of those descriptions did not include material on discourse, units of the language that are larger than the sentence, as Pike (1964) called them. In other words, the descriptions included information about phonology (sound systems), morphology and syntax, but omitted analyses of narratives or other texts. In some cases, texts were included, but more typically, sentences were the largest units described, analyzed or included. Because many kinds of discourses are important in the languages of the world, including oral histories, recitations of familial relationships, religious or ceremonial material and other types of narratives, this oversight seems all the more serious. It is unlikely that the linguists who assembled these descriptions failed to notice the importance of these materials to the people with whom they worked. Indeed, my own experience among the Hobongan suggests that larger-than-sentence discourses are vastly more important to native speakers of a language than are the syntactic, morphological or phonological details.
There could be a variety of phenomena that contribute to this omission. An important aspect might be the traditional inertia in the field of linguistics.1 If linguists are trained in field methods, much of that training is based on the resources already available. If linguists need to train themselves in field methods or update their knowledge about a specific language family, they rely on the materials that are already available. When a linguist settles into a language community to conduct linguistic research and work towards a description, all of the linguistic information, including sociolinguistic information and the discourse types important to that community, is freshly available to the linguist. The task of sorting through it is immediately overwhelming, and field linguists must make decisions about how to proceed in collecting and organizing data. Consulting previous language documentation is a natural strategy: if a language does not have a description, it is likely that closely related languages have been described, and their patterns can provide a way into the analysis. But in the past, phonology, morphology or syntax have been the aspects of language most often included in language descriptions, with no explanation or theory-driven account provided for the customary omission of anything beyond syntax. Thus, phonology, morphology and syntax still constitute the core of most linguistic descriptions today.
This omission means that resources allowing linguists to compare and contrast structures cross-linguistically are predominantly available for the above-mentioned aspects (see Perkins, 2017, for a typological approach to analysis of narrative discourse). The common availability of certain kinds of information (phonology, morphology, syntax) and the corresponding unavailability of other kinds of information (semantics, pragmatics, sociolinguistics, etc.) pragmatically suggest that the phonology, morphology and syntax are more important to descriptions than other kinds of linguistic information are. In my experience with Hobongan, which would likely apply to work with other minority languages, the analyses of sound systems, morphology and syntax are relatively simplified because of the wealth of material on a wide variety of languages, while the analyses of other aspects of language are made more difficult by having to forge a new path. This difficulty, and the additional time and effort required to do the analyses of larger-than-sentence materials, no doubt inhibit some linguists and other field workers from going beyond the background that is available to them.
This privileging of phonology, morphology and syntax may reflect a bias towards what Joseph calls the ‘natural’ rather than the ‘human-oriented’ (see Joseph, this volume). The study of phonology, morphology and syntax has typically emphasized their systematic status more than their connections to societal concerns. In that regard, those areas of linguistics may seem further removed from societal or human contingencies, and thus may seem more ‘natural’, whereas sociolinguistic, discourse and pragmatic concerns may seem more ‘human-derived’. Joseph claims that this same bifurcation provides the foundation that divides descriptivism (natural) from prescriptivism (artificial). Descriptivism aims to identify what is natural about language, without recourse to the human element, which is inherently impossible because languages are imagined, constructed and used by real people in real situations. Without the human element, there is no language to be described. It could be that this same bias for the ‘natural’ ends up privileging the areas of language least connected to the societal concerns of the people who speak the languages.
This traditional omission of any aspect of language beyond syntax is sometimes enshrined in what linguists have come to accept in a published (or publishable) description. Comrie and Smith (1977) published an extensive outline of questions to ask when collecting and analyzing information for language descriptions. This outline includes material on sound systems, syntax and morphology, but not one question, much less a section, on any other domain. The questionnaire has been used as an informational template for several language descriptions. Comrie and Smith provide this questionnaire with the best intentions of helping linguists ensure that their descriptions are as complete as possible and that the material contained within their descriptions is comparable across descriptions, but this approach has a number of difficulties. One of the main difficulties is that nothing beyond phonology, morphology or syntax is included. As noted, aspects of language beyond these three elements are more important to native speakers, making this inventory of material inadequate for what native speakers need. Another drawback is that not every language has the same patterns as other languages, making large portions of the questionnaire irrelevant for many languages while potentially omitting material that is relevant, even within sound systems, morphology and syntax, particularly in languages that have not yet been descriptively analyzed. Descriptions written according to this kind of questionnaire are therefore linguistically incomplete in ways that could be avoided if a less prescriptive approach to the contents of descriptions were more acceptable within the field. Fortunately, this approach to description has been changing over the past couple of decades. More and more linguists are including more kinds of linguistic information in their descriptions, making their descriptions truer to the rich variety of languages and what language speakers can do with their languages.
This traditional focus on only certain aspects of language and the neglect of others is not usually included in the notion of linguistic prescriptivism, yet the effects are similar. Analogous to the ways in which language prescriptivism favors certain variants over others, professional biases favor certain language domains over others in language descriptions. Additionally, in the way that favored variants in language prescriptivism can come to be seen as more legitimate than others, the most-described domains of language can also come to be seen as more legitimate, even more linguistic. The bias already present in previous descriptions has a way of perpetuating itself, not unlike the traditional character of language prescriptivism. While the linguists who write and consult these descriptions have no intention of telling speakers how they should use or think about language, linguists might still be susceptible to the ‘is becomes ought’ process that Joseph outlines in this volume. What has been done in descriptions previously often continues to be done.
This kind of unstated bias is particularly difficult to identify and be honest about. Once recognized, however, it becomes clear that the kinds of information collected often reflect the ideals of linguists that have arisen through the history of the discipline more than they reflect realities among members of a language community. In recordings and observations that I have made in which people were spontaneously engaged in language use, the clear majority of the content comprised personal narratives, discussions of current and future events, recitations of genealogies and other types of discourse. Overt discussion of structures in words or sentences was rare, occurring only for about 15 minutes as part of a single and unusual discussion of sociolinguistic phenomena. The Hobongan used specific examples of words, as well as personal narratives of their experiences, to comment on their sociolinguistic situation; they framed their discussion and discourse sociolinguistically rather than syntactically. If linguists were objective and descriptive observers of the language around them in a field situation, then discourses, and the pragmatics that make discourses work, would be the major component of description, followed by sociolinguistics, with only a few examples of sentences and sounds-in-words – a pattern that is nearly opposite to how content is managed in traditional language descriptions.
Furthermore, a question is raised about how traditional limitations on descriptions might have affected or might continue to affect language conservation efforts. When
