Lexical choice in Karo narratives

Escolha lexical em narrativas karo

LEXICAL CHOICE IN KARO NARRATIVES^*

(Escolha Lexical em Narrativas Karo)

Nilson GABAS JÚNIOR

(MCT/Museu Paraense Emílio Goeldi, Belém)

ABSTRACT: This paper looks at two verbal descriptions of the Pear film and characterizes them according to the analysis proposed by Downing (1980) for factors influencing lexical choice. The two descriptions, one short and one long, were told by my Karo consultant, Mário Jorge Arara, after the exhibition of the film. Generally, the present article looks at Downing's assertion that "if the description is to be brief, words of broad referential scope are likely to be chosen (...). If the speaker opts for a more detailed description, more lexemes of narrower referential scope are likely to appear" (1980:90) and sees how this assertion applies to the two narratives. Specifically, it looks at each of the versions of the story and tries to explain the mentions of the referents by either basic or non-basic level categories in terms of cognitive, textual and contextual factors.

KEY-WORDS: Lexical choice; Codability; Brazilian Indian Languages; Karo.

RESUMO: Este artigo procura caracterizar duas descrições do filme 'A Pera' de acordo com a análise proposta por Downing (1980) sobre fatores que influenciam a escolha lexical. As duas descrições, uma curta e outra longa, foram colhidas junto ao informante Mário Jorge Arara, logo após a exibição do filme. De maneira geral, este trabalho investigará a hipótese de Downing de que "if the description is to be brief, words of broad referential scope are likely to be chosen (...). If the speaker opts for a more detailed description, more lexemes of narrower referential scope are likely to appear" (1980:90), verificando como esta hipótese se aplica no caso das duas descrições. Especificamente, este artigo procurará explicar as referências aos participantes do filme em termos de categorias de nível básico e/ou não-básico, tendo como parâmetro fatores cognitivos, textuais e contextuais.

PALAVRAS-CHAVE: Escolha lexical; Codabilidade; Línguas Indígenas Brasileiras; Karo.

0. Introduction

In this paper I look at two versions of the Pear Story, a non-verbal film designed for cross-linguistic comparison of verbalization processes (Chafe 1980), and characterize them according to the analysis proposed by Downing for factors influencing lexical choice in the same type of narratives (Downing 1980). The two versions of the story, one short and another long, were told by one of my Karo consultants, Mário Jorge Arara, immediately after the exhibition of the film, when he came for a visit in Belém in 1995.

From a broader point of view, I will be looking at Downing's assertion that "if the description is to be brief, words of broad referential scope are likely to be chosen (...). If the speaker opts for a more detailed description, more lexemes of narrower referential scope are likely to appear" (p.90) and see how this assertion applies to the two Karo narratives. From a specific point of view, I will look at each of the versions of the story and try to explain the mentions of referents by means of basic level categories as opposed to non-basic level categories in terms of cognitive, textual and contextual factors. A comparison of the analysis of the two versions will follow, in the search for possible differences on the categorization of the referents.

It is worth mentioning that Downing's analysis is also based on the same Pear Story, with a corpus comprised of several narratives as told by English and Japanese speakers, whereas my analysis is based upon just two narratives told by the same speaker, and has, for this reason, limited scope over broad generalizations.

This paper is structured as follows. In the next section I present the general points raised by Downing in her paper, showing, in section 3, the methodological background in which I base my analysis of the Karo narratives. In section 4 I provide the descriptions and explanations for the non-use of basic level lexemes in the Karo narratives, in the light of the explanations given by Downing for the English and Japanese narratives. In sections 5, 6 and 7 I deal with the three factors held responsible for the occurrence vs. non-occurrence of basic level lexemes in the Karo narratives, cognitive, contextual and textual factors, respectively. In section 8 I conclude by providing an explanation for the different number of occurrences of basic level lexemes in the first version as compared to the second version of the story.

Before proceeding, it is worth to provide the reader with the context in which the two versions of the same story were obtained. After exhibiting The Pear Story film to my consultant Mário Jorge, I asked him to tell what he saw in the movie, gathering the first version of the story. Then we continued to talk about the film and he realized that some parts of it could have been left out in the first telling. Then I played the tape to him, we both listened to his telling, and he mentioned that some parts of the movie were not told as accurately as it should be. I then asked him if he could tell me the story again, this time including whatever he thought necessary to include. This was the second and longer version of the same story.

1. Downing's paper

In studying English and Japanese tellings of the Pear film, Downing examines the way nominal expressions are employed to refer to concrete entities by considering the principles of prototypicality and lexical choice.

She found it was possible to recognize that although there was an extensive use of basic lexemes, a great number of non-basic (either subordinate or superordinate) lexemes are also used to refer to concrete entities within the narratives. In order to account for the occurrence of these non-basic lexemes Downing postulates a number of cognitive, stylistic and textual constraints which may cause the speakers to abandon the basic lexemes in favor of the non-basic ones when they categorize a given referent at a given point in the narrative. The specific ways in which she explains the occurrence of these constraints will be provided in section 4 below, before the discussion of the occurrences of non-basic lexemes in the Karo narratives.

Before starting with the analysis of the Karo versions of the Pear Story and compare them with Downing's analysis of the English and Japanese versions, it is necessary to provide the methodological background under which the analysis of the two versions was based.

2. Methodological background

The first point to mention concerns transcription. Each of the versions of the Pear Story was transcribed using the concept of intonation units (IU's), according to the principles of discourse transcription established by DuBois et al. (1992)¹. The first version of the story contains approximately 86 (IU's), and the second version approximately 172 IU's.

Second, in order to look closely at the content of the two stories (for a better understanding of some of the principles established by Downing), they were subdivided according to the episodes (or frames) they describe². Accordingly, in the first version, the consultant described 7 different episodes which were filled with 4 commentaries in different places of the narrative (3 in episode boundaries, and 1 inside an episode). In the second version, the consultant described 10 episodes, 8 of which were filled with commentaries (6 in episode boundaries and 2 inside episodes).

The episodes of each version were labeled (to facilitate their recognition) according to the most prominent protagonist on that specific episode. The different episodes of each version, including their labels and number of IU's employed in their characterization, are given below, as well as the places where the commentaries were made³:

The total number of mentionings of referents in the stories was 55 and 124, respectively, for the first and second versions, including NP's and pronominal markers (either free pronouns or bound prefixes). Not all of these mentionings, nevertheless, were of referential form, i.e. some of them did not actually refer to any entity in the narrative, and thus will not be counted for the establishment of differences of basic level and non-basic level categories.

Non-referential uses of NP's and pronominal affixes are characterized by Chafe (1994) as when "there is no referent at all, either particular or generic" (p.103). Among these uses he considers the occurrence of the pronoun it, as well as negatives (none) and universals (everyone), question words (what, who), event-modifying nouns (telling stories), non-specific nouns (I'll buy a newspaper) and predicate nouns (He's a jerk).

In both versions of the Karo narratives, non-referential uses of NP's and pronominal affixes were not extensively found (probably because of the fact that, more oftenly than in ordinary speech, the speaker was focused on the description of specific scenes with specific referents), but when they actually occurred they corresponded to the event-modifying nouns and non-specific nouns of Chafe's categorization.

The final number of referential uses of NP's and pronominal affixes, then, was of 54 in the first version, and 109 in the second version.

The mentionings of NP's and pronominal markers of actual referential use served to characterize different participants in both stories. As expected, since the first version of the narrative was shorter than the second version, a greater number of participants were mentioned in the second version when compared to the first. They are given as follows (the numbers in parenthesis correspond to the number of times each of the was mentioned):

a) version # 1: man (8); pears (8); boy (10); basket of pears (7); basket (1); bike (4); girl (1); boy's hat (4); stone (1); boys (7); one of the boys (3);

b) version # 2: man (21); pears (14); rooster (2); ground (1); pear (4); basket (1); cloth (1); boy (23); bike (5); man with the goat (3); goat (3); basket of pears (8); one of the pears (1); boy's hand (1); girl (3); boy's hat (4); stone (1); boys (7); one of the boys (4); boys' pathway (2).

3. Analysis of the Karo narratives

Having established the methodological background in which I will base the analysis of the narratives, I will pass now to the correlation between the points raised by Downing and the facts observed in the two Karo narratives. In different subsections below I present her postulations and then proceed with the description and explanation of the facts in Karo.

3.1. Preliminary facts

In her account of narratives in general, Downing establishes that in describing an episode the speaker has to decide how detailed the description would be, the consequences being: brief description = words of broad referential scope are likely to occur; detailed description = more lexemes of narrower referential scope are likely to appear. Her understanding of broad versus narrow referential scope include, for verbs, things like 'to pick' vs. 'to pluck', and for nouns, things like 'tree' vs. 'trunk', 'leaves', etc. She furthermore states that whenever words of narrow scope were used a word of broad scope was used first.

In analyzing the two versions of the Pear Story, it was possible to observe that the shorter version had more words of broad referential scope when compared to the second version⁴. For example, with relation to events, while in the first version a general verb is used for 'to pick', , to refer to the action of picking pears, in the second version the same verb 'to pick' is used followed by an elaboration of this action with the occurrence of the reduplicated form of the ideophone 'to pluck', meaning plural action. Example:

'Then the man was picking his pears. He was picking his pears (by) plucking (then).'

One exception to this general pattern was found regarding the uses of superordinate and basic level category lexemes to refer to the pears. Of the two narratives, the use of superordinate level category lexemes was surprisingly higher in the second (longer and more detailed) telling, as compared to the first one. I will come back to this point in section 8 below, with a more detailed description of the facts and a possible explanation for this exception.

As for the speakers' description of referents in the events, Downing observes that they could refer to these referents as wholes (human beings, goat, bicycle, etc.) or parts of wholes (head, hair, face, eyes, rack, wheels, etc.). When parts were mentioned, she found out that the whole was mentioned first, signaling a preference for wholes over parts⁵.

In the Karo narratives I could observe that Downing's statements seem to hold. It was true, for example, that whenever narrower words (either verbal or nominal) were used either to describe and event or to refer to a participant in the narrative, a correspondent broader word was used first. An example is the occurrence of boy's hand, reproduced below, in the second telling, much after the boy had been introduced into the narrative.

I will pass now to the analysis of how Downing describes and explains differences in the occurrence of basic and non-basic level categories in the English and Japanese narratives.

On conceiving the problem of how the speaker refers to an entity once s/he decides to talk about it, Downing proposes that cognitive, textual and contextual factors might come into play. For methodological and practical reasons, I will describe these three factors in separate sections below, each of them followed by the analysis of the facts presented by Karo.

4. Cognitive factors: codability and basic level categories

Downing characterizes codability as the way(s) in which the mention of a referent varies, depending on the number of lexicalized labels available and the speaker's familiarity with them. According to Downing, the measurement of codability can be complicated by the existence of a number of lexical alternatives, any of which could be used appropriately in referring to a given referent⁶. In cases like these, where the language provides a number of lexical options for referring to a given entity, the recognition of basic level categories (Rosch 1975; Rosch et al. 1976) becomes necessary.

Codability and basic level categories, thus, seemed to relate to each other in interesting ways. For example, Downing mentions that the use of superordinate categories occurred mostly when speakers referred to objects which are of poor basic level codability, either because 1) the speaker did not know the basic level label for the conceptually well-defined category in question, or 2) the specific referent at issue was a non-prototypical member of the category.

In the general sense, nevertheless, Downing observes that in both English and Japanese tellings of the Pear film the vast majority of the mentions of referents used basic level categories. References to people seemed to be an exception, as she points out, since the speakers used a wide range of different terms in referring to the human characters, making it difficult to come up with a clear taxonomy.

As a final point, she states that in most cases where superordinate categories were used they were accompanied by some sort of modifier, in order to narrow down the referent class in question and create and `ad hoc' basic level of categorization.

4.1. Codability and basic level categories in Karo narratives

The codability aspect had a striking effect over several of the participants in the story in which the speaker did not have a way to refer in the Karo language. In all cases, except 'pears', he used the Portuguese terms. This happened, for example, with bicicleta 'bicycle', chapéu 'hat', galinha⁷ 'chicken' and cabrito 'goat', which were also, except by 'bicycle', the only ways to which these participants were referred.

Furthermore, where the speaker did not have a term to refer to a participant and was not acquainted with the Portuguese term, he came up with a term on his own. This was the case of 'pears', which was referred to by means of wirik kanã (lit. 'edible thing') and kanã ('thing').

As for how systematically a participant was referred once the speaker had decided to mention it in the narrative, I found that a more extensive number of different labels were used to refer to humans as opposed to non-human participants. For example, 'man' was referred to as pé ('non-Indian'), a?=maneâ) (lit. '3sg=whole'), yét wirik kanã kóa (lit. 'this owner of the edible things'), and a?= ('3sg.'). Furthermore, non-human participants like 'the boy's hat', 'the stone', as well as all other participants referred to by means of Portuguese terms, were consistently referred to by means of a unique lexeme. The pears, as mentioned above, were referred to by means of two descriptive terms, wirik kanã and kanã.

As was the case for the English and Japanese narratives, it was also not possible to come up with a nice taxonomy for human referents in the Karo narratives.

Among non-human participants, the only ones mentioned with more than one label were the pears (cf. above), the basket of pears (referred to as either wirik kanã cit ¾ lit. 'cover of edible things' ¾ or kanã cit ¾ lit. 'cover of things'), and the bike (referred most of the times as bicicleta ¾ 'bike' ¾ and a few cases as cit ¾ lit. 'cover'). Based on the textual factors to be discussed below, I propose that wirik kanã, wirik kanã cit and bicicleta are the basic level lexemes for the pears, the basket of pears and the bike, respectively. Also, that kanã, kanã cit and cit are the correspondent superordinate level lexemes used to refer to the same participants.

I also observed that some mentions of both basic level and superordinate level category lexemes were in fact accompanied by some sort of modifier (either a possessive or a demonstrative pronoun), but in no case the modifier was used to alter the category level of the lexeme, i.e., in all occurrences of the modifier the purpose was to straighten down the reference of the participant, and not to change its means of categorization. A more detailed explanation of the occurrences of modifiers will be offered in the section where I discuss the textual factors influencing lexical choice.

As a final observation, I would like to point out that although reference to the 'basket (of pears)' was made only once in both versions, their labels do not match. In the first version the word na?no ?a? ('small basket') was used, while in the second version the more appropriate word namón ká? ('large basket') was used instead. This difference in use seems to be due to the fact that the second version was more elaborated as compared to the first.

Let us now consider the question of how basic level lexemes was always (or not always) the contextually neutral level of lexical choice. According to Downing, this is where the contextual and textual factors might come into play.

5. Contextual factors

Three types of contextual factors were considered by Downing as responsible for lexical choice: a) sociological factors, like age, sex, social and educational status, etc.; b) deixis; and c) style. Notwithstanding the postulation of these factors and the possibility of their interference on the choice of referent categorization, she ascertains that none of them played a fundamental role, for one reason or another, on the English and Japanese speakers' choice.

Contextual factors also did not have any effect on the lexical choices made by the speaker in the two Karo versions of the story.

6. Textual factors

As textual factors Downing postulates the occurrence of the following:

a) avoidance of the repeated use of the same term. Even though this is usually true for stylistic purposes, Downing says that, except when referring to humans, the speakers of the stories generally stuck with the terms they originally chose in referring to most entities in the film⁸;

b) position of the word in the narrative. Downing observed that certain lexical strategies seemed to be favored, for example, at the beginning of a narrative, at changes of location or character, or even in descriptive passages. Other times, lexical choice is needed to avoid ambiguity at a particular point in the narrative;

c) introduction of a new referent. It was pointed out that a speaker will often use different labels in introducing referents into the narrative than he will use subsequently once their identity has been specified.

Downing concludes by commenting that in some cases none of these textual factors mentioned above may come into play⁹ and that since "the choices are so highly context-bound and the alternative strategies so numerous that it is only post hoc that the determinants of any particular choice can be dimly perceived" (italics in the original).

6.1. Textual factors in Karo

In analyzing the textual factors responsible for the differences in the level of categorization of the Karo lexemes I will describe the references made to human vs. non-human participants in separate subsections, for methodological purposes.

6.1.1. Human participants

Among human participants in the narratives only a few were referred to by means of different lexemes: the man, the boy and the boys. The girl and the man with the goat were mentioned consistently by means of the same basic level category lexemes.

Avoidance of repetition seemed to be responsible, at least partially, for the use of a pronoun instead of a full noun phrase: the vast majority of the references to the man, the boy and the boys, after their introduction to the narrative, was made by means of a third person pronominal marker (singular or plural).

As for the factor introduction of new referents, all human participants in both versions, including the girl and the man with the goat, were introduced by means of a basic level category, in a special presentative construction. Some examples are:

Version # 1:

Version # 2

The position of the word in the narrative was an important factor to explain the uses of basic level category lexeme plus some sort of modifier (deictic element). In accordance with Downing's postulation, I observed that uses like these seem to have been made in order to help disambiguating the reference of a specific participant in certain problematic points of the narrative. In the following example from the second telling of the story, the speaker uses a basic level category lexeme, boy, plus a demonstrative pronoun, this, to make reference to a participant who could be misunderstood as another. It occurs almost at the end of the telling, in a scene where a boy is given his hat back by another boy. The first boy is one of the main characters of the story, whereas the second boy had just been introduced few IU's before.

In this example, 'this boy', yét nakõm ha, is used to refer to the boy who lost his hat, whereas the third person singular pronominal marker a?=, is used to refer to the boy who gives him the hat back. The description could otherwise have been ambiguous if the speaker was to use two third person pronominal markers (or it would not make sense if he had used boy twice).

6.1.2. Non-human participants

Among non-human referents, the three textual factors pointed out by Downing worked in the following way:

a) avoidance of repetition: Even though no further mentions of non-human referents were made by means of third person pronominal markers, some cases occurred in which a superordinate level category was employed. This can be observed in the following example, where kanã is the superordinate level category lexeme used to refer to the pears:

The occurrence of a superordinate lexeme in this example could be explained in terms of avoidance of repetition since, when describing this passage, the speaker had just mentioned pears a few IU's before, by means of a basic level category lexeme.

b) introduction of new referents: non-human participants were not introduced into the narratives by means of special presentative constructions, as was the case with humans. Notwithstanding this lack of special treatment, it was possible to observe a difference in the way some non-humans were introduced for the first time as opposed to others. All first occurrences of secondary participants used a basic level category lexeme in the grammatical core argument (absolutive) position, with variations occurring only later in the narratives. This was the case, for example, of the pears shown in example (3) above. All other unimportant (or trivial) non-human referents were introduced in oblique case. This was the case, for example, for bicicleta, in both the first and second versions:

c) position. When the description of non-human participants could cause ambiguity due to their position in the narratives, basic level category words plus a modifier (also demonstrative pronouns) were used to refer to them. A good example of this occurrence was found in the first version of the story, in reference to the pears.

In this example, it is possible to assume that, in order to straighten down the reference to the pears that a man was picking up from a tree apart from the pears that a boy was about to take (and run), in a subsequent scene, the speaker refers to the former by means of the basic level category lexeme wirik kanã plus a demonstrative pronoun yét. Notice that if the speaker had decided to use either the basic level category lexeme without the demonstrative pronoun or the third person singular pronoun, reference to the pears could have become vague.

7. Conclusion

I hope to have shown from the above discussion that the use of superordinate and subordinate level category lexemes to refer to the participants in the Karo narratives can roughly be explained by the same cognitive and contextual factors as postulated by Downing in her treatment of the same phenomenon in English and Japanese narratives.

In spite of the high degree of similarities regarding the motivations for lexical choice in the three quite different languages, one important exception seems to occur in the case of Karo.

If we look once again to Downing's postulation that a more extensive use of superordinate level category lexemes is expected to be found in shorter stories as compared to longer and more detailed ones, we observe that her postulation does not hold in reference to one of the most important elements in the film, the pears. Of the eight times in which pears were mentioned in the first telling, three (37.5%) were made by means of the superordinate level category lexeme kanã, whereas of the 14 times pears were mentioned in the second telling, nine of which (65%) were made by means of the same superordinate level category lexeme. This fact seems to contradict Downing's postulation, since the second telling was longer and more detailed than the first one, and still contained more uses of superordinate level category lexemes to refer to the pears, instead of basic level category lexemes.

A possible explanation for this exception, on the other hand, could be achieved if we inspect the content of one of the first comments made by the speaker at the beginning of the second telling, where the speaker expresses overtly his ignorance towards an appropriate label to refer to the pears. I reproduce his entire comment as follows:

Based on the exposed, it is possible to hypothesize that the awareness of the speaker towards the lack of an appropriate label to refer to the pears brought him to use a superordinate instead of a basic level category lexeme. From a strict cognitive point of view, it is plausible to assume that in situations like these the use of a superordinate level category lexeme like `thing' becomes the best option available, since it is vague enough to cover the reference to object for which the appropriate label is missing.

If this explanation is accepted, then the postulations made by Downing seem to apply without failing to the case of the Karo narratives. More cross-linguistic descriptions and analyses of the same Pear film would have to be scrutinized, nevertheless, to verify her points from a broader (maybe universal) perspective.

References

CHAFE, Wallace (1994) Discourse, Consciousness and Time. Chicago: University of Chicago Press.

¾¾¾¾¾ (ed.) (1980) The Pear Stories. Advances in Discourse Processes. vol. III. Norwood: ABLEX Publishing Corporation.

DOWNING, Pamela (1980) Factors influencing lexical choice in narrative. In: Chafe (ed.) 1980.

DUBOIS, John et. al. (1992) Discourse Transcription. Santa Barbara: Santa Barbara Working Papers in Linguistics.

ROSCH, Eleonor H. (1975) Human Categorization. In: N. WARREN (ed.) Advances in Cross-Cultural Psychology. vol. 1. London: Academic Press.

ROSCH, Eleonor H.; C. Mervis; W. Gray; D. Johnson; P. Boyes-Braem

(1976) Basic Objects in Natural Categories. Cognitive Psychology 8:382-439.

^*

Paper presented at the Society for the Study of the Indigenous Languages of the Americas (SSILA) meeting, in Albuquerque, New Mexico, USA, in July of 1995.

¹

DuBois et al. define an Intonation Unit as "a stretch of speech occurring under a single unified intonation contour (pp. 100).

²

Epidodes should be understood here as each one of the (complete) scenes shown/described in the film. Accordingly, the episode number 1 of the first telling, called "The Man", for example, is a scene where a man is pictured standing up in a ladder, picking up pears from a tree.

³

In the 'label' column, the slash bar after the actual label followed by "(comment.)" is used to characterize the commentaries made inside episodes. In the '# of IU's' column, the first number that appears refers to the number of IU's of the entire episode, except the commentary, and the number in parenthesis represents the number of IU's of the commentary.

⁴

As in Downing's analysis, I only included content words in this survey.

⁵

As the reasons for this preference she postulates two factors: a.) the greater salience of wholes over parts: wholes act as wholes, while parts don't; b.) in general, wholes are preferred because they tend to avoid underspecification, redundancy, overspecification and other hazards attendant on the mention of either collections or parts of wholes in the same contexts. As for the motivations for the actual occurrences of parts instead of wholes on her narratives, Downing presents one of the following three possibilities: a.) when the parts belonged to a highly salient, significant, codable whole (human body parts); b.) the part mentioned was a clearly delineated element of the whole and was directly involved in the action being described; c.) the whole was uncodable and could best be described as the sum of its parts (paddleball ball, paddle and string).

⁶

She points out that this was enormously true for the human referents in both English and Japanese narratives.

⁷

In both tellings of the story, the consultant mistakenly refers to the rooster as 'chicken'. This may be due to the fact that he is not 100% fluent in Portuguese.

⁸

Humans, once more, did not follow this general tendency, as she proposes, probably because there exist in the lexicon significantly more terms denoting humans among which it would be possible for the speaker to switch.

⁹

Downing postulates that once the speaker has decided how to structure the narrative, he may find that there is one perfectly appropriate basic level label available for the entity he wishes to refer to and used thoroughly the narrative.

Publication Dates

• Publication in this collection
  02 May 2002
• Date of issue
  2001