (1988), there is no consensus as to the notional boundaries of evidentiality (DENDALE; TASMOWSKI, 2001); recent studies have stressed that mirativity is an independent conceptual category, albeit related to evidentiality (DeLANCEY, 1997).
According to Dendale and Tasmowski (2001, p. 343), mirativity is "a sub-domain situated between evidentiality (direct source of information) and modality (speaker's attitude: surprise)." Following Delancey (1997) and Dickinson (2000) states that mirativity may be a universal semantic category, and all languages probably have the means to code an event or state that is unusual with regard to expectations.Yet according to Dickinson (2000), it is possible to find (i) systems that fuse evidentiality, mirativity, and epistemic modality (she gives an example of the "must-have" constructions in English); (ii) systems that fuse evidentiality and mirativity (e.g., Turkish); and (iii) languages that have evidential systems, independent of epistemic modality, and vice versa.Nevertheless, distinguishing mirative and evidential systems is not always easy.
The goal of this paper is to present data from the Brazilian Portuguese dialect spoken in in the city of Belém, state of Pará, northern Brazil, which can be interpreted as having mirative overtones.
Languages like Brazilian Portuguese or English (CHAFE, 1986) do not encode evidentiality as an obligatory grammatical category, but such languages can develop devices to indicate mirativity, providing evidence to support the claim that the latter is indeed a distinct universal semantic category, as DeLancey (1997) suggests.

Overview of the Data: Marking New, Unexpected Knowledge
Data from the Brazilian Portuguese dialect spoken in the city of Belém, state of Pará, presents a curious grammatical construction with mirative meaning, without any indication that the given information comes from either first or second hand or from hearsay; that is, this construction is not related to evidentiality.In this grammatical construction, free pronouns in subject function (S or A) appear both at the beginning and at the end of the main declarative clauses.Besides this unusual doubling, there is also a rising intonational pattern in this construction, similar to that of some exclamative sentences.The following example comes from a free conversation among teenagers.They were talking about a very shy friend of theirs, who went out to a night club for the first time in his life and danced the whole time, to the astonishment of the speaker, who was narrating what had happened to friends who had not been there: (1) ele dançou muito na festa ele he dance-IPP3s much in+DSFA party he 'he danced a lot at the party' lit.'he danced very much in the party he' The three following sentences below could allanswer the question: "Where is my father?"In (4), the speaker expresses his own surprise about the fact.
(2) ele saiu He go.out.IPP3s 'he went out' (3) ele?Ele saiu HeTop he go.outIPP3s'he?, he went out' (4) ele saiu ele He go.outIPP3s he 'he went out' lit.'he went out he' The context in which I obtained example (4) was the following: An individual wanted to talk to his father by phone, and the father knew he had to wait for his son's call at a certain time; however the father could not wait for it.The person who answered the call uttered the example (4) to inform the father had just left.This information was surprising to the speaker, since the father had been waiting for the call some few minutes ago.The speaker who used (4) was surprised.
Another example involves the first-person singular: (5) eu fiz o trabalho de I do.IPP3s DSMA work of química eu chemistry I 'I did the chemistry homework' lit.'I did the chemistry homework I' The statement above was made in reference to a difficult high-school chemistry exercise.All the students had tried their best to do the exercise and only one of them got it.Example (5) suggests a mirative interpretation.However, whereas DeLancey (1997, p. 38) states that "an inferential marker, lõ [the Hare mirative marker] with first person actors requires a context involving inattention or lack of consciousness," the Brazilian Portuguese mirative-like construction intends to stress the new and in fact unexpected information that the speaker did the exercise before everybody else.
The difference between the Hare and Brazilian Portuguese constructions lies in their relationship to the person.In Hare, the mirative marker used with the first person implies "lack of control," whereas in Brazilian Portuguese it implies new and unexpected information.
The common aspect of the examples given above is the similar contexts in which they occur: (a) new knowledge: all of the examples provide a piece of information that is new to the addressee.
(b) surprise: all of the examples convey the speaker's surprise/admiration concerning this piece of information.
The occurrence of the first-singular and thirdsingular persons is well attested in the data.Upon direct elicitation, some speakers accept the reduplicated use of other pronouns, such as nós 'we,' vocês 'you.PI,' and also the expression a gente1 ; but all of them (including myself as a Brazilian Portuguese speaker) report that although it is possible to duplicate both second-person singular and second-person plural pronouns, as well as the first-person plural pronoun, speakers normally would not do so.
One instance that seems to be an exception to this reluctance is the doubled occurrence of the pronominal expression 'a gente,''we'; this is undoubtedly due to its third-person singular form, in spite of its first-person plural meaning.
Some assumptions can be made with respect to the semantics of the pronouns that appear duplicated in the data: forcefully he 'he blew forcefully' lit.'he blew forcefully he' (e.g., the boy blew out the candles forcefully) However, it seems that this constraint is related to the ambitransitive meaning of such verbs, which demands an agentive subject, since it is possible to have the duplicated pronoun construction with copula verbs.In this case the pronoun can refer to a [-animate] entity.I obtained the following example from a speaker who related an instance "when my sister had lost one of her books in my house.I was looking for this book and I asked her what that book was like, and she replied": sense, given that the speaker is talking about himself with overtones of surprise/admiration; according to DeLancey (1997), evidentiality and mirativity are closely related notions.In contrast to the duplicatedpronoun constructions, example (16) combines both mirative and evidential aspects.In this case the speaker removes himself to a removed position and does not say directly (with a boasting impication) 'I went to do that';instead, he reports his action as though it was that of someone else.

Conclusion
As DeLancey (1997, p. 47) points out, languages such as Albanian, Georgian, Washo, and Akha have mirative-like constructions.The duplicated personal-pronoun constructions in the Brazilian Portuguese dialect of northern Brazil is simply another one of such constructions.
Natural discourse research offers a valuable approach to languages such as Brazilian Portuguese in which mirativity exists as a covert category.Roughly speaking, evidentiality in Brazilian Portuguese is expressed lexically by (i) verbal expressions employing verbs of hearing or seeing; (ii) adverbial expressions; and (iii) mirativity, which can also be coded in this language in different ways: (a) by the lexicon; (b) by, for instance, exclamativeintonational patterns; and (c) by mirative-like constructions that combine unusual grammatical devices and intonational patterns.