This chapter highlights the development of language documentation in Indonesia, both at universities and at research institutions. It also presents the roles of linguistics associations in Indonesia as actors of supporting language documentation. Finally, it discusses what it takes to build a national language archive, and describes how a language documentation archiving project intends to preserve language diversity in Indonesia.
With its nearly 300 languages — around 219 non-Austronesian and 57 Austronesian — Western New Guinea is home to about 38 percent of all languages in Indonesia (Lewis et al. 2014). Against the background of this immense linguistic diversity, this chapter investigates how the documentarist turn has entered the Indonesian research landscape, especially in Western New Guinea (WNG). This article first provides an overview of the linguistic and sociolinguistic landscapes in WNG, highlighting the importance of language documentation in the region. Second, it discusses different efforts on the part of various institutions to document languages in WNG using classical or traditional methods. Third, it discusses the era when modern language documentation was introduced to the Indonesian and Western New Guinea research landscape, which then contributed to the establishment of the Center for Endangered Languages Documentation (CELD), the country’s first centre for the documentation of indigenous languages in Indonesia.
This chapter reflects on over two decades of linguistic research in Indonesia, examining collaborations between native- and non-native-speaker linguists and local speech communities in documenting and describing minority and endangered languages. Drawing on case studies from Balinese, Rongga, Marori, and Enggano, it explores how positionality, authority, and co-production shape grammar writing and broader documentation outcomes. Situating these experiences within the shift toward community-led, ethically grounded, and ethnographically informed fieldwork, the chapter underscores the importance of capacity building for native-speaker linguists and sustained engagement with both communities and local institutions. It argues that partnerships grounded in mutual respect and shared epistemic goals can produce richer, more culturally embedded, and socially impactful linguistic descriptions, advancing a more inclusive and decolonial linguistic science.
The documentarist turn remains lopsided with respect to the form<>meaning mapping: we risk knowing what speakers said, but not what they meant.
Original recordings thus need to be supplemented by discussions with speakers that use methods more similar to traditional methods of textual hermeneutics.
Here we exemplify these problems, drawing on our own work with previously untranscribed, untranslated audio recordings in Dalabon. We examine four key issues: an unreported “reversal” construction; adding senses to known vocabulary; unearthing previously unrecorded new food-preparation vocabulary; reference of “triangular kin terms” in actual conversation.
Such interpretive discussions with later-generation community members reach beyond the audible to increase the reach of semantic interpretation.
Language archives are a type of research infrastructure that grew out of requirements and efforts of language documentation. From the very beginning of language documentation as a separate research paradigm, it has been accompanied by a strong and vibrant methodological discourse. This discussion of methods and data has not only led to a clear framework of documentary linguistics, but also specified requirements for data and metadata formats, software tools, language archives and ontologies. A whole ecosystem of data infrastructure along with new data types has grown out of this methodological discourse and in close dialogue with language documentation practice.
How do public media represent language documentation? We offer a first pilot study informed by the Public Understanding of Science framework and recent media history. We employ a mixed-methods approach on a corpus of selected media coverage, using Topic Modelling to identify and quantify recurring topics, and corpus-assisted discourse analysis in the form of KWIC and collocations for qualitative analysis. We show that only a good one in two mentions of language documentation in media coverage corresponds to the way it is accepted in academia, whereas outlets tend to highlight certain characteristics of the discipline, overemphasise certain facets, and sometimes ignore relevant strands. In addition, there are competing meanings in discourse which belong to unrelated fields.
The aim of language documentation is to provide a comprehensive record of the observable linguistic behaviour and practices of a language community to be used now and in the future. The endeavour goes beyond the mere description of the linguistic system. To serve as a multipurpose record, the documentary record must capture language use in its fullest form — as a fundamentally multimodal and interactional phenomenon. This means that video recording is essential to the collection of data in language documentation efforts, and that the description and analysis of data must ultimately extend to and integrate multimodal and interactive components of language use. In this chapter, we outline the implications for language documentation from such a conception of language.
This chapter argues for a decisive shift in corpus-building practices toward a Multimodal Turn. For decades, smaller-scale spoken and signed languages were less available for linguistic corpus research due to the scarcity of corpora, but the situation has changed: language documentation now provides rich multimodal resources, and many sign languages are recorded in high-definition video, offering unique insights into communicative practices. In contrast, corpora of majority languages often remain limited in video data. We contend that they can draw inspiration from the practices developed in language documentation and sign language research to close this gap. Only by systematically integrating multimodal interaction into corpus design can linguistic research capture the full complexity of human communication.
The study of multimodality in communication is a field of research that has been receiving increasing attention in various humanities disciplines since the 1990s. While research on multimodality deals extensively with communicative situations in recent European and Anglo-Saxon cultures, such situations in ancient times have hardly been considered. The article discusses some of the challenges and potentials of multimodality research for the study of historical languages using the example of ancient Egyptian monumental inscriptions. It focuses on two topics. Firstly, the question is examined whether and to what extent historical inscriptions can be used for gesture research. Secondly, the parameter of size scaling is used to show that perceptual saliency and visual prominence are relevant for the constitution of meaning in monumental inscriptions.
This chapter is concerned with improving our understanding of literacies through language documentation. It addresses different types of writing practices and the use of documentation corpora in this regard. While writing is often taken for granted for languages with a tradition in this modality, it can be seen as “intrusive and alien” (Finnegan 2008: 22) in other contexts. Yet Himmelmann (cf. 2018: 36) points out that hardly any language community nowadays has not encountered a written standard. Accordingly, this chapter argues to increase the efforts in documenting the rich perspectives on and practices of literacy in linguistic communities worldwide. The rationale is that this may enhance knowledge about literate structures in the languages documented as well as in linguistic theory.
This paper surveys and spotlights historical layers within language documentation corpora, with a particular emphasis on the many languages for which we lack historical written attestation. I highlight the potentials of lesser-prestige and lesser-accessible, often more conservative varieties, as well as the role of oral art. Paying greater attention to existing depth, and intensifying efforts to document lects and genres that point to greater age, will help advance both historical linguistics and typology in significant ways. On the one hand, conservative lects are particularly suitable for the investigation of (micro-)changes given the unusually tight control for other influencing factors. On the other hand, oral art may preserve what is often the only direct evidence of more ancestral forms and structures.
Songs are expressions of creativity and emotion that have great intrinsic value not only to the cultures that produced them, but also to academics from fields such as linguistics and anthropology, who seek to gain deep and meaningful insights into these cultures. Here, we analyse some fascinating aspects of the language of song and related speech acts in four languages of New Guinea. Both structural and semantic aspects of song language are discussed, in order to show the differences between it and everyday speech. Repetitions, metaphor, allusion and unintelligibility are deliberate and prominent features in song language, and this can pose difficult, but not insurmountable, challenges to outsiders who wish to understand (and by extension, translate) the text of the songs.
Transcripts are an integral part of documentary corpora, and transcription, conducted in close cooperation with a language’s speakers/signers, is central to fieldwork activities. Yet, transcription practices — involving decisions that reduce the complexity of an audio/video signal into a written form — remain surprisingly under-researched within the language documentation context. This contribution investigates variability and agreement among transcribers across three languages (Iha, Qaqet, Dëne Sųłıné), focusing on their phonemic and phonetic spellings of nasalization, and asks to what extent the observed transcription strategies provide a window into metalinguistic knowledge. The results paint a complex picture, partly corroborating the view that speakers are aware of phonemes, but also demonstrating that speakers are aware of considerable phonetic detail and automatic (morpho-)phonological processes.
This study investigates how pauses before nouns and verbs interact with word order and prosodic structure across a worldwide sample of 27 languages from the DoReCo corpus. Whereas prior research has suggested that more pauses occur before nouns in general, the results here show that this asymmetry holds consistently only in verb-final languages, while in other languages, pausing patterns are mixed. Since word order mediates pausing asymmetries, it likely also plays a major role in explaining prefixing asymmetries — namely, the global dispreference for prefixes on nouns compared to verbs. The study demonstrates the potential of comparative corpus linguistics for examining typological patterns in speech, especially when incorporating prosodic annotation.
Macro-rhythm, the tonal rhythmicity within phrases, constitutes a typological dimension in prosody. However, little empirical research supports this concept. Work on the rhythm-class hypothesis (i.e., stress-, syllable and mora-timing) and on macro-rhythm quantification based on regular intervals (isochrony) has led to unsatisfactory results. This study therefore abandons the assumption that macro-rhythmic differences should be sought in some form of temporal regularity. The current exploratory approach to macro-rhythm quantification uses a measure of f0 contour length on five languages (Greek, German, European Portuguese, French and Seoul Korean). This measure generally distinguishes the languages in the expected way, and its potential and limitations are discussed in the context of the prosodic typology of rhythm and its applicability to under-documented languages.
This paper examines the acoustic cues and phonological behaviour of vowel length in Meto (Austronesian, Timor), which is described as having a distinction between “double” (/aa/ → [aː]) and “single” (/a/ → [a]) vowels. We examine acoustic evidence for this distinction, finding that putative double vowels are ~40% longer than single vowels. There is also some evidence of following consonant duration and vowel quality differences between single and double vowels.
We then examine the phonological behaviour of the putative double vowels, showing that their phonotactic patterning and (morpho)phonological behaviour is identical to unambiguous vowel sequences. This study highlights how field data can be used in phonetic and phonological research, and the importance of analysis in language documentation.
Languages with “simple” phonological inventories are often neglected in traditional descriptive phonetic research in favour of those with more varied segmental and prosodic features. Tahitian (Eastern Polynesian) is a prime example with no experimental phonetic investigation of the language in spite of a relatively long history of language documentation. This chapter explores the interaction between prosody (stress, phrasing), glottal stop variation and vowel glottalization. Phonemic glottal stops in many languages of the world have a range of variants, from fully occluded voiceless glottal stops to creaky voiced vowels. Tahitian is no exception but there is a relatively complex relationship between glottal variation and prosodic function (i.e. prominence and phrasing), suggesting that languages with “simple” phonological systems illustrate phonetic complexity.
Since Givón (1983) and Ariel (1990), a large body of research continues to assume that zero anaphors reflect a higher degree of accessibility than overt pronominal forms. However, when the features of animacy or person are taken as its metric, existing case studies do not consistently confirm a correlation between high accessibility and zero form, and some report the opposite tendency. In this chapter we bring a broader cross-linguistic perspective to the question, leveraging data from a multilingual richly annotated spoken language corpus covering 16 typologically diverse languages. We identify a surprisingly robust trend in support of a reversal of expected accessibility-accounts: Overall, it is non-human referents rather than human, and third person rather than first and second person, which favour zero expression. Our findings provide cross-linguistic support for the constraint ‘avoid non-human pronouns’, found by Genetti and Crain (2003) for Nepali, and a tendency to overtly realise speech-act participants as pronouns. Furthermore, we identify a significant effect of syntactic role, such that these regularities are more clearly evident for direct objects than for transitive subjects. We explore possible explanations for these results, noting similarities between our findings and well-documented patterns of differential indexing, informativity and surprisal (Haspelmath 2021). More generally, these findings call for a re-assessment of the view that a single principle such as accessibility can account for referential choice among all form types. For the choice between reduced forms (pronouns and zero), distinct principles may be operative.
Austronesian voice has been claimed to originate in an ancient reanalysis of participant nominalization as matrix clause predicates. A survey of participant nominalization as a relativizing strategy reveals that Austronesian voice has many analogues throughout the world that show certain typological hallmarks such as apparently null-headed relative clauses, genitive marked agents and a lack of dedicated relativizers and copulas. Many Austronesian languages maintain remnants of the original verbal forms alongside the nominalized forms that took over main clause functions in most languages. New observations about the case assignment properties of these forms in several Philippine languages strengthen the argument that they remain as a distinct verbal syntactic category in modern languages, in comparison to the participle-like forms stemming from nominalization.
In languages of western Indonesia with symmetrical voice systems, apparent passive constructions often display overlapping features with transitive P-oriented voice (PV) constructions. Drawing on a corpus of everyday conversations in Besemah, this chapter demonstrates that it is untenable to distinguish transitive PV constructions from such passive constructions, calling into question core features of both: A arguments are not obligatory in unmarked PV constructions; ‘by’-phrases serve multiple functions, and it is unclear whether A demotion is in fact one of these functions; PV constructions without A arguments exhibit the same patterns of unrealized A reference found elsewhere in the language. This chapter has broader implications for the analysis of both passives and symmetrical voice alternations in the languages of western Indonesia.
Language documentation has the broad aim of building a representative evidence base across typologically diverse languages and contexts of use. One such context that is gaining increased attention is language acquisition, owing to its importance in language maintenance in an era of significant language endangerment and loss. In this chapter we discuss the acquisition of a typologically important feature of Austronesian languages — symmetrical voice — to which Nikolaus Himmelmann has made distinguished contributions. We consider three languages that span the commonly made divide between Philippine-type and Indonesian-type languages: Tagalog, Jakarta Indonesian, and Totoli. We first compare the previous literature on the acquisition of Tagalog and Indonesian, for which there is significant data. We then consider how Totoli differs from Philippine- and Indonesian-type languages, and, based on a preliminary corpus observation, make predictions on the acquisition process of the Totoli voice system.
According to Himmelmann’s (2008: 249) Syntactic Uniformity Hypothesis for Tagalog, words from different lexical categories “may all occur in essentially the same basic syntactic positions”. This is also true of Movima (isolate, Bolivia). Based on a documentation corpus of spontaneous Movima discourse data, this paper furthermore shows that argument DPs containing a verb are particularly frequent in combination with nonverbal predicates (nouns, quantifiers, demonstratives), i.e. their syntactic distribution is diametrically opposed to that of nominal DPs. It is suggested that in a system with syntactic uniformity, a DP containing a verb functions in a way similar to a headless or light-headed relative clause in a language whose verbs and nouns are more strictly linked to the syntactic categories predicate and argument.
This paper investigates correlations between information structure and constituent order patterns in transitive clauses extracted from recorded texts in Jaminjung-Ngaliwurru, an Australian language of the Mirndi family without a dominant constituent order at clause level. In transitive clauses with two overt arguments, and considering only linear order, the findings support a widely reported cross-linguistic preference for orders where the agent precedes the patient. However, the prevalence of agent-first orders is an epiphenomenon of two related preferences: a high probability for topics to be agents, and a preference for a rheme-internal placement of the patient, in contiguity with the verb. The result is that agents, in the case that they are overt at all, tend to be placed in rheme-external position, usually as initial topics.
Reference to an implicit argument in a bridging context is made by definite determiner phrases (DPs) (a book... the author), and it is assumed that demonstrative DPs are not felicitous in such contexts (a book... #that author). However, the literature reports bridging contexts with demonstrative DPs. We argue that the recognitional use of demonstratives, first described in Himmelmann (1996, 1997), can license demonstrative DPs in such contexts. Bridging contexts rely on general knowledge, making definite DPs the default in the “larger situational use” (Hawkins 1978). We argue that demonstrative DPs are possible in such contexts if they signal special shared knowledge between speaker and hearer, i.e. if they are recognitional. Two rating tasks with evaluative adjectives provide evidence for the recognitional function of bridging demonstratives.
Under-described languages may present diverse strategies of structuring narrative discourse. Investigations into the structure of narratives of such languages, in particular ones with a predominantly oral tradition, are still scarce, and even more so studies that relate the structure with the corresponding linguistic devices. A critical component of the hierarchical structure of discourse is captured in semantic-pragmatic relationships within and across fragments, i.e. strings of text of the length of one or more paragraphs. We investigate texts from two less described languages (Samoan, Austronesian and Chipaya, Uru-Chipayan) to delineate their hierarchical structure on a local and a global level applying three different semantic-pragmatic approaches (among them SDRT) in a comparative way.
Co-present conversation is the primary habitat of human language and thus most probably constitutes an important locus of language change. However, language change is observable only at much larger timescales. How, then, is it possible to study the real-time conversational dimension of language change? In this paper, we argue that language documentation corpora, often covering spontaneous conversational language use, will constitute a crucial data source for such an endeavor, next to historical sources. We report two case studies illustrating the investigation of the mechanism of conversational priming in repetitional responses, where a responding interactant may repeat an innovative form used by the other interlocutor, which may in turn facilitate the spread of this innovation across a population.
The marker yimarne(k) in Kunbarlang can be used to signal “mistaken belief”, whereby a proposition held as a belief at some point is now considered false. The various functions and formal types of expressions yimarne(k) combines with point to an analysis in terms of two separate markers, a short form with a core function of marking similarity, and a long form expressing counterfactuality. This suggests that the long form’s mistaken belief use might stem from an original similarity marker with an added element of unclear origin. To further assess the potential pathway whereby a similative acquires counterfactual uses including mistaken belief, we compare the case of yimarne(k) to attested markers of mistaken belief both within the Gunwinyguan family and beyond.
Reported speech constructions in many Papuan languages exhibit a range of functions that go beyond the primary function of representing an utterance. For some of these languages, it has been pointed out that a specific verb form — the so called medial-verb form — is required to introduce reported speech in these contexts. This paper presents synchronic evidence for a grammaticisation cline that describes the development from a medial speech introducing clause that introduces reported speech proper in a clause chaining construction to a marker of extended reported speech, a quotative marker, a multi-purpose NP marker, and a marker of epistemic stance.
Little is certain about the origins of Papuan Malay. The literature offers two opposing viewpoints: one is that Papuan Malay arose out of centuries of Malay use in New Guinea connected with indigenous trade networks centred on Ternate and the North Moluccas, the other is that Papuan Malay arose very late as part of colonial policies in the 20th century. The chapter presents the first examination of these conflicting claims on the basis of an in-depth analysis of the lexicon of Papuan Malay and Malay loanwords along the north coast of New Guinea. Papuan Malay is shown to have a significant lexical component from Ternate, like other vehicular Malays in the region. Despite this, the evidence as a whole supports the theory of a recent, not historically distant, origin for Papuan Malay.
This paper investigates the origins of Old and Modern Javanese negators.
Modern Javanese distinguishes four negators.
Predicative ora (‘no(t)’) derives from Old Javanese tan+wwara ‘there is not’ and its high register counterpart botən from tan+wontən.
Prohibitive aja (‘don’t’) and its Old Javanese predecessor haywa must derive from Old Javanese hayu ‘beauty, goodness, rightness; beautiful, good, right’ + subjunctive -a: hayu-a was used in desiderative and hortative phrases (‘it would be good if…’), and, with tan, in prohibitives. After loss of tan, hayu-a became lexicalised as haywa, a prohibitive. Its high counterpart sampun is basically an aspect marker (‘already’), which acquired a cessative and prohibitive function.
Contrastive dudu (‘no [+noun]; not this [but another]’) originally meant ‘other’.
The origin of Phasal duruŋ ‘not yet’ needs further investigation.
This chapter highlights the development of language documentation in Indonesia, both at universities and at research institutions. It also presents the roles of linguistics associations in Indonesia as actors of supporting language documentation. Finally, it discusses what it takes to build a national language archive, and describes how a language documentation archiving project intends to preserve language diversity in Indonesia.
With its nearly 300 languages — around 219 non-Austronesian and 57 Austronesian — Western New Guinea is home to about 38 percent of all languages in Indonesia (Lewis et al. 2014). Against the background of this immense linguistic diversity, this chapter investigates how the documentarist turn has entered the Indonesian research landscape, especially in Western New Guinea (WNG). This article first provides an overview of the linguistic and sociolinguistic landscapes in WNG, highlighting the importance of language documentation in the region. Second, it discusses different efforts on the part of various institutions to document languages in WNG using classical or traditional methods. Third, it discusses the era when modern language documentation was introduced to the Indonesian and Western New Guinea research landscape, which then contributed to the establishment of the Center for Endangered Languages Documentation (CELD), the country’s first centre for the documentation of indigenous languages in Indonesia.
This chapter reflects on over two decades of linguistic research in Indonesia, examining collaborations between native- and non-native-speaker linguists and local speech communities in documenting and describing minority and endangered languages. Drawing on case studies from Balinese, Rongga, Marori, and Enggano, it explores how positionality, authority, and co-production shape grammar writing and broader documentation outcomes. Situating these experiences within the shift toward community-led, ethically grounded, and ethnographically informed fieldwork, the chapter underscores the importance of capacity building for native-speaker linguists and sustained engagement with both communities and local institutions. It argues that partnerships grounded in mutual respect and shared epistemic goals can produce richer, more culturally embedded, and socially impactful linguistic descriptions, advancing a more inclusive and decolonial linguistic science.
The documentarist turn remains lopsided with respect to the form<>meaning mapping: we risk knowing what speakers said, but not what they meant.
Original recordings thus need to be supplemented by discussions with speakers that use methods more similar to traditional methods of textual hermeneutics.
Here we exemplify these problems, drawing on our own work with previously untranscribed, untranslated audio recordings in Dalabon. We examine four key issues: an unreported “reversal” construction; adding senses to known vocabulary; unearthing previously unrecorded new food-preparation vocabulary; reference of “triangular kin terms” in actual conversation.
Such interpretive discussions with later-generation community members reach beyond the audible to increase the reach of semantic interpretation.
Language archives are a type of research infrastructure that grew out of requirements and efforts of language documentation. From the very beginning of language documentation as a separate research paradigm, it has been accompanied by a strong and vibrant methodological discourse. This discussion of methods and data has not only led to a clear framework of documentary linguistics, but also specified requirements for data and metadata formats, software tools, language archives and ontologies. A whole ecosystem of data infrastructure along with new data types has grown out of this methodological discourse and in close dialogue with language documentation practice.
How do public media represent language documentation? We offer a first pilot study informed by the Public Understanding of Science framework and recent media history. We employ a mixed-methods approach on a corpus of selected media coverage, using Topic Modelling to identify and quantify recurring topics, and corpus-assisted discourse analysis in the form of KWIC and collocations for qualitative analysis. We show that only a good one in two mentions of language documentation in media coverage corresponds to the way it is accepted in academia, whereas outlets tend to highlight certain characteristics of the discipline, overemphasise certain facets, and sometimes ignore relevant strands. In addition, there are competing meanings in discourse which belong to unrelated fields.
The aim of language documentation is to provide a comprehensive record of the observable linguistic behaviour and practices of a language community to be used now and in the future. The endeavour goes beyond the mere description of the linguistic system. To serve as a multipurpose record, the documentary record must capture language use in its fullest form — as a fundamentally multimodal and interactional phenomenon. This means that video recording is essential to the collection of data in language documentation efforts, and that the description and analysis of data must ultimately extend to and integrate multimodal and interactive components of language use. In this chapter, we outline the implications for language documentation from such a conception of language.
This chapter argues for a decisive shift in corpus-building practices toward a Multimodal Turn. For decades, smaller-scale spoken and signed languages were less available for linguistic corpus research due to the scarcity of corpora, but the situation has changed: language documentation now provides rich multimodal resources, and many sign languages are recorded in high-definition video, offering unique insights into communicative practices. In contrast, corpora of majority languages often remain limited in video data. We contend that they can draw inspiration from the practices developed in language documentation and sign language research to close this gap. Only by systematically integrating multimodal interaction into corpus design can linguistic research capture the full complexity of human communication.
The study of multimodality in communication is a field of research that has been receiving increasing attention in various humanities disciplines since the 1990s. While research on multimodality deals extensively with communicative situations in recent European and Anglo-Saxon cultures, such situations in ancient times have hardly been considered. The article discusses some of the challenges and potentials of multimodality research for the study of historical languages using the example of ancient Egyptian monumental inscriptions. It focuses on two topics. Firstly, the question is examined whether and to what extent historical inscriptions can be used for gesture research. Secondly, the parameter of size scaling is used to show that perceptual saliency and visual prominence are relevant for the constitution of meaning in monumental inscriptions.
This chapter is concerned with improving our understanding of literacies through language documentation. It addresses different types of writing practices and the use of documentation corpora in this regard. While writing is often taken for granted for languages with a tradition in this modality, it can be seen as “intrusive and alien” (Finnegan 2008: 22) in other contexts. Yet Himmelmann (cf. 2018: 36) points out that hardly any language community nowadays has not encountered a written standard. Accordingly, this chapter argues to increase the efforts in documenting the rich perspectives on and practices of literacy in linguistic communities worldwide. The rationale is that this may enhance knowledge about literate structures in the languages documented as well as in linguistic theory.
This paper surveys and spotlights historical layers within language documentation corpora, with a particular emphasis on the many languages for which we lack historical written attestation. I highlight the potentials of lesser-prestige and lesser-accessible, often more conservative varieties, as well as the role of oral art. Paying greater attention to existing depth, and intensifying efforts to document lects and genres that point to greater age, will help advance both historical linguistics and typology in significant ways. On the one hand, conservative lects are particularly suitable for the investigation of (micro-)changes given the unusually tight control for other influencing factors. On the other hand, oral art may preserve what is often the only direct evidence of more ancestral forms and structures.
Songs are expressions of creativity and emotion that have great intrinsic value not only to the cultures that produced them, but also to academics from fields such as linguistics and anthropology, who seek to gain deep and meaningful insights into these cultures. Here, we analyse some fascinating aspects of the language of song and related speech acts in four languages of New Guinea. Both structural and semantic aspects of song language are discussed, in order to show the differences between it and everyday speech. Repetitions, metaphor, allusion and unintelligibility are deliberate and prominent features in song language, and this can pose difficult, but not insurmountable, challenges to outsiders who wish to understand (and by extension, translate) the text of the songs.
Transcripts are an integral part of documentary corpora, and transcription, conducted in close cooperation with a language’s speakers/signers, is central to fieldwork activities. Yet, transcription practices — involving decisions that reduce the complexity of an audio/video signal into a written form — remain surprisingly under-researched within the language documentation context. This contribution investigates variability and agreement among transcribers across three languages (Iha, Qaqet, Dëne Sųłıné), focusing on their phonemic and phonetic spellings of nasalization, and asks to what extent the observed transcription strategies provide a window into metalinguistic knowledge. The results paint a complex picture, partly corroborating the view that speakers are aware of phonemes, but also demonstrating that speakers are aware of considerable phonetic detail and automatic (morpho-)phonological processes.
This study investigates how pauses before nouns and verbs interact with word order and prosodic structure across a worldwide sample of 27 languages from the DoReCo corpus. Whereas prior research has suggested that more pauses occur before nouns in general, the results here show that this asymmetry holds consistently only in verb-final languages, while in other languages, pausing patterns are mixed. Since word order mediates pausing asymmetries, it likely also plays a major role in explaining prefixing asymmetries — namely, the global dispreference for prefixes on nouns compared to verbs. The study demonstrates the potential of comparative corpus linguistics for examining typological patterns in speech, especially when incorporating prosodic annotation.
Macro-rhythm, the tonal rhythmicity within phrases, constitutes a typological dimension in prosody. However, little empirical research supports this concept. Work on the rhythm-class hypothesis (i.e., stress-, syllable and mora-timing) and on macro-rhythm quantification based on regular intervals (isochrony) has led to unsatisfactory results. This study therefore abandons the assumption that macro-rhythmic differences should be sought in some form of temporal regularity. The current exploratory approach to macro-rhythm quantification uses a measure of f0 contour length on five languages (Greek, German, European Portuguese, French and Seoul Korean). This measure generally distinguishes the languages in the expected way, and its potential and limitations are discussed in the context of the prosodic typology of rhythm and its applicability to under-documented languages.
This paper examines the acoustic cues and phonological behaviour of vowel length in Meto (Austronesian, Timor), which is described as having a distinction between “double” (/aa/ → [aː]) and “single” (/a/ → [a]) vowels. We examine acoustic evidence for this distinction, finding that putative double vowels are ~40% longer than single vowels. There is also some evidence of following consonant duration and vowel quality differences between single and double vowels.
We then examine the phonological behaviour of the putative double vowels, showing that their phonotactic patterning and (morpho)phonological behaviour is identical to unambiguous vowel sequences. This study highlights how field data can be used in phonetic and phonological research, and the importance of analysis in language documentation.
Languages with “simple” phonological inventories are often neglected in traditional descriptive phonetic research in favour of those with more varied segmental and prosodic features. Tahitian (Eastern Polynesian) is a prime example with no experimental phonetic investigation of the language in spite of a relatively long history of language documentation. This chapter explores the interaction between prosody (stress, phrasing), glottal stop variation and vowel glottalization. Phonemic glottal stops in many languages of the world have a range of variants, from fully occluded voiceless glottal stops to creaky voiced vowels. Tahitian is no exception but there is a relatively complex relationship between glottal variation and prosodic function (i.e. prominence and phrasing), suggesting that languages with “simple” phonological systems illustrate phonetic complexity.
Since Givón (1983) and Ariel (1990), a large body of research continues to assume that zero anaphors reflect a higher degree of accessibility than overt pronominal forms. However, when the features of animacy or person are taken as its metric, existing case studies do not consistently confirm a correlation between high accessibility and zero form, and some report the opposite tendency. In this chapter we bring a broader cross-linguistic perspective to the question, leveraging data from a multilingual richly annotated spoken language corpus covering 16 typologically diverse languages. We identify a surprisingly robust trend in support of a reversal of expected accessibility-accounts: Overall, it is non-human referents rather than human, and third person rather than first and second person, which favour zero expression. Our findings provide cross-linguistic support for the constraint ‘avoid non-human pronouns’, found by Genetti and Crain (2003) for Nepali, and a tendency to overtly realise speech-act participants as pronouns. Furthermore, we identify a significant effect of syntactic role, such that these regularities are more clearly evident for direct objects than for transitive subjects. We explore possible explanations for these results, noting similarities between our findings and well-documented patterns of differential indexing, informativity and surprisal (Haspelmath 2021). More generally, these findings call for a re-assessment of the view that a single principle such as accessibility can account for referential choice among all form types. For the choice between reduced forms (pronouns and zero), distinct principles may be operative.
Austronesian voice has been claimed to originate in an ancient reanalysis of participant nominalization as matrix clause predicates. A survey of participant nominalization as a relativizing strategy reveals that Austronesian voice has many analogues throughout the world that show certain typological hallmarks such as apparently null-headed relative clauses, genitive marked agents and a lack of dedicated relativizers and copulas. Many Austronesian languages maintain remnants of the original verbal forms alongside the nominalized forms that took over main clause functions in most languages. New observations about the case assignment properties of these forms in several Philippine languages strengthen the argument that they remain as a distinct verbal syntactic category in modern languages, in comparison to the participle-like forms stemming from nominalization.
In languages of western Indonesia with symmetrical voice systems, apparent passive constructions often display overlapping features with transitive P-oriented voice (PV) constructions. Drawing on a corpus of everyday conversations in Besemah, this chapter demonstrates that it is untenable to distinguish transitive PV constructions from such passive constructions, calling into question core features of both: A arguments are not obligatory in unmarked PV constructions; ‘by’-phrases serve multiple functions, and it is unclear whether A demotion is in fact one of these functions; PV constructions without A arguments exhibit the same patterns of unrealized A reference found elsewhere in the language. This chapter has broader implications for the analysis of both passives and symmetrical voice alternations in the languages of western Indonesia.
Language documentation has the broad aim of building a representative evidence base across typologically diverse languages and contexts of use. One such context that is gaining increased attention is language acquisition, owing to its importance in language maintenance in an era of significant language endangerment and loss. In this chapter we discuss the acquisition of a typologically important feature of Austronesian languages — symmetrical voice — to which Nikolaus Himmelmann has made distinguished contributions. We consider three languages that span the commonly made divide between Philippine-type and Indonesian-type languages: Tagalog, Jakarta Indonesian, and Totoli. We first compare the previous literature on the acquisition of Tagalog and Indonesian, for which there is significant data. We then consider how Totoli differs from Philippine- and Indonesian-type languages, and, based on a preliminary corpus observation, make predictions on the acquisition process of the Totoli voice system.
According to Himmelmann’s (2008: 249) Syntactic Uniformity Hypothesis for Tagalog, words from different lexical categories “may all occur in essentially the same basic syntactic positions”. This is also true of Movima (isolate, Bolivia). Based on a documentation corpus of spontaneous Movima discourse data, this paper furthermore shows that argument DPs containing a verb are particularly frequent in combination with nonverbal predicates (nouns, quantifiers, demonstratives), i.e. their syntactic distribution is diametrically opposed to that of nominal DPs. It is suggested that in a system with syntactic uniformity, a DP containing a verb functions in a way similar to a headless or light-headed relative clause in a language whose verbs and nouns are more strictly linked to the syntactic categories predicate and argument.
This paper investigates correlations between information structure and constituent order patterns in transitive clauses extracted from recorded texts in Jaminjung-Ngaliwurru, an Australian language of the Mirndi family without a dominant constituent order at clause level. In transitive clauses with two overt arguments, and considering only linear order, the findings support a widely reported cross-linguistic preference for orders where the agent precedes the patient. However, the prevalence of agent-first orders is an epiphenomenon of two related preferences: a high probability for topics to be agents, and a preference for a rheme-internal placement of the patient, in contiguity with the verb. The result is that agents, in the case that they are overt at all, tend to be placed in rheme-external position, usually as initial topics.
Reference to an implicit argument in a bridging context is made by definite determiner phrases (DPs) (a book... the author), and it is assumed that demonstrative DPs are not felicitous in such contexts (a book... #that author). However, the literature reports bridging contexts with demonstrative DPs. We argue that the recognitional use of demonstratives, first described in Himmelmann (1996, 1997), can license demonstrative DPs in such contexts. Bridging contexts rely on general knowledge, making definite DPs the default in the “larger situational use” (Hawkins 1978). We argue that demonstrative DPs are possible in such contexts if they signal special shared knowledge between speaker and hearer, i.e. if they are recognitional. Two rating tasks with evaluative adjectives provide evidence for the recognitional function of bridging demonstratives.
Under-described languages may present diverse strategies of structuring narrative discourse. Investigations into the structure of narratives of such languages, in particular ones with a predominantly oral tradition, are still scarce, and even more so studies that relate the structure with the corresponding linguistic devices. A critical component of the hierarchical structure of discourse is captured in semantic-pragmatic relationships within and across fragments, i.e. strings of text of the length of one or more paragraphs. We investigate texts from two less described languages (Samoan, Austronesian and Chipaya, Uru-Chipayan) to delineate their hierarchical structure on a local and a global level applying three different semantic-pragmatic approaches (among them SDRT) in a comparative way.
Co-present conversation is the primary habitat of human language and thus most probably constitutes an important locus of language change. However, language change is observable only at much larger timescales. How, then, is it possible to study the real-time conversational dimension of language change? In this paper, we argue that language documentation corpora, often covering spontaneous conversational language use, will constitute a crucial data source for such an endeavor, next to historical sources. We report two case studies illustrating the investigation of the mechanism of conversational priming in repetitional responses, where a responding interactant may repeat an innovative form used by the other interlocutor, which may in turn facilitate the spread of this innovation across a population.
The marker yimarne(k) in Kunbarlang can be used to signal “mistaken belief”, whereby a proposition held as a belief at some point is now considered false. The various functions and formal types of expressions yimarne(k) combines with point to an analysis in terms of two separate markers, a short form with a core function of marking similarity, and a long form expressing counterfactuality. This suggests that the long form’s mistaken belief use might stem from an original similarity marker with an added element of unclear origin. To further assess the potential pathway whereby a similative acquires counterfactual uses including mistaken belief, we compare the case of yimarne(k) to attested markers of mistaken belief both within the Gunwinyguan family and beyond.
Reported speech constructions in many Papuan languages exhibit a range of functions that go beyond the primary function of representing an utterance. For some of these languages, it has been pointed out that a specific verb form — the so called medial-verb form — is required to introduce reported speech in these contexts. This paper presents synchronic evidence for a grammaticisation cline that describes the development from a medial speech introducing clause that introduces reported speech proper in a clause chaining construction to a marker of extended reported speech, a quotative marker, a multi-purpose NP marker, and a marker of epistemic stance.
Little is certain about the origins of Papuan Malay. The literature offers two opposing viewpoints: one is that Papuan Malay arose out of centuries of Malay use in New Guinea connected with indigenous trade networks centred on Ternate and the North Moluccas, the other is that Papuan Malay arose very late as part of colonial policies in the 20th century. The chapter presents the first examination of these conflicting claims on the basis of an in-depth analysis of the lexicon of Papuan Malay and Malay loanwords along the north coast of New Guinea. Papuan Malay is shown to have a significant lexical component from Ternate, like other vehicular Malays in the region. Despite this, the evidence as a whole supports the theory of a recent, not historically distant, origin for Papuan Malay.
This paper investigates the origins of Old and Modern Javanese negators.
Modern Javanese distinguishes four negators.
Predicative ora (‘no(t)’) derives from Old Javanese tan+wwara ‘there is not’ and its high register counterpart botən from tan+wontən.
Prohibitive aja (‘don’t’) and its Old Javanese predecessor haywa must derive from Old Javanese hayu ‘beauty, goodness, rightness; beautiful, good, right’ + subjunctive -a: hayu-a was used in desiderative and hortative phrases (‘it would be good if…’), and, with tan, in prohibitives. After loss of tan, hayu-a became lexicalised as haywa, a prohibitive. Its high counterpart sampun is basically an aspect marker (‘already’), which acquired a cessative and prohibitive function.
Contrastive dudu (‘no [+noun]; not this [but another]’) originally meant ‘other’.
The origin of Phasal duruŋ ‘not yet’ needs further investigation.