Toward a Typology of Ranking Elements of Narrative Discourse in Languages and Cultures : A Cross-Linguistic Survey

It has been noted (Perkins, 2009; Zwaan, 1999; Zwaan & Radvansky, 1998) that causality, character, location, and time are the four main aspects of narrative discourse, even if not attended to by listeners or readers in equal ways. For example, character is highly ranked, and the locational/spatial components have often been underestimated for English narratives (see Perkins, 2009, for a review). Relative to the ranking, there is no inherent reason why character needs to be highly ranked, and locational/spatial information is in fact important in English narrative discourse (Perkins, 2009). I instead suggest that there are linguistic and cultural factors in the ranking of these aspects of discourse. Specifically, I suggest that causality is (probably) the highest ranked component, in languages that have a ranking, with the other three elements being linked to causality more or less strongly, depending on linguistic and cultural factors; it is possible that some languages do not rank narrative elements or that some elements are ranked as highly as others. In English, the strongest link is between causality and character. However, this is not universal. In a survey of fifty-eight languages from thirty language families, including an in-depth study of Hobongan, an Austronesian language spoken by approximately two thousand people on the island of Borneo that I am in the process of describing, it is found that there is a great deal of cross-linguistic variation, to the extent that it is possible that each logically possible combination of narrative elements is present in the world’s languages.


Introduction
Studies on people's processing of narrative discourse (Zwaan, 1999;Zwaan & Radvansky, 1998; see also Perkins, 2009 for a review) have identified four main elements of narrative discourse: causality, character, location, and time, broadly defined: characters do not need to be animate, for example, and 'time' information can range over the full variety available to languages, including aspect and tense, as well as duration and sequence.A narrative discourse is a narrative discourse if it contains each of the four elements, even if not as equal components.The psycholinguistic studies that have been completed (Zwaan's aforementioned works; Zwaan & van Oostendorp, 1993;Blanc & Tapiero, 2001) also indicate that readers or listeners do not attend to the various aspects of narrative discourse equally.In studies on English (Zwaan & colleagues) and French (Blanc and Tapiero), subjects tracked causal information most closely by linking causality in narrative to character information.Temporal and spatial information were relatively backgrounded in subjects' processing of narrative discourse, with subjects rarely paying attention to spatial information unless explicitly instructed to do so.Studies in which subjects did not attend to spatial information have provided support for the idea that spatial information is unimportant in English (e.g., David Mark, 2004, p.c.).However, my own work has suggested that, although spatial information may be relatively backgrounded in English, spatial information is important and can be used as a crucial component of narrative by skilled creators of narrative discourse (Perkins, 2009).The relatively close link between causality and character in English has also prompted some authors to suggest 'motivation' as perhaps a fifth component of narrative discourse; however, because the close link between causality and character appears to be a language-specific phenomenon, I use only the four elements that have been noted in prior research on narrative (Longacre, 1968) and in the aforementioned psycholinguistic literature.
Because little psycholinguistic data is available for non-Indo-European languages, in order to analyze narrative discourse in additional languages, it is helpful to understand some of the strategies available in the languages that have been studied (mostly English) in order to compare similar strategies in other language and to identify additional strategies that might not be available in English.According to the aforementioned psycholinguistic literature, people use a variety of cues.One is the level of specificity provided to the character information as linked to causality in English; another is the coherence and consistency for character in the narrative discourses in English (a lack of coherence and consistency in spatial information in the textoids (Graesser et al., 1997), short narrative discourses created specifically for experimental purposes, could have contributed to subjects' difficulty in processing the spatial information in the textoids); another type of evidence is from prescriptive 'rules' for creating narratives that are culturally presumed to be better than others, one of the main ones being that authors should develop characters in order to create narratives that are compelling for readers (Simons, 1996, p.c.).English thus provides a direction to take when beginning to evaluate narrative discourses in other languages, but as might be expected, languages vary in the available strategies for managing information in narrative discourse.However, English is not the only language possibility, so I begin this study with an in-depth analysis of narrative in Hobongan, an Austronesian language spoken by approximately two thousand people on the island of Borneo.I then continue with a survey of some of the literature on narrative discourse in additional languages, with the attempt being to identify additional strategies available in the world's languages for indicating priority of the narrative elements and to categorize narrative discourses by their rankings.
Using these initial data and information from additional languages, I suggest a typology of sociolinguistic elements of discourse and culture, with all of the logical possibilities of relationships among elements of narrative discourse probably being available in the world's languages and cultures.Admittedly, in the comparison conducted, not all of the logical possibilities were found, perhaps for a number of reasons.One reason could be the inadequacy of language descriptions that often do not examine language structures beyond the level of the sentence.Given the importance of narrative discourse to languages, cultures, and individuals within those cultures, the lack of inclusion of narrative discourse in language descriptions represents a major gap in the information available about many of the world's languages and hinders the investigation of narrative phenomena cross-linguistically. Another reason could be that not all possibilities are instantiated after all.Finally, it could be that not finding all of the logical possibilities is an artefact of the necessarily limited materials examined in a study of this size (an opportunity for future research is to expand the field work and examinations of descriptions).
This study presents typological challenges to assumptions about the ways in which narrative discourse is constructed (e.g., Longacre, 1968, whose work was, and remains, the basis for cross-linguistic studies of discourses, including narrative discourses), such as the idea that character is universally primary in narrative discourse, as well as assumptions about the ways in which the information inherent to narrative discourse is managed in the world's languages and in people's cognitive conceptualizations.Although there are patterns, claims for universality of any pattern are premature.This cross-linguistic comparison examines the basic elements of narrative discourse, the sine qua non for building a theory of literariness (literary quality): before it can be determined that a discourse qualifies as literature, it must be determined that it is a discourse and which elements are available in the language that can be used to determine, relative and relevant to cultural and linguistic requirements, the aesthetic value of a discourse as determined by people who are the audience for the narratives.
The approach used in this study of comparison and contrast of the availability and ranking of elements of narrative discourse in multiple languages could facilitate considerations of the relevance of the elements of narrative discourse, elements that are necessarily prior to the building of a generalized theory of literary quality in the world's languages, in addition to providing a starting point for additional cross-linguistic comparisons of the ways in which narrative discourses are structured and evaluated.The study is also prospective in that a case is made for the evaluation and comparison of narrative discourses and for defining the components that make narrative discourses literary in a variety of languages and language families so that a thorough grounding is provided prior to attempting to make generalizations about the patterns that exist.

Theoretical assumptions
In this study, I avoid choosing a specific theory of discourse analysis in order to give as much weight to the data in each language and language description as possible.The choice of a specific theory often limits analyses to data that either confirms or denies a certain approach, either intentionally or unintentionally; for example, the assumption that the height of language existence is the sentence, a remnant of the success of Chomskyan theories in describing English syntax, has led many people who describe languages to limit their analyses to sentence-level phenomena.As a more particular example, various theories of narrative structure have been offered, beyond the character-focused theory noted by Longacre (1968), focusing on causal sequences of events (a correlation of causality-based and time-based focus in this analysis).Two of these are Rimmon-Kenan (2002) in literature, and Labov and Waletzky (1967) in linguistics.As noted previously, the cross-linguistic comparison suggests that character-focused narratives are not universal; taking an approach that assumes a certain focus or interaction with causality limits the nature of what can be examined.Because of the foundational nature of this study, theoretical considerations must be kept minimal.However, because it is impossible to proceed without some underlying assumptions, I state them here; these stated assumptions have the status of an approach, rather than a theory.Falsifiability, if desired, can be achieved as new data from the world's languages become available.
The first and primary assumption is that all levels of language, phonetics through sociolinguistics, both content and organization, interact to create units of discourse.Discourses are structured and rule-based, not random or chaotic.This study is therefore syntactic discourse analysis, which focuses on the structures in languages that create and emerge from discourse, rather than critical discourse analysis, which focuses on the ways in which ideology affects uses and users of language.The interactions of socio-cultural phenomena and ideology with language, while intriguing, are therefore beyond the scope of this study.
The second assumption follows from and relates to the first: without analyzing all levels of language available in discourse, a description of what is available in discourses, both with regard to content and structures, is not possible.
The third assumption is that written language represents at least possibly spoken language; both written and oral media rely on human cognition and linguistic patterns for their structures.Analysis and descriptions of discourses and their structures can therefore be conducted on any discourse, written, oral, or signed.This assumption does not make claims about similarities and differences between written and spoken discourses; the requirements of the media can be met differently without violating the requirements of syntactic discourse analysis or the requirements of the linguistic structures that are specific to any given language.This assumption does prevent the idea that literature might be only what is written.The Hobongan, for example, have ideas about the literary quality of oral instantiations of traditional narrative discourses.
The fourth assumption, inherent to any field work and language analysis, is that data from the language as given by native speakers provide the essential basis for linguistic analysis.Data are data and can be described, analyzed, and reanalyzed without any impact on the facts of the language.Human participation in the process, as a native speaker or as a linguist, should not be conflated with raw data.When a language description does not include discourse analysis, but does include a discourse, analysis of the given discourse and description of discourse structures can be conducted, with considerations for the lack of context that is often necessary in descriptions of a manageable length.Having some language data means that some language data is not available (having all data is impossible), and the selection of the data may be strategic on the author's part, with positive or negative consequences; a discourse makes discourse analysis possible, but it may not be sufficient for adequate analysis and description of discourse structures.The awareness of the necessity and limitations of data is crucial to responsible analysis.
A fifth assumption is that qualitative linguistic description and analysis must precede quantitative investigations.It is necessary to determine what is countable and why and how the countable elements are relevant to the questions being investigated before quantitative investigation is relevant.This is therefore a qualitative study rather than a quantitative study, which aims to address the fundamental questions that arise when comparing discourse elements and structures cross-linguistically.
A sixth assumption, particular to this study and the approach I am pioneering here, is that causality is (probably) ranked highest in any narrative, as a means to provide global coherence (Graesser et al., 1997).The other elements of narrative would then link more or less closely with causality, depending on language-specific requirements.This assumption is open to revision if empirical data require.

Methods
The Hobongan case-study is based on fieldwork conducted in 2012-2015 in Indonesia among the Hobongan.The type of field work conducted is generally known as Community-Based Language Research.It has been described (Czaykowska-Higgins, 2009) as language research conducted on a language or languages, for the language community, with the language community, and by the language community.In other words, the linguist(s) involved are active participants as opposed to external observers (Dimmendaal, 2001), and native speakers are intimately involved in the process of data collection and analysis, and implicitly, there is some benefit to the community from the language research.In the case of the Hobongan, that benefit has yet to be realized but is in progress; language documentation is part of the process toward gaining minority rights under Indonesian law.
The language survey consists of two parts.The first part was based on informal interviews with linguists who are native speakers of the languages or who are engaged in linguistic analysis of a language.Linguists presumably have the meta-awareness necessary to comment relevantly on their languages' discourse structures.However, in cases in which the linguists did not seem to understand the explanation of the structures being examined, written descriptions were consulted when available.The second part consisted of examinations of descriptions of languages as written by professional linguists or people with the necessary linguistic training.The survey is therefore not systematic, relying instead on availability.Future research could resolve this potential weakness.

Terms
As mentioned in the introduction, this analysis and survey rely on four central concepts in narrative discourse: causality, character, space, and time.For the purposes of comparison across languages and analyses, I interpret and use these terms as broadly as possible.Characters need not be human, for example; a 'character' in a narrative can be anyone or anything that has aspects of personhood such as thinking, feeling, interacting (with self or others).'Space' can be any aspect of space, such as motion through space, location, position.'Time' can be any aspect of time, such as sequence, duration, pastness/presentness/futureness, other aspectual categories.'Causality' is perhaps the most restricted because it cannot include conflations of sequence with causality, but it can include any aspect of causality, such as primary, secondary, tertiary causation and can be focused on causes or effects or both.

Hobongan narrative discourse
This study begins with a study of narrative in Hobongan, an Austronesian language spoken by approximately 2000 people on the island of Borneo, in order to show how the main narrative elements of discourse are used in a language that differs typologically from English.Specifically, Hobongan links causality and location most closely, as indicated by a focal particle that is used almost exclusively for locational information (Perkins, 2013), and by nativespeakers' unwillingness to accept as a unified discourse (a discourse that can be accepted as a unit of language) any units of language that do not include explicit locational information.
Because this study was initially inspired by looking extensively at patterns in narrative discourses in Hobongan, I begin with an overview of that case study.The case study reveals variations in one language that call into question English-based assumptions that are inadequate for many non-English and non-Indo-European languages.Hobongan narrative discourse is organized around locations, a pattern that is evidenced throughout the language.All examples of Hobongan are written in the Hobongan writing system; example sentences are taken from a narrative description of a field trip to the ancestral burial grounds.I proceed through levels of language from pragmatics to sound system in order to identify the ways in which the Hobongan language structures its narrative information.
Pragmatically, information about location is managed specifically and continuously throughout a narrative, in contrast to the ways in which information about the people involved in the narrative, which is relatively underspecified, in comparison to the spatial information.In these two sentences, the spatial information is specified to the river and the navigational activity, which is in the focused clause.In the second sentence, the information is further specified by designating a certain intersection of the river with one of its tributaries.In the rest of the narrative, the specificity of the spatial information continues.The 'we' in the text is never specified.All of the audience members present at the telling of the narrative were included in the field trip; therefore, it might be suggested that the information does not need to be further specified.However, if it were the case that known information does not need to be specified, the entire narrative would be unnecessary because all of the people in the audience were also on the field trip and therefore knew both the route to the destination and the destination.
Another piece of evidence is the way that one child used to encourage his mother to continue a narrative.Rather than asking "Now what?" or "What is he/she doing," the child asked, repeatedly, "And here what?" (moq mo?).The spatial information was what the child was seeking in order to keep a narrative flowing, not information about any other aspect of the narrative.
Lexically and semantically, Hobongan has a number of discourse markers, such markers being defined as words or particles or phrases that have semantic import across clauses without being anaphoric or pronominal (Perkins, forthcoming).One of the markers is a focal marker, joq, that is used frequently in narrative discourses and almost always refers to locational information.I have found three possible exceptions, in which joq appears to be focusing on a character, but in each of those exceptions, it is also possible that joq is highlighting locational aspects within the narrative, such as reminding listeners to pay attention to where a given character is located or focusing on a character's position or a specific part of the character.Joq is also used in an idiomatic expression that tells someone not to tell a story because the events that are part of the story are considered too bad to tell: joq aiq kisa, in which the focal particle is used to indicate that the story is located; in other words, there can be no story without a location, according to the idiom.
In the narrative already mentioned, joq is used twenty-six times in twenty-two sentences, each time to guide focus to the spatial information provided, in the case below on the known information (the cardinal direction: upriver) as opposed to the unknown information indicated by being a little bit lost.With regard to the sound system, the focal particle joq draws attention to the clause not only by acting as a focal particle but by its uniqueness in the lexicon.
In a lexicon of over 6200 entries, the /ʤ/ sound occurs word-initially in only twenty-five entries.Rarity or uniqueness is one way in which the human attentional system is focused (Marrone et al., 2002;Hyönä, 1995;Tepin & Dark, 1992;Berlyne, 1958).The sound system of Hobongan therefore allows the focal particle to reinforce the semantic import of the particle, thus affording two ways, at two levels in the language (semantic and phonetic) for joq to direct attention toward the locational information in a discourse.In a primarily oral culture, as Hobongan currently is, the sounds used in any discourse can be a strong contributor to how an audience understands a discourse.Each major aspect of Hobongan, and the overall patterns of the language, work together to reinforce the primacy of locational informationn in the narrative discourse of the language.
Hobongan narrative suggests a number of ways that a language might indicate priority for one or another of the narrative elements, in addition to those already noted for English.In Hobongan, focal particles can be especially useful because they designate certain types of information as more salient in a narrative than others.All levels of language, however, can and do include strategies for prioritizing

Language survey
The language survey was conducted on fifty-eight languages from thirty language families (includes Hobongan).Information on defining languages and language families is taken from Ethnologue (2016), unless Ethnologue did not contain the information, in which case, Glottolog (2016) was used.This portion of the study is a meta-analysis of the data available, usually from linguistic descriptions of languages.No systematic approach was used in the selection of the languages, except in attempting to access as many different language families as possible; the descriptions were those available from the academic library of the university where I was affiliated at the time.Adding descriptions to the study is an area for future research.
As noted, the language survey is comprised of two parts, of information acquired from native speakers of and linguists specializing in some languages, and of information acquired from an examination of written descriptions of languages.Because most written descriptions do not include discourse analysis, many languages that were examined could not be included here.Also, because those descriptions that do include discourse do not approach discourse analysis from a standard perspective (the lack of an agreed-upon approach to discourse analysis remains one reason for the ongoing underestimation for the need for discourse analysis in analyses of language (e.g.Ruiz, 2009) and therefore make methodological consistency difficult across languages and scholars, I have analyzed the data provided in the language descriptions using the theoretical assumptions as noted previously.The languages' ranking of the four narrative elements, causality, time, character, space, will be described along with the main evidence for the ranking, evidence to include the factors noted previously (focus markers, elements that are obligatory in narrative, rhetorical strategies, focal strategies at any level of language) as well as any factors that are language-specific and identified in the descriptions.
The languages are organized according to language family, in alphabetical order, in order to facilitate cross-linguistic comparison.Where a forward slash appears between aspects of narrative, the two aspects cannot be ranked given the available information (for example, space/time).

Afro-Asiatic
Sumerian is a dead language that was spoken in southern Mesopotamia.The information on Sumerian is unique in that is came from an annotated translation of the Gilgamesh Epic (Gardner & Maier, 1984).The translators included some insightful linguistic analysis in which they noted that narrative units are organized by time consistently in Sumerian (233) but that the translation does not always follow that pattern because the target language of English requires more variety in transitions but consistency in character.

Algic
Nishnaabemwin, also known as Eastern Ojibwe, is an Algic language.Based on information in Valentine's (2001) description of the language, it appears that Nishnaabemwin ranks the four aspects of narrative discourse as causality, time, character, space.There is a focal particle ([mi:]) that directs attention to temporal sequence in narrative (151).

Athabaskan
Navajo is a southern Athabaskan language.According to Midgetted (1995), the "overall emphasis" of a narrative "is on the simple sequence of events" (86); in addition, however, the "story is rooted in the spatial realities" (83).Given this information and the texts provided, it seems that the ranking is probably causality, time, character/place because character and place tend to be collocated, with certain characters being used in certain places, and certain places being specific to certain characters.The places and character occur within the "overall sequence" of events, which is why I suggest the current ranking.Character appears to be "rooted in" spatial realities, which might explain why Midgetted suggested that the story as a whole was rooted in spatial realities, but characters and places cannot exist in narrative without a sequence, which is why the ranking is suggested to be what it is.

Australian
Dyirbal is an endangered language spoken in Queensland, Australia.Dyirbal was famously described by Dixon (1972), but the discourse material is minimal.Dixon did note that word order within sentences is constrained by discourselevel considerations (77, 149); however, Dixon did not include a description of those constraints.Nevertheless, the evidence provided in the description and some ancillary information suggest that the ranking of narrative elements is causality, space, time/character.Dyirbal speakers "pay great attention to shape, location, and direction" (30), and time is largely an extension of spaces.Characters can be topicalized, but locations on characters' bodies are often the focus, rather than the character as a whole (72).
Gunwinggu, which is also called Kunwinjku, Bininj Gunwok, and Mayali, is a language spoken in Australia.According to the description by Oates (1964), actions are routinely attributed to characters, and spatial information is optional and not often developed, which suggest the ranking of causality, character, space/time.

Austronesian
Central Bontok is a language spoken in the Philippines; it is also called Finallig.
According to a description by Reid (1970), ranking of narrative elements is causality, character, time, and place; however, the entire description makes Central Bontoc look suspiciously similar to English, which makes it difficult to determine whether the discourse analysis is true to the language or is linguistically inadequate in ways that are found throughout the description.
Daqan is a language spoken on the island of Borneo.According to my investigations during 2014-2015, the discourse elements are ranked causality, time (duration and sequence), character, space.Durations are given as framing elements of narrative discourses, and within those framing elements, sequences provide the narrative structure and drive the narrative causality.Without the duration and sequence information, narrative discourses cease to function as units of language and lose coherence and cohesion.
Embalo is a language spoken on the island of Borneo.Preliminary results from my field work suggest that the elements are ranked causality, character, time/place.Embalo language and culture prioritize characters, with people making things happen and each individual being held accountable for the quality of her/his contributions, both in concrete and abstract domains.Although all of the communities that I have investigated have ideas about what is better or worse in community storytelling and public discourse, it is only in the Embalo community and among Embalo individuals that people think for noticeable amounts of time before deciding how best to contribute to the interaction.Inadequate contributions, even in speech, are dismissed.Paucity of information prevents determining whether there is a ranking between time/place and if so, what that ranking is.
Ponapean is also called Pohnpeian after the island of Pohnpei in the Caroline Islands, where it is spoken.According to a description by Rehg (1981), the elements of narrative discourse are ranked as causality, character, time, place.
A specified character who is tracked throughout the narrative (in contrast to Hobongan, which can leave character underspecified) is required for narrative discourse, which indicates its primacy in the structures of discourse, and place is optional and requires an additional marker when it does appear, which suggests that its use in narrative discourse is non-canonical.
Tagalog is an Austronesian language spoken throughout the Philippines.Based on information in Schachter and Otanes (1972) on focal structures and discourses, it appears that Tagalog ranks the narrative elements as causality, character, space, time.Focus is morphologically marked on verbal elements and relates the action of the verb to the focused argument of that verb, usually a character (69-71).However, space can also be brought into focus, and when that occurs, direction is the characteristic of spatial information that is focused.Temporal information is marked in a complex system but is rarely if ever focused; the description did not contain any example of temporal information being focused.
Taman is an Austronesian language spoken in Borneo.Based on preliminary evidence from my field research, Taman appears to rank the elements as causality, character, space/time.People make things happen in the community; the group-oriented social structure is not used to background individual contributions, as is the case in Hobongan.The Taman interact closely with majority Indonesian language and culture, which has been extensively influenced by western norms and ideals, so some of the emphasis on personal responsibility could be a result of the interactions with the majority.There is currently inadequate evidence to determine whether there is a ranking between space/time, and if so, what that might be.

Eskimo-Aleut
Aleut, also called Unangam, is a critically endangered language spoken in Alaska.According to Bergslund's description (1997), it appears that the ranking is causality, character, space, time.However, this language is particularly difficult to analyze for discourse patterns because some forms and structures have been imported into the language during translation of religious texts, as well as from Russian.In the description, no interlinear narratives are provided.Characters make things happen and are repeated as agent-as-anaphor, similarly to Russian and Spanish (see below), but sentence-finally, which is a rhetorically powerful position in the language (128).Time is grammatically imprecise, and space is typically more precisely described than time.The ranking of elements in narrative discourse is therefore probably causality, character, space, time.

Eyak-Athabaskan
Navajo is a southern Athabaskan language.According to Midgetted (1995), the "overall emphasis" of a narrative "is on the simple sequence of events" (86); in addition, however, the "story is rooted in the spatial realities" (83).Given this information and the texts provided, it seems that the ranking is probably causality, time, character/place because character and place tend to be collocated, with certain characters being used in certain places, and certain places being specific to certain characters.The places and characters occur within the "overall sequence" of events, which is why I suggest the current ranking.Character appears to be "rooted in" spatial realities, which might explain why Midgetted suggested that the story as a whole was rooted in spatial realities, but characters and places cannot exist in narrative without a sequence, which is why the ranking is suggested to be what it is.

Indo-European
English can be categorized as causality, character, time, space.English speakers typically do not recognize discourses as units of language unless there is a character, stated or implied.Evidence for this occurs throughout the language and throughout analyses of language.In literary analyses, the level of literariness is often associated with the level of character development (Simons, 1996, p.c.), with better character development being more literary and less adequate character development being less literary, or non-literary.Time is a close second to character, as evidenced by the common conflation of time with causality.It is possible that older forms of English were ranked with time above character, but that remains an investigation for future study.Space can be almost ignored, as evidenced by the consistent underestimation of the possibilities for space in narrative discourse.Indeed, some have claimed that English is impoverished with regard to strategies to describe spaces (Mark, 2004, p.c.).However, I have noted that spatial information is both describable in English and important to literary narrative discourses, and that lack of spatial information, and a lack of coherent spatial information, can be disastrous in narrative discourses (Perkins, 2009).
Ancient Greek is a dead language that has been extensively studied by many scholars.According to de Kreij (2013, p.c.), Greek organizes discourse elements as causality, character, space/time.Modern Greek has maintained this pattern, according to Antoniou (2014, p.c.).
According to Petrova (2012, p.c.) and Sandler (2013, p.c.), Russian ranks the discourse elements as causality, character, time, and space.The evidence of repeating agents in subject position in narratives as in "the father, he…" suggests that this is a rhetorical device for drawing attention to characters (repetition is a major strategy for guiding attention).Evidence for the ranking of time over space is weaker, but both native speakers/linguists believe that this is the correct order.
The Spanish pattern is similar to the Russian pattern: causality, character, time, and space, with similar rhetorical and syntactic evidence (Smith, 2012, p.c.).

Isolates
Abun, also known as Yimbun, A Nden, Manif, and Karon, is a language spoken in Papua New Guinea.The description by Berry and Berry (1999) suggests that the ranking of narrative elements could be causality, space, character, time.There is a focal particle that is used primarily for spatial information; other than that, the evidence for the hypothesized ranking is weak.
Waorani, also known as Huaorani, Sabela, Wao, Huao, Auishiri, Aushiri, Ssabela, Auca, and Auka, is spoken in Ecuador and possibly Peru.A collection of essays on discourse structures edited by Pike and Saint (1988) includes details of narrative structures such as paragraph breaks and genre distinctions; however, narrative elements are treated in ways that suggest that English-type assumptions guided the analyses.Characters appear to be important and can be focused on and tracked in narratives, yet spatial information can be similarly managed, and little material exists to clarify the authors' focus on character when other phenomena appear to be similarly important in the narratives.However, given the information available, it is possible that Waorani ranks narrative elements as causality, character/space, time.It is also possible that Waorani is a language in which narrative elements are not ranked.
Basque is an isolate spoken in Europe, which makes analysis of its structures difficult because it has millennia of influence from other languages.However, given the information currently available (De Rijk, 2008), it appears to rank the narrative elements as causality, time, character, location.The main evidence for the primacy of time is that Basque has morphological ways to indicate quantities that increase with time, including characters ( 718), but it also appears that characters make things happen in narrative, which might be evidence for a historical shift in process, either from character to time or from time to character.
It is also possible that there is no ranking of narrative elements in Basque; it was one of the most balanced languages examined, 'balanced' meaning that the elements are treated fairly equally with regard to detail, morphology, organization and focus in narrative, and other ways of determining ranking that are relevant in other languages.
Timucua is an extinct language that was spoken in the southeastern United States of America.The language description by Granberry (1993) contains an extensive analysis of the temporal distinctions that are possible in the language that appear to be important at the discourse level; this information suggests a ranking of narrative elements of causality, time (process), character, space.However, Granberry claims that "Time is distinctly not of the essence in Timucua" (89), despite the importance of certain aspects of temporal information that is not governed by clocks.The importance of that information is why I have chosen to present the above ranking; there are many kinds of temporal information, and the lack of clock-centric time is not a deterrent to the use and importance of other types of temporal information.

Japonic
Japanese is an interesting case because, despite the popularity of English-Japanese contrastive studies (Tomotsugu, p.c., 2016), little work has been done on identifying the ranking of elements in narrative discourse.According to Tomotsugu, causality is ranked first because it operates at multiple levels.
Japanese ranks spatial information above information about character, with temporal information being ranked last.In Japanese narratives, locations provide justification for what characters can do, with characters often being identified with and by the locations in which they operate, rather than with personal pronouns.

Mayan
Itzaj (also written Itza' and Itza) Maya is a language spoken in Guatemala.The language was competently described by Hoffling (2000).Based on that information, it is probable that the discourse elements are ranked causality, time/character, space.The language has focus markers that affix to verbs that focus temporal information but that can also focus character in some contexts.
Jacaltec, also written Jakalteko and also known as Popti', is a language spoken in southern Mexico and Guatemala.The description by Craig (1977) is primarily about syntax, but it does include texts that were helpfully transliterated.Based on those texts, the ranking of narrative elements is probably causality, character, time, space, based on the ways in which those aspects of narrative information are used in the sentences.
Mam is a language spoken in Guatemala and southern Mexico.The ranking of narrative elements is probably causality, time/character, space, according to information provided by England (1983).Focal elements are usually personal possession affixes that mark information about temporal order.Character and time are therefore closely linked and may be equally important in Mam discourse.

Niger-Congo
Engenni is a language of the Niger-Congo family, spoken in Nigeria.The language organizes narrative elements as causality, time, character, space.Thomas (1978) notes that the focal marker /ni/ is used sometimes as an aspect marker but has more syntactic flexibility than other aspect markers, which suggests that it is in a different syntactic, and perhaps semantic, category.It invariably marks and focuses verb-with-aspect, when it occurs (74), which suggests that the temporal information that is prioritized in narratives is aspectual.There are also a discourse particles that provides linking among sentences in discourse in a certain order, [ka], thus providing aspect to the discourse as a whole (20-21).
Si-Luyana, also known as Luyana, Luyaana, Luyi, Louyi, Lui, and Rouyi, is a language spoken in Botswana and Zambia.The grammar by Givón ( 1970) is a preliminary study and does not include adequate data to hypothesize about the ranking of narrative elements.

Nilo-Saharan
Anywa, also called Anuak, is a language spoken in western Ethiopia.Reh's description (1996) suggests that the ranking in Anywa narratives is causality, character, time, space.The evidence is not as complete as a discourse analyst might prefer.Characters clearly are most closely associated with causality, there are somewhat few options for time information, and spatial information is optional and oblique.

Salish
Bella Coola, also called Nuxálk, is a language spoken in British Columbia, Canada.The evidence from a description by Nater (1984) suggests an ordering, but more data is required.Aspect is important in Bella Coola and is invariably marked; characters are pragmatically assumed, often based on the semantics of the verbs.The ranking is therefore dependent on whether the language prioritizes stated information or pragmatically implied information.However, given the information that is available, the ranking of narrative elements can be understood as causality, time (process)/character, space.

Sepik
Manambu is a language spoken in Papua New Guinea.Its narrative elements are probably ranked causality, time, character, place.Focal particles are typically used for time information (Aikhenvald, 2008, 204-206), although characters can be marked.Marking characters is a rare phenomenon.Spatial information is usually provided with respect to something else, usually a character, which suggests that the character outranks the spatial information.

Sino-Tibetan
Mandarin is a language spoken by over a billion people in China and around the world.Although it is a major language, it is difficult to find relevant sources on discourse.Li and Thompson's (1981) description includes some relevant information, however.Discourse-level focus tends to emphasize the contributions of an interlocutor (150-154), which suggests causality, character, space/time as a possible ranking of narrative discourse elements.
Classical Tibetan is a language that resulted from attempts to standardize the language of translations of religious texts, usually translated from Sanskrit.As such, the discourse could be influenced by the source language and the target language and the standardization in complex ways.The description by Beyer (1992) hints at a possible ordering of narrative elements: causality, character, space, time.Characters are typically marked in multiple ways (particles, position, semantics: 197), which provides rhetorical reinforcement of attention for information about characters.Spatial information is typically presented with regard to characters (206-207), and adverbs formed from spatial words are used for time (248-9).The language also contains topicalizers that can be used for any element of discourse (275-279), however, which makes determining a clear ranking difficult.

Tai-Kadai
Thai is a language spoken in Thailand.According to Smyth (2002), Thai probably ranks narrative elements as causality, character, time, space.Thai has a rhetorical strategy of repeating a character both nominally and pronominally (the character, she…); sequences of time are fairly precise, unlike space, which is generic and mostly given via closed-class morphological markers.

Totozoquean (Glottolog)
Totonac is a language spoken in Mexico.Its narrative elements are probably ranked as causality, time/space, character, although determining what the ranking could be is difficult.Reid, Bishop, Button, and Longacre (1968) note that setting and orientation components of spatial information provide relationships of participants in narratives, who then do not need to be mentioned because they are obvious from their given relationships (29).However, spatial information can be optional (76), and what defines a unit of event-ness in narrative is a shift in time (77)(78)(79)(80)(81), which leaves open the possibility that perhaps Totonac lacks a ranking of narrative elements.

Tsimshian
Coastal Tsimshian, also called Sm'algyax, is an endangered language spoken in coastal Alaska and British Columbia.It does not appear to be related to the Eskimo-Aleut languages or to other North American languages, unless the Penutian language family can be demonstrated.Coastal Tsimshian ranks narrative elements as causality, character, space, time.Character relates closely with causality both in discourse and in syntax.Spatial relationships are grammatically abstract but more necessarily specified than temporal information (Dunn, 1995, 41-45), as evidenced by the requirement to quantify different types of spatial data differently.

Uto-Aztecan
Comanche is a language that is spoken primarily in Oklahoma in the United States of America.Its narrative elements are probably ranked causality, character/time, space.The description by Charney (1993) suggests that although time/process could outrank character, characters are typically more specified than might be expected with a more narratologically important time/process.Spatial information is always related to process and character, with a speaker's perspective determining deictic reference.
Nahuatl is a language spoken in central Mexico.According to a description by Sullivan (1988), the narrative elements are probably ranked character, space/time.Characters are given structural detail throughout the language, and space/time are handled relatively equally in the language.
Sonora Yaqui is a language spoken in Mexico and in the southwestern United States.The narrative elements are probably ranked causality, character, space, time.Characters are typically focused (Dedrick & Casad, 1999, 44), and units of discourse are units because a character is consistent.Character transitions are marked, and time information can be used to indicate shifts, as well.Spatial information is typically linked to a character.

Conclusions
The preceding notes on the narrative elements in the languages surveyed revealed some patterns in languages and patterns and possibilities in finding information about narrative based on language surveys, and I note those patterns here.

Language Patterns
Although it was assumed for the purposes of the survey that narrative elements probably have a ranking in each language, it is not an essential commitment.Many of the languages, such as Nahuatl and Mam) did not appear to have clear rankings of two or more of the elements, in which case it is possible that there is no ranking of those elements.It is also possible that in some languages, such as Basque and Waorani, there is no ranking of elements, just different ways of handling the elements based on stylistic, rhetorical, or literary constraints and preferences.
As suggested previously, it is possible that all logical possibilities for rankings of the narrative elements, including a lack of ranking, could occur in the world's languages.However, this study did not identify all of the possibilities; notably lacking is any pattern in which causality is ranked below other narrative elements or any pattern in which character is the least crucial element of narrative.It is possible that these patterns could be cross-linguistic patterns, but in such a small sample of the languages of the world, it is premature to commit to claims of consistent patterning.The following chart summarizes the patterns that have and have not been identified in this study.

Implications
This study examined fifty-nine languages from thirty language families (counting the isolates as each being its own family); it is a beginning of an examination and comparison of narrative structures in the world's languages, but as the languages included here are less than one percent of the languages currently spoken in the world and represent approximately twenty percent of the language families (Ethnologue, 2016), the study is far from complete, and future research is needed in order to have a more complete understanding of what is possible in the world's languages and what patterns might exist.
Unfortunately, many language descriptions are incomplete, which makes the task of developing a more complete understanding of patterns and possibilities in narratives difficult.Because of a common focus on sound systems, morphology, and syntax, and the analytical methods that have become widely accepted for those aspects of languages, many descriptions of many languages do not include any analysis of discourse, and sometimes no samples of discourse, and in cases in which discourses and analyses of discourses are included, the methods used often do not suit the data available or the languages being examined.Literary linguistics is thus crippled at the outset as a comparative linguistic science because the narratives of the world's languages have traditionally been neglected in linguistic descriptions.Given how many of the world's languages are in danger of being lost in the relatively near future, it is deeply unfortunate that so little attention has been paid to units of language that are larger than the sentence.
However, even having information available regarding narrative elements and structures is not sufficient to establish the literary quality of any given narrative.Nevertheless, it is necessary to examine the narrative elements and structures in order to establish what any given language requires and allows in its narratives in order to determine a direction in which to begin to examine what various languages require for discourse units and for literary quality.
Indeed, given the variety of ways in which narratives are managed in the world's languages as evidenced by this preliminary summary, it is likely that there is a parallel variety of ways in which languages and speakers of languages evaluate the literariness of narratives.Without the comparison of literary ideas within and across languages and speakers, scholars potentially limit themselves to the ideas that seem to be the most relevant to them, such as the primacy of character, thus ignoring ideas that could be important to other speakers of other languages.This study provides a necessary prologue to the study of literary linguistics generally and of literature cross-linguistically more specifically, but the analysis of what individual languages require for narratives to be literary remains to be undertaken.It is hoped that this study inspires both broader examinations of cross-linguistic patterns as well as more thorough examinations of intra-language narrative phenomena, including literary phenomena.
up the Hobongan to where the Hoborit joins it.' The strictly syntactic, sentence-level evidence for the focus on location in narrative discourse is perhaps the weakest component of the evidence.Hobongan is SVO as is English, yet the languages have very different discourse patterns.'Joq' focuses the clause in which it is located, sometimes distinguishing main clauses from subordinate clauses.It most often focuses clauses in which spatial information is given, particularly new spatial information.