1
Topics and Conditionals in A’ingae
Silvia De Grandis 11311517
University of Amsterdam Research Master Thesis in Linguistics
Supervisor: Prof. dr. Kees Hengeveld Second reader: Prof. dr. Enoch Aboh
Abstract
A relationship between Topics and Conditionals has been known since Haiman (1978). A’ingae, a language isolate spoken by approximately 1400 people in Colombia and Ecuador, is particularly interesting in this respect since there are two clitics that can be used to mark Topics as well as Conditional clauses. The present study focuses on these clitics and its purpose is primarily descriptive, that is, to give a picture of the use, and the difference in use, of the two clitics. Moreover, I investigate the hypothesis that the two clitics may mark different types of Conditionals (Ebert et al., 2014). A corpus approach is taken, with a dataset collected and analysed by Borman (1990). The framework chosen is Functional Discourse Grammar (FDG. Hengeveld and Mackenzie, 2008); since FDG predicts that Conditionals are marked at the Episode layer, that is, the layer that is formed by one or more States-of-Affairs that are thematically coherent, I hypothesise that the Topic clitics are markers of Episodes. The results show that the clitics are markers of Information Structure, and that it is not possible to state that Topic markers are Episode markers. Finally, a significant relation between different types of Conditionals with respect to the two clitics was not found .
2 Table of Contents Acknowledgments ... 3 List of abbreviations ... 4 1. Introduction ... 6 2. A’ingae ... 7
3. Conditionals are Topics 3.1. A Definition of Topic ... 10
3.2 Conditionals and Information Structure ... 10
4. Functional Discourse Grammar (FDG) 4.1. General outline ... 14
4.2. Information Structure ... 15
4.3. Conditionals ... 16
5. Research Questions & Methodology ... 20
6. Results 6.1. New Topic ... 21
6.2. Contrastive Topic ... 25
6.3. Conditionals ... 31
7. Discussion 7.1. New and Contrastive Topic ... 33
7.2. Conditionals ... 35
8. Conclusion ... 36
3 Acknowledgments
I would like to briefly thank some important people who made this thesis possible. First and foremost, Kees Hengeveld for his support, inspiration and help in conducting this extremely interesting and challenging research. Secondly, Enoch Aboh for taking the time and the effort to read, and question it. I am also grateful to my friends back in Italy who always encouraged me, even from a long distance, in particular Marianna Morao, Carlo Andretta, Serena De Valentin and Piera Stragliotto. Not to mention the incredible support system I have had in Amsterdam, not only during the thesis, but also during the two years of the Master’s: thanks to Felicia Bisnath, Catarina Loureiro Soares and Jorge Viterbo Ferreira for having been a home when I did not have a home. And, of course, Veronica Miatto for being my Italian friend outside Italy, and Jerrod Maddio for the proof-reading. I am also grateful to my parents, my sister and my cat, who always believed in me: without you, I would not have had the strength to go through this. Finally, I would like to thank Marco Guidolin, who thinks that I have no reason to thank him. Grazie.
4 List of abbreviations German 3=third person NOM = nominative ACC = accusative SG = singular M = masculine Polci 1= first person 2=second person ACC =accusative COP =copula DEICT =deictic INJ =injunctive S =singular SUBJ = subjunctive A’ingae 1 = first person 2 = second person 3 = third person ACC1 = accusative 1 ACC2 = accusative 2 ADD = additive focus ADJR = adjectivalizer ADVR = adverbializer AN = animate
ANA = anaphoric reference to entity or event ANA.LOC = anaphoric reference to location or
time ASS = assertive ATTR = attributive AUG = augmentative CAUS = causativizer CMP = comparative
CONTR = contrastive topic
DAT = dative
DIST = distal
DS = different subject cosubordinator
DUR = durative
EXCL = exclusive focus
FRT = frustrative HES =hesitation H = human IGN =ignorative IMP = imperative IMPF = imperfective INAN = inanimate INF = infinitive INT = interrogative IRR = irrealis ITER = iterative LOC = locative
MANN = manner, path NEG = negation
NEGP = negative predicate
NEW = new topic
NR = nominalizer
NR.ANG = nominalizer angular shape
5 NR.NLR = nominalizer linear shape
PL = plural PLS = plural subject PREC = precumulative PROSP = prospective PROX = proximal QUAL = quality Q1=question word 1 Q2=question word 2
REFL = reflexive pronoun
RPT = reportative
SG = singular
SIMIL = similative
SO =ablative
SS = same subject cosubordinator SUB = subordinator
6 1. Introduction
Conditional clauses and Topics are marked in similar ways in a considerable number of unrelated languages. A relation between these two entities has been acknowledged since Haiman (1978), who argues that Conditional clauses and Topics can be defined in the same way and that, consequently, Conditionals are, in fact, Topics. Such definitions rely on the idea that both Conditionals and Topics are entities whose existence is agreed upon by the speaker and his audience. As such, they constitute the framework which has been selected for the following discourse. In other words, they both refer to information presupposed by both the speaker and his audience and, in this sense, they establish the context for what comes next in the discourse.
A’ingae is a language of particular interest with respect to Haiman’s hypothesis. It is a language with no genetic affiliations, spoken by approximately 1400 people in Colombia and Ecuador. This language encodes New and Contrastive Topics with two distinct clitics which, at the same time, can mark Conditional clauses as well. One goal of this research is to investigate the use of the two particles using corpus data to corroborate Haiman’s hypothesis. Another goal is to investigate the prediction that the different Topic clitics mark different types of Conditionals (Ebert et al., 2014) in A’ingae.
In order to bring together Information Structure (hereafter IS) and Conditionals into one analysis, an existing framework is needed. The framework adopted to accomplish this purpose is Functional Discourse Grammar (Hengeveld & Mackenzie, 2008. Hereafter FDG). FDG is a typologically based, structural-functional theory of language. It has a top-down architecture with four distinct levels of analysis – Interpersonal, Representational, Morphosyntactic and Phonological – which interact with one another. The model starts with the speaker’s intentions and then works down to the articulation, referring therefore continuously not only to a Grammatical Component, but also a Conceptual, Contextual, and Output Component. Following a recent proposal in FDG, as will be further explained, a Conditional clause is marked with different combinations of functions and operators, all operating at the Representational Level, i.e. in semantics. IS, on the other hand, is encoded as a function at the Interpersonal Level, i.e. in pragmatics. It is the present research’s intention to bring together these two levels of analysis, studying how they interact with one another, to confirm that Conditionals are in fact Topics. The existence of a relation between these two levels appears to be confirmed by FDG considering that New and Contrastive Topics seem to mark the beginning of an Episode at the Representational Level in A’ingae. As will be further illustrated, the Episode is also the layer at which most of the Conditional clauses and the realis/irrealis distinction are marked.
Besides the Topic/Conditional issue, this research has another purpose, which is purely descriptive. A full grammar of A’ingae is in preparation (Fischer & Hengeveld, in prep.), but the
7
remaining literature concerning this language does not analyse, in detail, the use of these clitics. Hence, this study aims to widen the general knowledge with respect to this language.
This paper is structured as follows. In Section 2, A’ingae is presented, with particular focus on how IS and Conditionals are encoded in this language. In Section 3, some previous literature on the relation between Topics and Conditionals is reported. In Section 4, a description of FDG is provided. The intention of this section is to provide a summary of the basic features of this language model, paying special attention to the levels of analysis relevant for the discussion: namely, the Interpersonal Level and the Representational Level. In Section 5, the methodology and the specific research questions are outlined. The results are presented in Section 6 and discussed in Section 7. I conclude in Section 8.
2. A’ingae
A’ingae (ISO 639-3 identifier con, Glottolog code cofa1242) is a language isolate, endangered and yet still developing. It is spoken by approximately 600 people in Colombia and 800 people in Ecuador (Fischer & van Lier, 2011: 222). Despite its being more commonly known as Kofán, Kofane, or Cofán, this paper will refer to this language with the name used by the native speakers, A’ingae. This name consists of the stem A’i, ‘person’ and the manner clitic =ngae. As has been mentioned, a full grammar of A’ingae is currently in preparation (Fischer & Hengeveld, in prep). Other major publications include studies on different aspects of A’ingae’s language and culture. Apart from the study by Fischer & van Lier (2011), already mentioned, these include Borman (1962, 1976, 1977, 1981), Fischer (2002, 2007) and Tobar Gutiérrez (1995).
The most interesting trait of A’ingae’s morphology may be the extensive use of clitics. Only enclitics and suffixes are present, whereas proclitics and prefixes are not found (Fischer & Hengeveld, in prep: 10). Noun phrases and subordinate clauses present clitics in a fixed order which signals the role of the NP itself in the clause and in the discourse. Interestingly, this order includes IS (1).
(1) NP=Case=Focus=Givenness
An example is given in (2).
(2) Kha=nga=yi=ta other=DAT=EXCL=NEW
8
There are two Focus markers and two Givenness markers, as presented in Table 1.
Clitic Gloss Information status Example
=khe ADD Additive Focus afa me'in ña=khe se'je-y usha=mbi=gi.
say NEGP 1.SG=ADD heal-INF able=NEG=1
‘He said, "No, I can't heal you either.”.’ (BC20.100) =yi/ =ñi EXCL Exclusive Focus akhia tsuve=yi khûi a'ta.
just head=EXCL lie dawn
‘Just his head lay here at dawn.’ (BC19.042) =ta/ =nda NEW New Topic U'ti=e ungû=ta a'i.
first capuchin=NEW man ‘At first Capuchin was a man.’ (BC06.001)
=ja(n) CONTR Constrastive Topic fûe=sû afa ña=ja fathukhu=ve da=ya.
other=ATTR say 1.SG=CONTR cliff=ACC2 become=IRR
‘Another said, "I'm going to become a cliff.".’ (BC07.136)
Table 1. Clitics marking Information status
Since the present study is concerned with the relation between Topics and Conditionals, I will limit my discussion to the New and Contrastive Topic markers1. Both these clitics can mark a change of different elements of the discourse, including time (3) (4), location (5) (6), or participant (7) (8).
(3) Ûfa-je=ni=nda dû'shû a'tû=tsh=e kan khûtsû=fa. blow-IMPF=LOC=NEW child hide=QUAL=ADVR watch stand=PLS
‘When he blew, the young men being hidden stood and watched.’ (BC02.019)
(4) injan=mba ru'nda=ni=ja a'tû=tsh=e tutu-fû=chu-a=ma think=SS wait=DS=CONTR hide=QUAL=ADVR white-??2=SUB-ADJR=ACC1
un'jun tive=nga. bathe hand=DAT
‘Having thought, while he was waiting secretly, she bathed, rubbing white powder on her arms.’ (BC23.1.005)
1 Given Topics are either not overtly expressed or positioned as the last constituent of the sentence. 2 This is marked by Borman as a nominal classifier, however it does not occur elsewhere in the data.
9
(5) Da va=ni=nda tsa='kan=mbi.
HES PROX=LOC=NEW ANA=CMP=NEG ‘Here is not like that.’ (BC24.055)
(6) Ke yaya tse=ni=ja kan'jen. 2.SG father ANA.LOC=CONTR live
‘Your father lives there.’ (BC12.052)
(7) ña=nda=ngi utifa=pa ji=ya. 1.SG=NEW=1 crown=SS come=IRR
‘I will come wearing a headband.’ (BC12.075)
(8) Ña=ja asithae=ngi [kinikhu=ve da=ye]. 1.SG=CONTR think=1 tree=ACC2 become=INF
‘I think I’ll become a tree.’ (001-04-03-LC-Unfendyu'ndyu-027)
As is clear from examples (3-6), the locative clitic marker =ni can be used to mark either time (even though, in this case, it is not mandatory) or location, whereas this clitic does not occur when the Topic clitics mark a change of participant.
As noticed by Fisher (2007), the use of these two clitics in marking Topic is not always immediate. Specifically, he argues that while =ta/=nda is used to introduce or reintroduce a topic, =ja(n) is used to reactivate a Topic (sometimes through contrast or emphasis) that is either in the Common Ground or that was mentioned earlier, and which in any case is understood by the addressee.
In addition to these roles, these two clitics can also mark a Conditional clause. This is exemplified below in (9) for =ta/ =nda and in (10) for =ja(n).
(9) Tsa=mbi=e=ka=nda ukû=ve me’i=ye=kan.
ANA=NEG=ADVR=CMP=NEW biting_gnat=ACC2 NEGP=INF=CMP
‘If it hadn't been like that there wouldn't be any biting gnats.’ (BC06.028)
(10) Matichi ûkha=ni=ja ma'kaen. machete break.off=LOC=CONTR how
10
It seems clear from these examples that there is some sort of relation between Topics and Conditional clauses. In the next paragraph, I will summarise some relevant literature which has already acknowledged a relation between these two entities.
3. Conditionals are Topics
3.1. A definition of Topic
In the present research I adopt the definition of Topic provided by Krifka (2008), according to whom the main feature that distinguishes Topic from Focus is the absence of alternatives. Specifically, while a “property F of an expression α is a Focus property iff F signals (a) that alternatives of (parts of) the expression α or (b) alternatives of the denotation of (parts of) α are relevant for the interpretation of α.” (Krifka, 2008: 249), the “topic constituent identifies the entity or set of entities under which the information expressed in the comment constituent should be stored in the CG [Common Ground] content” (Krifka, 2008: 265). This does not necessarily imply that Topics always encode old information, but they can introduce new elements into the CG without them being in Focus. Specifically, this is the case of New Topics. Similarly, Topics can be Contrastive and, in this case, they combine features of Topics and of Focus into a single category. That is to say, they are constituted by a Topic that contains a Focus which, as explained above, indicates an alternative. An example of Contrastive Topic is reported in (11).
(11) A: What do your siblings do?
B: [My [SISter]Focus]Topic [studies MEDicine]Focus, and [my [BROther]Focus]Topic is [working on a FREIGHT ship]Focus.
(Krifka, 2008: 268)
3.2. Conditionals and Information Structure
Haiman’s 1978 article on Topics and Conditionals is the first study which endorses the idea that these two entities may actually be assimilated into one single definition. This idea arises from morphosyntactic evidence from a specific language analysed in the paper, namely Hua. Haiman’s results show how Conditional clauses and Topics (but also Questions) are all established in the discourse as given facts and are expressed in the same way in this language. For instance, they are both left-dislocated constituents, and Topic markers are Conditional markers. The definitions of the
11
two entities drawn by the article is the following: “A conditional clause is (perhaps only hypothetically) a part of the knowledge shared by the speaker and his listener. As such, it constitutes the framework which has been selected for the following discourse.” (Haiman, 1978: 583). Equally, “The topic represents an entity whose existence is agreed upon the speaker and his audience. As such, is constitutes the framework which has been selected for the following discourse.” (Haiman, 1978: 585). The conclusion of Haiman’s paper is thus that Conditional clauses and Topics signify the same thing and are equivalent to one another.
Starting from this hypothesis, Ebert et al. (2014) study German Topics, arguing that different types of Topics resemble different types of Conditionals. Consider (12) and (13).
(12) If Peter went shopping, then there is pizza in the fridge. (13) If you are hungry, there is pizza in the fridge.
(Ebert et al., 2014: 354)
Following Austin (1961), the type of Conditional in (13) is defined as a Biscuit Conditional (hereafter BC). The BC shows a Conditional in which the truth of the consequent does not depend on the truth of the antecedent. Conversely, (12) represents a Normal Conditional (hereafter NC) and shows the scenario in which the truth of the consequent does depend on the truth of the antecedent: that is, there is pizza in the fridge iff Peter went shopping. Ebert et al. (2014) argue that these two types of Conditionals are linked to German Left Dislocation (hereafter GLD) (14) and to Hanging Topic Left Dislocation (hereafter HTLD) (15).
(14) Den Pfarrer, den kann keiner leiden. The.ACC pastor the.ACC can nobody like
‘The pastor nobody likes.’
(15) Der/den Pfarrer, keiner kann ihn leiden. The.NOM/ACC pastor nobody can 3SG.M.ACC like
‘The pastor, nobody likes him.’
(Ebert et al., 2014: 364)
The article refers to Frey (2004) in claiming that GLD resembles aboutness topicality. That is, a sentence like the one reported in (14) is appropriate for a context in which a specific request for information about the left dislocated constituent has been made. Conversely, HTLD expresses
12
relevance topicality, a construction in which the Topic has a more general relevance and where a conventionalised expression (e.g. as for X, … or concerning/regarding X, …) is used to indicate that the fact expressed in the subsequent clause is pertinent with respect to an information seeking concerning X. Ebert et al. (2014) speculate that the similarity of the two different Conditionals and the two different Topics relies on three distinct factors, namely:
1) Prosodic Integration. In case of HTLD, the left peripheral phrase is separated from the rest of the sentence by a short pause, whereas the same is usually not observed in the case of GLD. Similarly, in BCs, the left peripheral if -clause is separated from the rest of the sentence by a short pause. By contrast, in the case of NCs with left dislocation, such a pause is either not found or tends to be shorter.
2) Resumption. GLD requires the presence of a resumptive d-pronoun which must be weak, and which is preferably realised in the prefield of the matrix clause. The left peripheral element must be marked for the same case as the resumptive pronoun. In the case of HTLD, such restriction does not apply and the resumptive element may occur in the form of a personal pronoun, a weak d-pronoun, a strong pronoun, an epithet, or a definite description. The position of such an element is also less restrained, since it can be found either in the prefield or in the middlefield of the matrix clause. Finally, the left peripheral element may either be in the nominative case or in the same case as the resumptive element. Concerning the Conditionals, Ebert et al. follow previous literature (e.g. Comrie 1986; Iatridou 1994; von Fintel 1994; Izvorski 1996) in arguing that then can be regarded as a proform which relates back to the possibilities introduced by the if -clause. Hence, it can be regarded as a resumptive element that corresponds to the weak d-pronoun in the case of the GLD constructions. As already explained, the presence of then is strongly linked to NCs.
3) Binding. Where in GLD binding of a pronoun contained within the left peripheral constituent by a quantifier in the matrix clause is possible, this is not the case for HTLD. Similarly, while binding into the if-clause seems to be possible in the NC reading, it is not possible in the BC reading.
As is clear from this summary, GLDs resemble NCs, whereas HTLDs are related to BCs. Therefore, not only do Topics look like Conditionals in multiple morphosyntactic respects in various languages, but different type of Topics can also be linked to different types of Conditionals. This conclusion is, for the moment, proven only for German, but it may be the case that this research will strengthen this hypothesis with data from another language. That is, it may be the case that in A’ingae the two clitics =ta/ =nda and =ja(n) are linked to different types of Conditional clauses, which do not necessarily have to be BCs and NCs.
13
Another interesting point investigated by this article is the possibility of stacking. Following Bhatt and Pancheva (2006) - who observed that, in case of stacking, ordering restrictions apply such that, in sentence-final if-clauses, those clauses belonging to NCs occur before those of BCs - Ebert et al. notice that the same restriction applies to left-dislocated clauses. Moreover, they point out that NCs can occur inside NCs, and that the same applies for BCs. The conclusion is that Topics can occur as nested elements and, specifically, that aboutness Topics can be nested inside relevance Topics.
A relation between Topics and Conditionals, also with respect to nested Topics, was also investigated in Sign Languages by Pfau (2008). These two entities are similar in NGT (i.e., Sign Language of the Netherlands) and in ASL (i.e., American Sign Language), given that they are both sentence initial, they present similar non-manual markers (NMMs) and they constitute independent prosodic units. However, there are some relevant differences. For instance, Liddell (1986) argues that head thrust (HT) is found in ASL conditionals but not in Topics. Pfau (2008) also notices that embedded topicalisation is possible in Sign languages: that is, it is possible for Topics to appear within a Conditional Clause. Consider, for instance, example (16) from NGT.
_____________________________________cond
(16) a. SUPPOSE CARi INDEX1 FATHER ti LEND AUX1, 1VISIT2 ‘If the car my father lends to me, I will visit you.’
________________________________cond ______neg b. SUPPOSE BOOK INDEX3i INDEX2 ti READ, SLEEP CAN^NOT ‘If this book you read, you will not be able to sleep.’
(Pfau, 2008: 9)
While the eyebrows remain raised throughout the Conditional clause, the Topic may additionally be marked by a NNM, such as head forward and/or sharp head nod. This indeed suggests that it is possible for Topics to appear within the Conditional clause in NGT.
All these studies indeed suggest that Conditionals are Topics. However, other studies argue that Topics and Conditionals are, to a certain extent, dissimilar. For instance, Iatridou (1991: 53) distinguishes between Hypothetical Conditionals and Relevance Conditionals, which are comparable, respectively, to NCs and BCs. She claims that in NCs both the protasis and the apodosis are part of the assertion. Conversely, in the BCs, the if-clause, in contrast with Topic, is not presupposed. Another argument against the idea of the equivalence between Topics and Conditionals is adduced
14
by Caron (2006), who argues that Conditionals share properties with Antitopics, which makes them incompatible with the status of presupposed Topics. Specifically, he shows that in some African languages, such as Polci, Zodi and Banda Linda, Conditionals are syntactically and morphologically marked as Focus. For instance, as shown in (17), in Polci focused elements and Conditional clauses both appear in the left periphery and are both marked by the copula kən.
(17) a. Wún gi kən yu ɲen a ga: gi. Girl DEICT COP pour milk in calabash DEICT
‘The girl poured milk into the calabash.’
b. Gǎrbà kən nʤaŋ ɬo: wú ɗe kə fǔ:-m. Garba COP cut meat ACC INJ 2S.SUBJ tell-1S
‘If Garba slaughters an animal (lit. GARBA slaughters an animal), tell me.’
(Caron, 2006: 76)
However, he does not conclude that the Conditional clause is the Focus of the Conditional construction, given that this would make the apodosis the presupposition of the Conditional clause. As a solution, Caron (2006) proposes that protasis [p] and apodosis [q] are in a changing relation in which [p] implies [q] and nothing is said about [p]. In this sense, the protasis is an assertive fiction and the Conditional clause is a frame fiction.3 The marking of Conditionals as Focus in the analysis that the present research will conduct should not be completely surprising as, in A’inage, Conditionals are not marked as Given Topics, but as New and Contrastive Topics, which could explain why, in the languages analysed by Caron (2006), the Focus marker for Conditionals is preferred over the Topic marker.
In the next paragraph I will delineate the main features of the approach adopted in the present research, namely Functional Discourse Grammar (Hengeveld & Mackenzie, 2008).
4. Functional Discourse Grammar (FDG)
4.1. General outline
As is clear from the name, FDG is a functionalist theory of language. As such, it accounts not only for a Grammatical Component (Formulation and Encoding) and an Output Component (Articulation),
15
but for Conceptual and Contextual components as well. The Conceptual Component “contains the Speaker’s communicative intention and the strategies that he wishes to deploy in order to achieve that intention” (Hengeveld & Mackenzie, 2008: 47). On the other hand, the Contextual Component accounts for the fact that “the intention developed by the speaker does not arise in a vacuum, but in a multifaceted communicative context” (Hengeveld & Mackenzie, 2008: 9). According to FDG, both the Conceptual and the Contextual Components interact with the Grammatical one, which is the core of the theory. Specifically, FDG has a top-down architecture, following the way language is processed. This architecture is organised in four different levels of analysis. The first one is the Interpersonal Level (hereafter IL): it deals with the intentions of the speaker towards the addressee and consists of different layers including Moves, Discourse Acts, Illocutions, Speech Participants, Communicated Content, Ascriptive Subacts, and Referential Subacts. The second level, the Representational Level (hereafter RL), is concerned with the semantic aspects of linguistic units and consists of the following layers: Propositional Contents, Episodes, States-of-Affairs, Properties, Individuals, Locations and Times. The third level is the Morphosyntactic Level (hereafter ML), which, together with the last level of analysis, the Phonological Level (hereafter PL), is concerned with the encoding of the information from IL and RL. The former consists of Linguistic Expressions, Clauses, Phrases and Words. The latter consists of Utterances, Intonational Phrases, Phonological Phrases, Phonological Words, Feet, and Syllables. Each layer has the structure presented in (18).
(18) (π V1:H(V1): ΣN(V1))ϕ
In (18) H represents the head of the layer V. The other elements are optional and they are the following: Σ is the modifier of the head, π the operator of the variable and ϕ the function of the unit as a whole. Whereas the modifier represents a lexical strategy which further restricts the head H, operators and functions represent grammatical strategies. The difference between them, as will be further explained, consists in the latter being relational in nature.
In the following paragraphs I will explain how IS and Conditionals are defined and how they are encoded in the FDG framework.
4.2. Information Structure
In FDG, Information Structure is encoded at the IL. The role it plays within this level is the subject of some debate in the literature. I will follow the original proposal of Hengeveld & Mackenzie (2008), which states that Topic and Focus are pragmatic functions (but see, for instance, Smit (2010) for
16
another interesting proposal). Functions (ϕ in (18)) represent grammatical strategies that a language uses to specify the function of a layer within a level. They are relational in nature: they hold between the entire unit and other units at the same layer (Hengeveld & Mackenzie, 2008: 14). Therefore, the sentence in (19a) may be given the interpersonal representation as in (19b).
(19) a. (What did John buy?) John bought two books. b. IL: (R1)TOP (T1) (R2)FOC
(Keizer, 2015: 179)
Focus is defined by Hengeveld & Mackenzie (2008: 89) as a function that “signals the Speaker's strategic selection of new information, e.g. in order to fill a gap in the Addressee's information, or to correct the Addressee's information”. It exists in opposition to the Background, even though the marking of the Background seems to be rare in natural languages. On the other hand, Topic is defined as a function “assigned to a Subact which has a special function within the Discourse Act, that of signalling how the Communicated Content relates to the gradually constructed record in the Contextual Component” (Hengeveld & Mackenzie, 2008: 92). Information that does not receive Topic function constitutes the Comment.
4.3. Conditionals
The definition of a Conditional clause is also a controversial subject in the literature. I will follow Sweetser (1990) in distinguishing three domains of Conditionals, namely: Content conditionals, Epistemic conditionals, and Speech-act conditionals.
In the case of a Content conditional clause, the realization of the protasis is a sufficient condition for the realization of the apodosis. This is shown in (20), where Mary’s going to the party is a sufficient condition for John also going to the party.
(20) If Mary goes, John will go.
(Sweetser, 1990: 114)
On the other hand, Epistemic conditionals refer to those if-then clauses where the knowledge of the truth of what is expressed in the protasis is a sufficient condition to conclude the truth of the apodosis. Consider, for instance, (21).
17
(21) If she’s divorced, (then) she’s been married.
(Sweetser, 1990: 116)
In (21), knowing that it is true that “she’s divorced” is sufficient condition to conclude that it is true that “she’s been married”. The third and last type of Conditional identified by Sweetser is the Speech-act conditional. In this case, it is the performance of the speech-Speech-act in the apodosis that is a sufficient condition for the fulfilment of the protasis. In other words, “the state in the protasis enables or causes the following speech act” (Sweetser, 1990: 118). For instance, consider (22).
(22) If I may say so, that’s a crazy idea.
(Sweetser, 1990: 118)
In (22) an opinion is expressed conditionally on a permission given by the addressee. Note, however, that in some sense the conditional speech act is always accomplished given that the politeness condition does not prevent the speaker from actually stating his/her opinion in (22).
The distinction of Conditionals in these three domains is important for this research, as the definition of such a construction is not as straightforward as it may seem. Moreover, as already explained in Section 3.2., given that different types of Topics could mark different types of Conditionals in A’ingae, such a distinction may be useful to check potential differences in the use of the two clitics.
Within the FDG framework, Conditionals seem to be marked at RL, i.e. in semantics. That is, given that they allow “a Speaker to create a possible world in which another Propositional Content is true” (Keizer, 2015: 305), the first impression is that they are all marked with the hypothetical modality, which is treated as a subdivision of the category of subjective epistemic modality at the Propositional Content layer at the RL. Here it has the role of operator (i.e., it marks a grammatical strategy) but, unlike functions, it applies only to the unit itself and, therefore, does not have a relational nature. This is confirmed in Hengeveld & Mackenzie’s original proposal (2008) and is exemplified in (23) for English.
(23) a. If he comes, (I’ll leave). b. RL: (hyp p1: […])
18
However, I propose that this operator applies only to Epistemic conditionals, considering that this domain presents a Propositional Content layer, whereas the same is not true for Content conditionals. Hence, Epistemic conditionals are formed by two Propositional Content layers, each containing an Episode, which is defined as “one or more States-of-Affairs that are thematically coherent, in the sense that they show unity or continuity of Time (t), Location (l), and Individuals (x)” (Hengeveld & Mackenzie, 2008: 157). On the other hand, Content conditionals are formed by two Episodes, hence the apodosis is not marked with the hypothetical operator. In both cases the realis/irrealis distinction is encoded as an episode-oriented epistemic modality operator, which indicates the likelihood of the Episode taking place. Furthermore, the protasis is marked by a Conditional function, which clarifies its relationship with the apodosis. In the case of Content conditionals, such function is marked at the Episode layer, whereas, in the case of Epistemic conditionals, it is marked at the Propositional Content layer. This is exemplified in (24).
(24) a. Epistemic Conditionals: (hyp p1: (real/irreal ep1))cond (p2) b. Content Conditionals: (real/irreal ep1)cond (ep2)
On the other hand, this distinction does not seem to hold when it comes to Speech-act conditionals. This third type of Conditional seems to be marked at the RL in the exact same way as Content conditionals. However, as mentioned above, in Speech-act conditionals the condition is always accomplished and, in this sense, such Conditionals differ from the Content and the Epistemic domains. I propose that this distinction is encoded at the IL, i.e. in pragmatics. First of all, I assume that protasis and apodosis are encoded at the IL as two separate Discourse Acts, given that they can carry different Illocutions. Consider (25).
(25) If I may ask, what are you thinking about?
IL: (M1: ((A1: (F1: DECL (F1)) (P1)S (P2)A (C1)) (A1))Cond ((A2: (F2: INTER (F2)) (P3)S (P4)A (C2)) (A2))) (M1))
In (25) the protasis is a declarative sentence and the apodosis an interrogative one, which shows that there are two distinct Discourse Acts. I propose that, in the case of Speech-act conditionals, there is a relation of dependence between the two Discourse Acts and that, in particular, the protasis is marked with a Conditional function. Although this function is not acknowledged in Hengeveld & Mackenzie (2008), it is already mentioned as existing (at least in English) in Keizer (2015: 262), even though it is used for Content conditionals.
19
Dik et al. (1990) also distinguish different types of Conditionals that are comparable with Sweetser’s division, but in the context of Functional Grammar (hereafter FG). Since the FG framework is theoretically largely related to FDG, I will briefly discuss their proposal. In this article, Dik et al. include the Conditionals in the discussion of satellites, that is, adverbial lexical optional constituents of the clause. Following Hengeveld (1989), Dik et al. (1990) distinguish four types of satellites, namely:
- Predicate satellites: these specify additional information on the States-of-Affairs designated by a predication.
- Predication satellites: these specify additional information on the State-of-Affairs, locating it in a possible world. Hence, predication satellites delimit the set of possible referents to external situations.
- Proposition satellites: these specify additional information on the speaker’s attitude towards the proposition.
- Illocutionary satellites: these specify those lexical strategies through which the speaker modifies the illocutionary force of the illocution.
In FG Content conditionals are treated as Condition satellites within the class of Predication satellites. An example is given in (26).
(26) He will take his umbrella in case of rain.
(Dik et al., 1990: 34)
This satellite specifies a State-of-Affairs, on the occurrence of which depends the possible occurrence of another State-of-Affairs. Therefore, Content conditionals seem to be marked at the States-of-Affairs layer, and not at the Episode layer, as the present proposal suggests. On the other hand, in FG, Epistemic conditionals are considered Condition satellites within the class of Proposition satellites (27).
(27) If the lights are on, he must be home.
This difference is motivated by the idea that in Propositional satellites the speaker does not evaluate the State-of-Affairs, but the Propositional Content. In this sense, this proposal is closer to the one suggested by FDG and by the present study itself, given that the hypothetical modality is treated as an operator of the Propositional Content layer in Epistemic conditionals. Finally, Speech-act conditionals are treated in FG as Illocutionary satellites. Consider, for instance, (28).
20
(28) John has left, in case you haven’t heard.
(Dik et al., 1990: 39)
The placement of this type of Conditional in this category is motivated by the fact that it indicates a “condition on the felicity of the speech act” (Dik et al., 1990: 39). Therefore, the main difference between this Conditional and the other Conditionals is the fact that it refers to a unit with an illocutionary value, hence to pragmatics and not to semantics. In this case, FG’s proposal is strongly related to that of the present research, given that I have argued that Speech-act conditionals diverge in being marked at the IL rather than at the RL. Thus, it seems possible to distinguish these three domains - content, epistemic, speech-act - in FDG.
The idea that the realis/irrealis distinction is marked at the Episode layer comes from Hengeveld (2011, but see also Dall’Aglio Hattner & Hengeveld, 2016), whereas in Hengeveld & Mackenzie’s original proposal (2008) it operated at the State-of-Affairs layer. The idea that Conditionals are marked at the Episode layer is crucial for the present research, considering that A’ingae’s clitics seem to mark both Conditionals and the beginning of an Episode.
5. Research Questions & Methodology
The specific research questions this research aims to answer are:
- What are the uses (both individually and compared to each other) of =ta/=nda and =ja(n) in A’ingae?
- Are the two clitics exclusively markers of IS, or can they be considered Episode markers as well?
- Do =ta/=nda and =ja(n) mark different types of Conditionals?
Given that the clitics can signal a Conditional clause, which is marked at the Episode layer in FDG, it is reasonable to predict that both clitics are not only marker of IS, but also of Episode. Furthermore, I hypothesise that, although it is possible that the clitics mark different types of Conditionals, such a differentiation has not been shown to be relevant for a large number of languages, since the only study on the matter, that I am aware of, is the one of Ebert et al. (2014) on German. Hence, it is not yet clear if it is possible to generalise and extend this result to A’ingae, but the present research’s results will provide more evidence either in favour of, or against, this possibility.
The analysis will be based on a pre-existing set of data, consisting of legends narrated by one single monolingual A’ingae speaker (Enrique Criollo) around 1960 which were then analysed by Borman in 1990. The purpose of this data collection was not to analyse the language itself, but to give
21
an account of the enormous amount of legends possessed by this culture. In the present research, these data are coded by the abbreviation ‘BC’ followed by the legend number and the line number within the legend. Moreover, further examples from other major publications on A’ingae are used to analyse some grammatical phenomena when the data are not clear enough. In this case, the example is coded according to the secondary source.
The data will be analysed as follows. All examples containing the New and/or Contrastive Topic clitic will be analysed by checking their usage as regards:
- Marking of IS
- Marking of Conditional clause
- Marking of the beginning of the Episode
The goal is to verify that =ta/nda exclusively marks New Topic and =ja(n) only marks Contrastive Topic with respect to IS. On the other hand, by checking the marking of the Conditional clause, the investigation would like to verify whether there is any difference in use between the two clitics or if one or the other is used indiscriminately. If this latter case applies, the hypothesis that different Topics mark different types of Conditionals would be rejected. Finally, by checking if both clitics can mark the beginning of the Episode, the research aims to check if, given that the Conditionals are marked at the Episode layer in FDG, the clitics are markers of Episode. The data will be qualitatively analysed and described in Section 6, in which examples will be reported. The results will then be discussed in Section 7.
6. Results
6.1. New Topic
The data analysis shows that =ta/=nda is most of the time used to introduce (or reintroduce) a Topic. In fact, in 152 out of 222 tokens containing this clitic, it is possible to interpret it as a marker of New Topic. Within these 152 examples, 97 show a change of participant, 42 a change of time, and 13 a change of location. Contextualised examples of each type are given below.
(29) a. atesû=mbi=e ana=mba khaki a'ta tise pûshe=ve me'i=un. know=NEG=ADVR sleep=SS next_day dawn 3.SG wife=ACC2 NEGP=ADVR
22
b. me'i-u=si athe=pa ma=ni=ngae=tsû ja ña pûshe=ta. NEG=ADVR=DS see=SS Q1=LOC=MANN=3 go 1.SG wife=NEW ‘Seeing she was not there he asked others, "where did my wife go?"’ (BC12.014)
In (29a), the element wife is introduced. In (29b) the speaker asks the addressee where his wife went, introducing the element wife as New Topic. A similar example is shown in (30) for time.
(30) khaki=ta ja~ja=fa dûtshi'ye=ndekhû. next_day=NEW go~ITER=PLS child=H.PL
‘The next day the young men went out.’ (BC03.001)
The same phenomenon is found for location, as exemplified in (31).
(31) tsa=‘ma=tsû ju=ni Pasto=ni=nda=te kan'jen.
ANA=FRT=3 DIST=LOC Pasto=LOC=NEW=RPT live
‘But it is said that she lives in Pasto.’ (BC26.122)
In (31) Pasto is marked both with the locative clitic =ni and with the New Topic clitic =nda, marking a location change.
Despite this being the most attested use of =ta/=nda, these clitics where also found in other contexts. Let us start with tail-head linkage (hereafter THL). THL refers to a fairly common phenomenon concerning (dis)continuity in language. It occurs when a language links two sentences by repeating the last element of a sentence as first element of the sentence that follows. This phenomenon is attested in A’ingae, and the head part of the THL is occasionally marked with =ta/=nda. 21 out of 222 of the tokens containing this clitic in the set of data use it to mark THL. An example is reported in (32).
(32) a. tsai=mbi=tsh=e a'jû. much=NEG=QUAL=ADVR vomit
‘He vomited much.’ (BC12.026)
b. a'jû=ni=nda tetete sata=ma ankan=e=mba khûtsû-an=fa. vomit=DS=NEW savage spear=ACC1 secure=ADVR=SS stand_up-CAUS=PLS
23
This THL use seems to be generalised in certain contexts in which the clitic marks a change of Episode. In other words, the clitic marks not a specific change in either time or location or participant, but marks a more general discontinuity with the sentence that it is following. We consider these as cases of dummy head THL, since the repeated verb is a dummy verb. This particular construction was found in 2 examples, and one is reported in (33).
(33) a. ana-ye andûfa-khu phi-ña=ni=nda ana=mba sleep-INF carry-NR.ANG place-CAUS=DS=NEW sleep=SS
ana-je=nda kanjansi=ve=ta di'sha. sleep-IMPF=NEW boa=ACC2=NEW change
‘But when it slept in the carrying sling it became a boa.’ (BC14.036)
b. sin-fa-khu tsen'tse-fa-khu=ta khûi. black-NR.LAT-NR.ANG pattern-NR.LAT-NR.ANG=NEW lie
‘Then it was black and spotted.’ (BC14.037)
c. tsun=nda tise mama ji=pa uya-en khendya-en=ni=nda do=NEW 3.SG mother come=SS move-CAUS awake-CAUS=DS=NEW
a'i=ve=ta da-je.
person=ACC2=NEW become-IMPF
‘When its mother came and waked it, it became a person.’ (BC14.038)
As is clear in (33c), the presence of the clitic =nda on tsun (‘do’) marks a general switch of Episode through a dummy verb.
The =ta/=nda clitic was also found on other types of heads, that is, it is used on connectors that appear either in first position or in second position immediately after the (dummy head) THL, marking the beginning of an all new Episode. This use was found in 15 examples and the most common connectors are the temporal adverb tse’i ‘then’ (34) and the demonstrative tsa, which indicates an anaphoric reference to entity or event (35).
(34) tse’i=ta khase asûkhu va=ni=ngae ji=khia='kan=en then=NEW again look_up PROX=LOC=MANN come=SIMIL=CMP=ADVR
24
ji~ji=fa an=ni=ngae setsa=ni=ngae. come~ITER=PLS downriver=LOC=MANN prow=LOC=MANN ‘Then, when she looked up again, they were going down river.’ (BC24.038)
(35) a. tisû-pa Chiga Kitsa=ma ma'pi. REFL-NR god father=ACC1 flog
‘He himself beat Father God.’ (BC26.048)
b. ma'pi=pa khitsa=pa anga tsaikhi nane munda=ma=ki khitsa. flog=SS drag=SS carry trail really peccary=ACC1=2 drag
‘Having beaten him he dragged him on the trail, bringing him just as you would drag a peccary.’ (BC26.049)
c. andû=pa=khe anga ji=mbi. carry=SS=ADD carry come=NEG
‘He didn't bring him by carrying him.’ (BC26.050)
d. tsu=mba tsa=ngae=ta Chiga Chan=ja kan dyai. do=SS ANA=MANN=NEW god mother=CONTR watch sit ‘Then God Mother sat watching all of this.’ (BC26.051)
Another use of the clitic was found in 5 examples, in which =ta/=nda is not classifiable in any of the abovementioned categories and seems to carry some sort of finality meaning. An example is reported in (36).
(36) tsu=mba manda dû'shû=ma yaya-kashe’ye=ma iñajan-sepa-en-ye do=SS command child=ACC1 father-old_man=ACC1 interrogate-V-CAUS-INF
mi=ngae=ti=ki ke juva=ka=en tshan=mba
Q2=MANN=INT=2 2 DIST=CMP=ADVR pluck=SS
pa-ji=si thesi=u ji=ni=nda iyu ji=pa
25
sintû=ni=nda. bite-IMPF=NEW
‘Then they told the children to ask Grandfather, "What would you do, with your eyes pecked out like that, if a jaguar came or if a snake came to bite you?"’ (BC02.046)
In (36), =nda attaches to the verb sintû ‘to bite’, returning a finality meaning. As is clear, this example cannot be classified either as a change of time/location/participant or as a Conditional, and neither can it be classified as (dummy head) THL.
Finally, 5 tokens were excluded because they were not clear enough. Consider, for instance, (37).
(37) a. tsa=ka=en kan'jen~ː dû'shû=ma=nda isû
ANA=CMP=ADVR live-DUR child= ACC1= NEW give_birth
‘Later she gave birth to a child.’ (BC14.034)
b. isû=pa na a'i-khu=ta a'i=ve=ta a'i=ve be_born=SS flesh people-NR.ANG=NEW people=ACC2=NEW people=ACC2
isû=ma
be_born=ACC1
‘When it was born its body was like a person's.’ (BC14.035)
In (37b), the use of the clitics is not clear. The noun a’i is repeated three times and marked in three different ways. Considering that I cannot catch the meaning the clitic has in this context, these and another three tokens were excluded. The remaining 22 examples containing the clitic can be categorised as Conditionals. The examples of this use are set aside for section 6.3.
6.2. Contrastive Topic
The number of tokens containing the clitic =ja(n) was smaller than the one containing =ta/=nda. In fact, only 47 tokens were found. 40 of these can be considered to be marking Topic, even though not all of them are clearly Contrastive Topics. The canonical use marking a contrastive participant is reported in (38).
26
(38) a. tise chan tuya injan-je. 3.SG mother yet remember-IMPF
‘His mother still remembered.’ (BC12.047)
b. dû'shû=ja in'jan-ña=mbi. child=CONTR remember-CAUS=NEG
‘The boy didn't remember.’ (BC12.048)
Clearly, in (38b) the subject dû'shû ‘child’ is in clear contrast with the subject of the previous sentence, chan ‘mother’. A similar example is reported in (39) for the marking of contrastive location.
(39) a. ke yaya tse=ni=ja kan'jen. 2.SG father ANA.LOC=LOC=CONTR live
‘Your father lives there.’ (BC12.052)
b. va=ni aipa=ndekhû kan'jen=fa. PROX=LOC savage=H.PL live=PLS
‘These are savages that live here.’ (BC12.053)
In (39a) the location where the father lives, translated in English with ‘there’, is marked so that the contrast with the location where the savages live (‘here’ in the English translation) is made clear.
The use of this clitic in marking contrastive time, on the other hand, is not as clear. Consider, for instance, (40).
(40) tse’i ja=pa vachu kan=ni=ja na'e=ni kan'jen avû. then go=SS net try-IMPF=CONTR river=LOC live fish
‘Then (when men) went to net fish, there were fish in the river.’ (BC03.060)
In seems evident that (40) is a case of time; however, it is hard to catch the contrastive reading, since the whole preceding context is based on this fishing scene.
Examples (38-39) show that it is indeed possible for =ja(n) to mark a Contrastive Topic. Nevertheless, the number of examples for which this is not true is considerable. One example was already reported in (40), and now I will report other problematic examples for the definition of =ja(n) as Contrastive Topic clitic. Let us start with some examples of the clitic attaching to a discourse
27
participant. 34 examples were found, but only 20 can actually be considered as Contrastive Topics. Consider, for instance, (41).
(41) re'ri-khu tsa'u-ña=mba ke=ja va-thi khûi=ja. small-NR.ANG house-CAUS=SS 2.SG=CONTR PROX-NR.LOC lie=IMP
‘When it was built they said, "You lie here."’ (BC15.068)
The construction exemplified in (41), where =ja(n) marks a participant of the direct speech, is fairly common, as it is found in 7 examples. In these cases, it seems that the clitic is used to mark Focus instead of Contrastive Topic. This seems to be the case also for constructions other than direct speech, as exemplified in (42).
(42) ña=khe=ti ega=ve da=ya khen asi'thaen=mba 1.SG=ADD=INT evil=ACC2 become=IRR thus think=SS
tsa'ndû=ja fi'thi. husband=CONTR kill
‘Having thought, "Shall l also become evil like that?" the husband killed her.’ (BC23.2.026)
(42) shows =ja(n) used to mark Focus in a context other than direct speech. Some other cases turned out to be more problematic, as shown in (43).
(43) ka'ni-an ja-yi=si Napu=ja ja=pa bu'mbu tsûtu-pa=ne enter-CAUS go-PROSP=DS Quichua=CONTR go=SS palm end-NR=SO
mandyi indi. seize catch
‘They say that the Quichua ran out and took hold of the other end of the palm.’ (BC14.073)
In a sentence like (43), the participant Napu being introduced for the first time, we would expect the use of the clitic =ta/=nda instead of the clitic =ja(n).
28
The non-contrastive use of =ja(n) is present not only for participant and, as reported in (40), for time, but also for location. However, only 4 and 3 tokens containing the clitic were found respectively for location and time. 3 out of 4 examples of location marked with the clitic are indeed contrastive. Consider, however, (44).
(44) a. khûi=ni=ja tise pûshe=ja a'ta=ni=ngae ji='ya arapa lie=DS=CONTR 3.SG wife=CONTR dawn=LOC=MANN come=ASS rooster
seyu-je=ni. crow-IMPF=LOC
‘His wife returned at dawn when the rooster crowed.’ (BC23.2.024)
b. tsûi ji-ña=si athe=pa tse-thi=ja=te
walk come-CAUS=DS see=SS ANA.LOC-NR.LOC=CONTR=RPT
pu'ta-en kati. shoot-CAUS destroy
‘It is said that he saw her as she came walking in and he shot her.’ (BC23.2.025)
In (44b) =ja is attached to a location which does not appear in the translation. This location, however, does not seem to contrast with anything and would more likely be interpreted as Focus.
An example reporting the problematic interpretation of =ja(n) marking time was reported in (40). Similarly, consider (45).
(45) a. khen asi'thaen=mba ru'nda. thus think=SS wait
‘Thinking thus he waited.’ (BC23.1.010)
b. me'i. NEGP
29
c. bia=ni ja-je khen in'jan=mba khaki=ja khaki far=LOC go-IMPF thus think=SS next_day=CONTR next_day
kuse athe. night see
‘Thinking that she went far, the next day when it was night he saw it.’ (BC23.1.012)
In (45c) =ja seems to be marking a New Topic more than a Contrastive one.
Finally, =ja(n) was also found marking a head in 3 examples. In 2 of them, it seems that the contrastive reading holds, as shown in (46).
(46) a. ji=pa tsa'u=ni ji=pa ana-je=chu=ni ru'nda. come=SS house=LOC come=SS sleep-IMPF=SUB=LOC wait
‘Having come to their house, he lay waiting in the bed.’ (BC23.2.023)
b. khûi=ni=ja tise pûshe=ja a'ta=ni=ngae ji='ya arapa lie=DS=CONTR 3.SG wife=CONTR dawn=LOC=MANN come=ASS rooster
seyu-je=ni. crow-IMPF=LOC
‘ His wife returned at dawn when the rooster crowed.’ (BC23.2.024)
In (46b) =ja on the head seems to mark the action performed by the husband in opposition with the action performed by the wife. However, consider (47).
(47) a. kan'jen=mba ana=mba bitha='ya. live=SS sleep=SS awaken=ASS
‘Staying and having slept he awoke.’ (BC23.2.006)
b. bitha=ni=ja me-thi=e jangi ja-je. awaken=DS=CONTR no-NR.LOC=ADVR arise go-IMPF
30
In (47b) the clitic seems to have the same use as the =ta/=nda has in marking THL. That is, it marks a general discontinuity, but it does not seem to strictly contrast with the preceding clause.
An interesting and unusual example was also found to be marked with this clitic (48).
(48) injan=mba ru'nda=ni=ja a'tû=tsh=e
think=SS wait= DS=CONTR hide=QUAL=ADVR
tutu-fû=chu-a=ma un'jun tive=nga white-??4=SUB=ADJR=ACC1 bathe hand=DAT
‘Having thought, while he was waiting secretly, she bathed, rubbing white powder on her arms.’ (BC23.1.005)
In (48), it seems that the verb ru'nda is used to highlight the contrast between the action of one of the participants (waiting), as opposed to that of the other (bathing), using the clitic to create some sort of scene change. This was the only clear example of this use of =ja(n) that was found in the data.
Finally, 3 tokens were excluded because they were not clear enough. An example is reported in (49).
(49) tse’i kuankuan afa tisû-pa-pura ke=ja junguesû=ve then trickster say REFL-NR-N 2.SG=CONTR IGN.INAN=ACC2
ke=ja=e da=ya. 2.SG=CONTR=ACC2 become=IRR
‘Then the trickster said to his companions, "What are you going to become?"’ (BC07.134)
This example was excluded because of the order of the clitics in the word marked in ke=ja=e. In fact, as mentioned in (1), the canonical order of the clitics in A’ingae is NP=Case=Focus=Givenness. However, in (49) the order is NP=Givenness=Case and, considering that it is not possible here to understand and solve this discrepancy, I have chosen not to include the example in the analysis. The other example was excluded for the same reason. Other uses, such as finality, dummy verb head marking, or other types of head marking were not found for =ja(n).
31
6.3. Conditionals
Interestingly, no Conditionals marked with =ja(n) were found in the dataset. This already suggests a larger use of =ta/nda in marking Conditional clauses. However, in order to fully answer the research questions, examples from other datasets were analysed for a total of 62 examples, giving a more complete picture of the use of the Topic clitics in Conditional clauses. Of these 62 examples, 40 were marked with =ta/=nda, 9 with =ja(n), and the remaining 13 were marked neither with the one nor with the other.
Following Sweetser’s (1990) distinction, no Epistemic conditionals were found. 60 examples are classifiable as Content conditionals and 2 as Speech-act conditionals, one unmarked (50) and one marked with =ta (51).
(50) tsun tsa=ka=en=ti=ki me=chu. do ANA=CMP=ADVR=INT=2 no=SUB
‘If that is so, where is she?’ (BC15.011)
(51) Tuya'kaen majan chava-ye in'jan=ta=tsû tsa-ki=e
Moreover IGN.AN buy-INF want=NEW=3 ANA-NR.LNR=ADVR
ja-je='fa=ya Corombia=ni. go-IMPF=PLS=IRR Colombia=LOC
‘And if someone wants to buy something, they know how to go to Colombia by that river.’ (20040218-EC-Interview-145)
Most of the Conditionals appear to be realis. Only 3 irrealis examples were found, all marked with the New Topic clitic; one of them is reported in (52).
(52) tsa=mbi=e=ka=nda ukû=ve me'i-ye=kan.
ANA=NEG=ADVR=CMP=NEW biting_gnat=ACC2 NEGP-INF=CMP
‘If it hadn't been like that there wouldn't be any biting gnats.’ (BC06.028)
In (52) the story of how these biting gnats came to the world is told. The story calls to mind the myth of Pandora’s vase: given that the vase was opened, and that the gnats are in fact in this world, it is not possible to account for this example in any different way than as an irrealis Conditional.
32
Another interesting feature about Conditionals in A’ingae is found in those marked with Contrastive Topic, as these are all marked with the clitic =ni. One example is reported in (53).
(53) Ke=ve in'ja=mba indi='fa=ni=ja ke=ja
2.SG=ACC2 think=SS hold=PL=LOC=CONTR 2.SG=CONTR
asi'thaen=mbe va=ni kanse=ja. think=NEG PROX=LOC live=IMP
‘If they want you, you stay here without worrying.’ (20060119-AnC-Cunsiana-01-0187.19)
This clitic is reported as locative in Fisher & Hengeveld (in prep.). However, Borman suggests that it is not only a locative clitic, but is a different subject cosubordinator as well, next to the clitic =si, which appears to have the same function. Similarly, Fischer (2007) argues that the construction =ni=ta ‘=LOC=NEW’ is used in A’ingae to mark different subject condition clauses. Hence, when
appearing in Conditionals, I will consider the clitic as a different subject cosubordinator marker. Finally, it is interesting to point out that, out of the 13 examples that were not found to be marked with either =ta/=nda or =ja(n), 6 were marked with the different subject cosubordinator clitic =ni/=si and 3 with the same subject cosubordinator clitic =pa. (54) and (55) show examples for =ni and =pa, respectively.
(54) sisipa=ve da=ni=khe a'i pa'fû=ya' tsa'u khûtsû-an-ye. sand=ACC2 become=DS=ADD man dig=IRR house stand-CAUS-INF
‘If we become sand, people will dig in us to set up their houses.’ (BC08.037)
(55) Pûshe=pa=tsû ethi=ni kanje=ña khen=de in'ja=’ña. marry=SS=3 house=LOC live=IRR thus= RPT think=ASS
‘They thought that if he marries he will stay in the house.’ (20040215-02-LC-Eresiune-013)
The remaining 4 examples do not appear to be marked in any particular way. One of them is the Speech-act conditional already reported in (50) and the others do not show consistency in marking. One example is reported in (56).
33
(56) dishasha'khu sumbu-ye vûe=tsû pûtsa-khu-en=sû ja-yi. biting_ant emerge-INF now=3 brave=ATTR-CAUS-NR.ANG go- PROSP
‘If a biting ant came out, "Now my enemy is getting away!"’ (BC26.071)
As is clear, there is no apparent difference in the use of the clitics as regards the marking of Conditionals, with the possible exception of irrealis Conditionals.
7. Discussion
7.1. New and Contrastive Topic
One purpose of the study was to understand if and how RL and IL interact. From the analysis, it seems that the IS marking often coincide with Episode marking. Given that the definition of Episode relies on the idea that there is a discontinuity of time/location/individual, and that =ta/=nda and =ja(n) mark (although in different ways) a change of Topic, it is not surprising that the clitics mark Episode boundaries. Nevertheless, it seems that this definition cannot be generalised, due to different factors, such as =ta/=nda having a role other than Topic marking, namely finality. In addition to this, the clearest example of change of Episode was reported as being dummy verb head marking, which was not found to be ever marked with the clitic =ja(n). Furthermore, the grammaticalised form tsomba, which is formed by the dummy verb tsun, ‘do’, and the same subject cosubordinator clitic, =pa, seems to mark a whole new Episode in A’ingae, despite not containing any IS clitic. Consider (57).
(57) a. Fûe-khu ûfa=ma fi'thi=mbi=e pasa one-NR.ANG blow=FRT kill=NEG=ADVR pass
‘He shot at one but missed.’ (BC07.013)
b. Tsu=mba tse’i munda ja=si umbu=en ja do=SS then peccary go=DS pursue=ADVR go ‘So, then when the peccary left, he followed them.’ (BC07.014)
On the other hand, it seems clear that neither =ta/nda nor =ja(n) are pure markers of, respectively, New Topic and Contrastive Topic. The analysed data shows agreement with the hypothesis of Fisher (2007, see Section 2): namely, that =ta/=nda is mainly used to introduce or reintroduce a Topic. Fisher argues that this clitic is rather a discontinuity marker, however, this seems to conflict with the
34
IS use it has. I propose that =ta/=nda should be defined as an IS marker and, specifically, a marker that (re)introduce something (new) in the discourse. The discontinuity noticed by Fisher directly arises from this IS use: that is, discontinuity is likely to be a side effect of something new being re(introduced) by the speaker, so that a continuity interruption with the preceding discourse is inevitable. This side effect is especially evident in some uses other than Topic marking. First of all, this appears immediately clear when =ta/=nda is attached to the head in THL, since I presented in Section 6.1. examples that show that in these cases the performed action is discontinuous with the preceding one. In other words, this clitic not only causes discontinuity of participant/location/time, but also a discontinuity of the entire action, as shown in THL and in its generalization, i.e., dummy head THL. This discontinuity effect also explains why this clitic marks Conditionals and final clauses. Fisher (2007) proposes that =ta/=nda marks Conditional clauses because this construction could be seen as a discontinuity of realis mood. Similarly, it seems possible to account for what is expressed in a purpose clause as also unreal(ised) given that the action is intended, yet remains to be achieved. On the other hand, the data shows that it is not as easy to give a definition of =ja(n). Even though some examples do indeed look like Contrastive Topic, for plenty of others this does not seem to be the case. As already mentioned in Section 2, Fisher (2007) suggests that this clitic is used to reactivate a Topic. In other words, =ja(n) can further specify a Topic that is already in the Common Ground and that is usually the subject of contrast or emphasis for the speaker. This would explain why occasionally this clitic seems to mark Focus (44) and why sometimes it can be confused with New Topic (45). That is to say, it seems that sometimes the definition of (re)introduction of a Topic and reactivation of a Topic may overlap, even though the latter seems to refer to the CG whereas the former does not, with the results that =ja(n) is used to introduce what looks like a New Topic. Concerning THL, it is not easy to draw a conclusion, considering that only 3 examples were found. In 2 of them the reading is Contrastive, and in one of them the reading is not clear. It is likely to be the case that the Contrastive interpretation holds for THL as well, but, due to the lack of data related to this use, it is not possible to draw such a conclusion.
Finally, it is interesting to notice a factor in the data that may potentially change the definition of Episode originally given by Hengeveld and Mackenzie (2008: 157) and already reported in 4.3. Indeed, some examples were found to introduce, reintroduce or contrast an entity that did not appear to be definable neither as time, nor as location, nor as individual, but as what should be referred to as participant. Consider (58).
35
(58) Kuse=ta kuankuan a'ta=ve da. night=NEW trickster day=ACC2 become ‘The night is the day of the trickster.’ (BC07.082)
In (58) =ta does not mark a change of time, but rather a change of time participant. Hence, the Episode should be (and has been considered in the present research) not as continuity of time, location and individual, but as continuity of time, location and participant.
7.2. Conditionals
No evidence was found for A’ingae supporting Ebert et al. (2014)’s hypothesis that different types of Topics are linked to different types of Conditionals. This does not mean that this cannot be the case, but that the analysis of the data does not give enough evidence to support this theory. There is some confidence only in proposing that irrealis Conditionals are marked with the clitic =ta/=nda since this is the case for 3 examples out of 3. However, again, the amount of data does not allow one to draw a definite conclusion. Nevertheless, there are still some interesting features I will now describe.
First of all, it appears clear that the Topic clitics can indeed mark the Conditional clause. However, it is also evident that this is not the only way this construction is marked. In fact, several examples are marked with two other clitics: namely, with the different subject cosubordinator clitic =si/=ni and with the same subject cosubordinator clitic =pa. These two clitics are usually used in narrative chaining constructions to express whether the subject of the sentence is referentially the same as or different from that of the following clause. However, it is evident from the fact that they occur in Conditionals that their interpretation is rather contextual. That is, the reading of the clause is a conditional one, even though this is clear only from the context and not from the marking.
In addition to this, some examples were found to be marked with an interrogative clitic. This is not at all surprising, and is fairly common among languages. As already mentioned, Haiman (1978) himself noticed not only the relation between Conditionals and Topics in Hua, but also the relation of these two entities with Questions. Consider (59).
(59) a. If John comes, I’ll be happy.
b. John comes? That makes me happy.
For many languages, A’ingae among them, it possible to realise a sentence both like the one in (59a) and like the one in (59b), owing to the similarities shared by the two constructions. However, there
