Marking focus in Chatino

WORD

ISSN: 0043-7956 (Print) 2373-5112 (Online) Journal homepage: http://www.tandfonline.com/loi/rwrd20

Marking focus in Chatino Troi Carleton & Rachelle Waksler To cite this article: Troi Carleton & Rachelle Waksler (2002) Marking focus in Chatino, WORD, 53:2, 157-171, DOI: 10.1080/00437956.2002.11432525 To link to this article: http://dx.doi.org/10.1080/00437956.2002.11432525

Published online: 15 May 2015.

Submit your article to this journal

Article views: 26

View related articles

Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=rwrd20 Download by: [172.245.151.126]

Date: 20 February 2017, At: 17:12

TROICARLETONandRACHELLEWAKSLER 1 - - - - - - - - - -

Marking focus in Chatino Abstract. This paper presents and analyzes new data from the officially endangered language Chatino. We investigate the behavior of a Chatino preposition, jj?j, which is sometimes found preceding direct objects. Its patterning, with respect to when it will and when it will not precede a direct object, has so far eluded characterization. Using both natural discourse and elicited data, we examine and reject hypotheses ofjj?j's occurrence based on various properties of the direct object: animacy, specificity, thematic role, 3 ~ 2 advancement (i.e., the marked direct objects are underlyingly indirect objects). We propose an analysis in whichjj?j marks the discourse focus. We argue that the Chatino data support Lambrecht's ( 1994) position that focus is nonidentical with "new information", and that the Chatino data necessitate an expansion of the definition of focus beyond the propositional level. Our analysis incorporates a disambiguating function of jj?j to separate the subject and object when they are potentially parsed as a single NP.

1. Introduction. Chatino is a Zapotecan language spoken in the mountains of Oaxaca, Mexico by approximately 28,000 speakers. Of Chatino's three major dialects, Yaitepec, Tataltepec and Zenzontepec, Zenzontepec Chatino (representing 8,000 of the 28,000 speakers) is the most remote and underrepresented in the literature. Formal documentation of this dialect is extremely limited. 2 (So far, the only linguistic fieldwork and analysis done on Zenzontepec Chatino has been conducted by the authors of this paper. )3 Here, we concentrate on the patterning of the Chatino preposition jj'lj. This preposition has various functions in Chatino. It is used to mark indirect objects and to mark possession (Carleton and Waksler 2000). Also,jj'lj is sometimes used to mark direct objects. For example, in (1), jj'lj precedes the direct object, in (2) jj'lj does not precede the direct object, and in (3),jj'lj optionally precedes the direct object. (NB: Chatino is a VSO language.) (1)

N-tusu?-04 jj'lj nyat~-~'1 (H5-grab-3S prep person-spec) 'It is grabbing the person'

(2)

Nte-sukwa? Pedro tzukwa? 157

158

WORD, VOLUME 53, NUMBER 2 (AUGUST, 2002)

(H-husk Pedro com) 'Pedro is husking the com' (3)

Nte-sa'l-ya Ui'lj) kiti (H-tear-lP (prep) paper) 'We are tearing paper'

The goal of this paper is to determine whenjj?j is used with direct objects in Chatino. The paper is organized as follows. In section 2, we demonstrate that jj'lj's presence cannot be accounted for by positing subcategorization restrictions on particular verbs. Section 3 presents hypotheses positing jj'lj as an object marker based on the syntax and semantics of the object. Because none of these hypotheses proves tenable, we tum to a discourse analysis ofjj'lj as a focus marker in Section 4. Section 5 exarninesjj'lj's use in discourse, and argues that the definition of focus (Lambrecht 1994) must be expanded in order to provide an account of the Chatino data. In section 6, we demonstrate that another use of jj'lj, to disambiguate NP grammatical relations in a sentence, is accounted for by our analysis. 2. Evidence against sub-categorization. Before attempting an analysis of jj'lj based on any property of the direct object, we need to dispel any doubt thatjj'lj is simply subcategorized for by the verb in the lexicon. Examples (4) and (5) below are taken from the same narrative. In each example, the verb -su 'to cut' takes nchi'lyu 'fruit' as its object. (4)

nte-su-0 nchi'lyu, nte-su-0 nchi'lyu na nt-yot!-na tukwi nchi'lyu nte (H-cut-3S fruit H-cut-3S fruit neg H-know-lP what fruit H-be) 'He cuts fruit, he cuts fruit; we don't know what kind of fruit it is'

(5)

nku-tyejna ntu-su-ka'la na nyat~-~1 jj'lj na nchi'lyu (C-begin H-cut-again det person-spec prep det fruit) 'The man began to cut the fruit again'

Since in both (4) and (5), the same verb is used, but the sentences differ in their use of jj'lj, it is evident from these examples that we cannot account for the distribution ofjj'lj with sub-categorization for the preposition by particular verbs. 3. Hypotheses positingji'Ji as object marker. Because jj'lj is found preceding direct objects, we entertain hypotheses that it is a syntactic or

CARLETON & WAKSLER: FOCUS IN CHATINO

159

semantic property of those objects that is responsible for j{lj marking. In this section, we will examine hypotheses based on the objects' animacy, definiteness, and thematic relations.

3.1.jfli as marker of animate objects. Though many of the examples that use j{lj have animate objects, such as ( 1) above, this hypothesis is inadequate for the optionality in (3) above, where the inanimate NP 'paper' is the object. Also, it would incorrectly predictj{lj's optionality in (6) and (7): (6)

Juan 0-yu?u=nto:-yu (jj?j) Maria (Juan C-have=eye-3S (prep) Maria) 'Juan recognized Maria'

(7)

Nkw-ilo-0 (j{lj) ton~?~ (C-rescue-3S (prep) loc.home) 'He rescued the house'

In (6), the direct object 'Maria' is animate, but the sentence is acceptable withoutj{lj. In (7), the direct object 'the house' is inanimate, but the sentence is acceptable with jj?j. Animacy, then, is not a predictor of j{lj's use with direct objects.

3.2.ji?i as marker of specific objects. The examples in (8) and (9) show a pair of sentences, one withjj?j and one without. The consultant's translations of these sentences (into Spanish) differed by the specificity of the object. (8)

Nt-i:-ya ita (H-like-lP water) 'We like water (in general)'

(9)

Nt-i:-yaj{lj ita (H-like-lP water) 'We like the water'

However, many of the examples with specific objects do not take jj?j. In (10), 'Carmen' is a specific direct object, yet jj?j is not used. Thus, specificity of the direct object does not correctly predict the presence ofjj?j.

160

(10)

WORD, VOLUME 53, NUMBER 2 (AUGUST, 2002) ku-t~?~

Pedro Carmen (P-carry Pedro Carmen) 'Pedro is going to carry Carmen'

3.3.jj'lj as marker of goals. The prepositionjj?j is used as the indirect object preposition for goals, as in (11), where 'milk' is the direct object, andjj'lj marks the indirect object 'the woman'. (11) Nka-yuji?-0 xiti? jj'lj kun~1~ (C-sell-3S milk prep woman) 'He sold milk to the woman' Whenjj'lj marks direct objects, many of those direct object NPs can be analyzed as goals. We consider a hypothesis in whichjj'lj marks the thematic role of goal (or its equivalent in a framework using 3 ~ 2 advancement, where an indirect object in one stratum advances to direct object in a later stratum). Such a hypothesis would account for the jj'lj in sentences like (12), in which the direct object would be identified as a goal: (12) y-a:=na1a=ntQ-Q1 jj'lj Juan. (C-go=see=eye-1S prep Juan) 'I visited Juan.' In (12), the predicate is a complex verb including an incorporated noun. The person/number marker, which attaches at the end of a verb, follows the incorporated element (and provides a morphological cue that the noun 'eye' is a part of the verbal complex). The NP 'Juan' has the thematic role of goal in (12), as there is movement toward him. In a framework with 3 ~ 2 advancement (e.g., Relational Grammar, Arc Pair Grammar), 'Juan' could be an indirect object (i.e., a '3') in a stratum before incorporation, and a direct object (i.e., a '2') after incorporation. However, other objects in examples with jj'lj are less easily construed as goals, as in (13)-(14). (13) Nka-?ni=yaka Pedrojj'lj nta (C-beat=wood Pedro prep bean) 'Pedro beat-with-stick beans' (14) Nka-1ne=tike1 Juanjj'lj ita. (C-make=warm John prep water) 'Juan warmed the water'


161

In (13), the complex verb nka-'lni=yaka, 'beat=wood,' takes the direct object nta, 'bean.' (NB: There is no plural marker on nouns in Chatino.) 'Bean' is thematically the patient in this sentence, since it is directly affected by the action, and there is no movement towards an NP goal. The incorporated element 'wood' in this sentence is an instrumental, so in a multistratal analysis, 'wood' would not be holding the direct object relation on any stratum. Thus, in a multistratal analysis, there would be no reason to posit a 3 ~ 2 advancement for the NP 'bean,' since the grammatical relation of that NP with respect to the verb does not change. In (14), the complex predicate includes an incorporated adjective. Here, the only object in the sentence, ita, 'water,' is more likely to be analyzed as a patient (as it is directly affected by the action of the predicate) than a goal (as there is no movement toward it). Furthermore, there are many examples using jj'lj in which the objects are clearly not goals. (15)-(17) provide some ofthese examples. (15) Nkw-i:=?ya Maria (jj'lj) pi:? (C-aux=carry Maria (prep) baby) 'Maria carried her child' (16) Nk-ya-yujj'lj tzaka tzQ?yta (C-meet-3S prep one Spaniard) 'He met a Spaniard' (17) Nt-i:-yajj'lj liti-ya (H-like-1P prep house-1P) 'We love our house' In (15)-(17), none of the NPs following jj'lj are goals, as there is no movement towards them. The only objects present in (15)-(17) are patients, being directly affected by their predicates. In a multistratal analysis, the NP objects followingjj'lj would not be indirect objects (i.e. '3s') on any stratum, and so there would be no support for an analysis of 3 ~ 2 advancement. Thus, we conclude that syntactic and/or semantic properties of the direct object cannot account for the presence of jj'lj. We tuin to a discourse level account of the pattern.

4.ji'li as focus.

We propose thatjj'lj is a focus marker in Chatino. It is a perceptual signal to the hearer, identifying a location in the discourse model where the speaker wants to bring the center of the hearer's atten-

162


tion. This marker is used as an aid in processing, for the speaker to make his/her interpretation of the model known to the hearer. It is analogous to a film director's use of a zoom shot. The film director, like the speaker, has control of setting the scene and its characters, and may choose to zoom in on one character or element in the tableau. Current research characterizes focus as a pragmatic relation at the level of a single proposition. Lambrecht (1994:213) provides a clear definition of focus at this level. Focus: The semantic component of a pragmatically structured proposition whereby the assertion differs from the presupposition.

This definition can account for the optionality ofjj'lj, since the same NP direct object might be the focus in one proposition, and not the focus in another. But examples like (1), in which all examples in the corpus use jj'lj, and examples like (2), in which none of the examples in the corpus use jj'lj are still unaccounted for. 4.1. Lexical semantics of predicates and focus-marking. The choice of predicate is usually irrelevant to the focused status of its arguments. However, certain predicates, because of their lexical semantic features, will either make it extremely likely that their objects will be focused, or extremely unlikely. In analyzing our Chatino natural discourse data, we have found that when the predicate su'l- 'grab' is used,jj'lj always marks its object (as in (1), repeated below for convenience). (1)

ntu-su'l-0 jj'lj nyat~-~1 (H-grab-3S prep person-spec) 'It is grabbing the person'

A lexical semantics analysis of su'l- entails 'quick movement used to gain control of an object'. So when a speaker chooses to use the predicate su'l-, s/he has made the choice of a predicate whose object is something that must be seized quickly, or it will escape. su'l- does not entail any particular semantics of its object, so the object itself is unknown. This particular predicate, then, makes it extremely likely that its object be focused; we have to pay attention to this object because its presence in the discourse model is suspensefully tentative. Conversely, a predicate whose lexical semantics entails most or all of the semantic features of its object makes it extremely unlikely that its


163

object be placed in focus. The speaker does not need to call the hearer's attention to that object, because its presence is already understood or expected. Predicates like extinguish, which entails the presence of its object, fire, or husk, which entails the presence of its object, corn, or earn, which entails the presence of something of value, (always money in our data), make it unlikely that their objects be placed in focus. And the Chatino examples in (18)-(20) support this hypothesis. (18a) Nkw-i:=suwi? Pedro ki: (C-aux=extinguish Pedro fire) 'Pedro extinguished the fire' (18b) *Nkw-i:=suwi? Pedrojj?j ki: (C-aux-extinguish Pedro prep fire) 'Pedro extinguished the fire' (19a) Nte-sukwa? Pedro tzukwa? (H-husk Pedro corn) 'Pedro husked the corn' ( 19b) *Nte-sukwa? Pedro jj?j tzukwa? (H-husk Pedro prep corn) 'Pedro husked the corn' (20a) Nka-?ne=kana Mariajne (C-make=gain Maria prep money) 'Maria earned money' (20b) *Nka-?ne=kana Maria jj?j jne (C-make=gain Maria prep money) 'Maria earned money' In each of these examples, the set of lexical entailments in the semantics of the predicates indicates the features of the corresponding objects. Thus, it would be very unlikely for the speaker to need to bring the hearer's attention to them. The lexical semantics of the predicates ensures that their objects are already in the hearer's awareness. To sum up so far, the hypothesis thatjj?j marks a focused direct object accounts for each type of example presented in (1)-(3). In examples like ( 1), where jj?j precedes the direct object in all instances in the corpus, the lexical semantics of the predicate sets up an unknown object to

164


be focused. In examples like (2), where jj'lj does not precede the direct object in any instance in the corpus, the lexical semantics of the predicates entail their respective objects, making it very unlikely that those objects be focused. And in sentences like (3), which comprise the bulk of the examples in the corpus, an NP might be focused or it might not, depending on the discourse and the speaker's desired interpretation of it. The optionality of jj?j corresponds to the optionality of any particular NP being placed in focus in the discourse. Thus, all the data presented so far are accounted for by the hypothesis thatjj'lj marks the focus in Chatino. Usually (following Grice's Maxim of Relation), the element the speaker will want the hearer's attention focused on will be the new piece of information which the speaker is bringing into the model. However, just as a film director may zoom in on a character already included in the visual field, so, too, can a speaker place focus on an element or relationship already present in the discourse model. The Chatino data suggest an expansion of the definition of focus to be able to include information already in the discourse model, if the speaker has reason to bring the hearer's attention to that information. The next section will present natural discourse evidence from Chatino for this expansion of focus.

5. Distribution ofjj?j in discourse. 5.1. Previous treatments of focus. Previous definitions of focus which we have found are inadequate in accounting for the Chatino discourse data. Most traditional approaches to focus employ the notion of "new information" (e.g., Vachek 1966; Halliday 1967; Jackendoff 1972; Akmajian 1973; Selkirk 1984). Halliday (1967) proposes that focus is information that is not recoverable from the preceding discourse, while Jackendoff (1972) posits focus as the information in the sentence that is not presupposed. In his discussion of intonation, Bolinger (1954:152) claims that focus is the "greatest concentration of information, that which the hearer would be least likely to infer without being told." It will be shown in the following sections that a definition of focus which entails new or unrecoverable information will not account for our data. Further, while Lambrecht's (1994) definition of focus does notrequire the focused constituent to be new information, he does define focus as a component of a proposition. We argue below that the domain of focus should be expanded to non-arbitrarily determined stretches of discourse.


165

5.2. Focus is not new information. As discussed above, many of the traditional approaches to focus define focus as being new information in a sentence or discourse. The following examples culled from several texts show unequivocally that this definition does not account for Chatino data. In all of the examples of jj?j preceding a direct object in the corpus, the direct object was a definite NP. Examples (21)-(23), (with direct objects underlined), are representative of the corpus. (21) nka-lo-yujj?j na kuchilu-u1 ntu-si?yu-yu yane (C-remove-TOP prep det knife spec P-cut-spec neck) 'He [the man who is topic] took his knife and began to cut his [the eagle's] throat.' (22) ntu-ra=ka?a mastru sukwela jj?j-q? nu kwenta na nti-ka ch-akw~?-~1 jj?j kwiti-i? (H-hit again master school prep-1S rel account neg H-can P-take-1S prep medicine-spec) 'The school master would hit me again because I could not swallow the medicine' (23) ynq?q ku-tze-Q1 jj?j kosa na tzo1Q (hence P-fear-3P prep thin~ ne~ ~ood) 'Hence, they fear these bad things/creatures' Examining the underlined constituents in (21)-(23), all objects followingjj?j are previously mentioned in their respective texts. na kuchilu-uJ 'the knife (specific)' in (21) is preceded by the definite article na and is marked as specific with the suffix on the noun which lengthens and glottalizes the final vowel, -u?. kwiti-iJ 'medicine (specific)' in (22) is marked for specificity, again, with the lengthened and glottalized final vowel-iJ. kosa na tzoJo 'bad things' in (23) is neither marked for specificity nor marked as definite with a preceding definite article; however, this NP appears at the end of a narrative which revolves around kosa na tzoJo and was translated into Spanish by the consultant as esas cosas malas 'these bad things'. Not only is the NP given information, it is the thematic subject of the narrative.

5.3. The focus relation. According to Lambrecht (1994), a focused constituent need not be new information itself, but rather the pragmatic relationship in a proposition which the focused constituent enters into must be new information. Lambrecht's analysis will not account for all

166


of the Chatino data. In some cases, the relationship between the proposition and the focused constituent is new, in others it is not.

5.3.1. New pragmatic relationship for focused NP. In (24), 'the knife' is the object NP. As was established in the preceding section, 'the knife' itself is not new information in this text. (24) nka-lo-yuji?j na kuchilu-u'l ntu-si'lyu-yu yane (C-remove-3S prep det knife-spec H-cut-3S neck) 'He [the man] took his knife and began to cut his [the eagle's] throat.'

The discourse in which (24) appears is a narrative about an eagle and a serpent that terrorize a community in the mountains of Oaxaca by swooping down onto the main road, grabbing people and livestock from it, and taking them up the mountain to devour them. The first half of the narrative involves the community's successful effort to kill the serpent. The second half involves a scheme to kill the eagle. Mention of the knife purchased for this purpose appeared in the clauses preceding the event depicted in (24). While the knife itself is not new information, the relationship between the knife and the proposition nka-lo-yu 'he removed [something]' is new. Lambrecht's (1994) proposal for focus accounts for (24) and similar examples.

5.3.2. Already-established focused NP. While the data above support a characterization of focus as a new relationship between a constituent and its proposition, the data presented in this section do not. Examples (25)-(27) provide evidence to suggest that the definition of focus needs to be expanded beyond the level of a single proposition. The relationship between the focused NP kosa na tzo?o 'bad things' and the proposition [they fear x]' is not new in example (25). (25) yn