The aim of this study was to pursue greater knowledge of vocal expressions in the borderland between speech and song through collaboration between researchers with different approaches, with a view to developing an interdisciplinary method for the analysis of such expressions. The research presented here is the outcome of the research project ‘In the Borderland between Song and Speech. Vocal Expressions in Oral Cultures’ carried out in 2011–14 with support from the Swedish Research Council. The starting point was a conference on endangered languages and musics, Humanities of the Lesser-Known, organized in Lund in 2010, which brought together those who came to be the members of the Borderland project.¹

The material is intercultural and includes a variety of language and music contexts: Kammu (Laos), Akha (Thailand), Seediq (Taiwan), Tanana (Interior Alaska), and Ryukyuan (Okinawa, Japan). A long-term aim has also been to play a part in the revitalization of such oral traditions and to contribute to their sustainability. The languages belong to different language families, Austroasiatic (Kammu), Sino-Tibetan (Akha), Austronesian (Seediq), Athabascan (Tanana), and Japonic (Ryukyuan). They are spoken in different parts of the world, but they also have much in common. Except for Ryukyuan, they lack a written tradition, and most of them are endangered to some degree. Tanana and Ryukyuan have very few speakers, and even Kammu, with at least half a million speakers and Akha with still more, are under constant pressure from the majority languages in their areas. Kammu and Akha are fully fledged tone languages and Tanana and Ryukyuan use lexical tones to some extent.

Our main interest has not been in what is performed – be it ‘song’, ‘recitation’, ‘prayer’, or ‘narration’ – but in the techniques that make improvisation, variation, re-creation, and creation possible in the kind of performances we include in the concept vocal expression. Thus, we focus on performance and not on the vocal expressions as artefacts. When we speak of transmission, it is not about the transmission of individual ‘poems’ or ‘songs’ but about the intergenerational transmission of the techniques used in order to realize vocal expressions in performance. When we study material from diverse cultures, the goal is not comparison in itself but, rather, understanding the diversity of culturally specific techniques of performance.

Background

The question of how music and language relate to each other, how they are combined or integrated, has been approached from different perspectives which have increased both our knowledge and the number of questions that arise. A recurring subject is the origin of song and speech.² Recent theories tend to see a common origin, believing that song and speech have diversified during the evolution of humans as biological and social beings.³ From another perspective, it has been shown that language and music share the same cognitive resources, and even that the phonology of an individual’s native language can influence both musical appreciation and composition.⁴ A further approach is the particular study of tone languages and music that involves transcription methods.⁵ In this study, we will focus on the language–music relation in sung or recited performances from different, predominantly oral, cultures by combining methodologies from musicology and linguistics.

In everyday situations, the meanings of ‘speech’ and ‘song’ are quite commonly agreed upon. However, turning to music traditions in cultural settings where music is mainly orally transmitted, the term ‘song’ is often problematic. It is, in itself, ethnocentric and may therefore in many ways be misleading. Furthermore, in research about orally transmitted singing the concept of ‘song’ has been closely associated with the concepts ‘original’ and ‘variant’. These concepts are ideologically burdened, since they imply that it is possible to find one original for a number of versions of the same song, which are thus considered as variants of this known or implied original. This approach is relevant in some situations, but there are also contexts where there is no such thing as an ‘original’. This has been expressed by Albert Lord in an approach that permits one to understand something as multiform:

In her discussion about ‘oral composition and oral literature’, Ruth Finnegan examines Lord’s view on oral composition, using examples from oral traditions in the Pacific where she also finds fixity in the pattern composition–rehearsals–performance that involves memorization. This leads her to conclude that there is more than one method of composition in oral contexts.⁷

In the English language, concepts like ‘song’, ‘declamation’, ‘recitation’, ‘incantation’, and ‘chant’ are – without being exactly defined – normally used to distinguish different genres of vocal performance. In a similar manner, many cultures use different names for more or less distinctly different forms of expression. In some cases, these different forms are quite clearly defined. In the Kammu language, spoken by an ethnic group in northern Laos, for instance, these terms may be understood as levels of ‘speech’ or levels of ‘song’ that represent different positions on a continuum from ‘speech’ to ‘song’, in which there is much overlapping in matters like prosody vs. melody, or rhythm vs. rhythmic patterns. What we refer to as the ‘borderland between song and speech’ is a segment of this continuum.

The term vocal expression will be used as a neutral term for expressions that cannot easily be defined as either speech or song.⁸ It is a term that is not limited to one culture, but can be used globally. As far as possible, indigenous terms will be used, normally rendered in italics. Sometimes it is necessary to talk about ‘song’, ‘recitation’, and the like in a very general sense – especially in discussions of, or communication with, related research when referring to the borderland between song and speech or when discussing ‘school songs’.

Among ethnic minority groups in South-East Asia, the existence of levels of performance is closely related to the re-creation, extemporization, or improvisation of vocal expressions. This is realized in performance by the combination of traditional sets of words or newly created sentences with pre-existing melodic and poetic templates. This is typical of the mono-melodic organization of music culture, in which a limited number of melodic formulae are used for many sets of words.⁹ In such cases, the interplay between language and music is crucial.¹⁰ In the special case of tone languages, yet another interplay between language and music occurs, which may take on different forms.¹¹ Since vocal expressions are closely linked with language, there is a risk that they will disappear as language-knowledge disappears. Most of the languages studied are endangered to some degree, and these musical/poetical traditions are consequently endangered to the same extent as the language they represent.

The borderland between song and speech

Research on the relationship between music and language carried out by musicologists or linguists has a long history. When the ethnomusicologist Steven Feld analysed this field of research in 1974, he found that

In his study of the Kaluli people in Papua New Guinea, Feld worked with a combined ethnomusicological/linguistic approach in order to interpret the social and symbolic levels of Kaluli aesthetics in which nature metaphors play a central role.¹³ In collaboration with Aaron A. Fox a decade later, he recognized four main categories in research on music and language:

1. Music in language: the musical dimensions of language (prosody, rhythm, and timing);
2. Music as language: application of formal linguistic models to music, creating different ‘grammars’ of music;
3. Language about music: the discourse that surrounds and interacts with musical practice;
4. Language in music: different ways in which language and music interact in musical contexts, and especially in song texts.¹⁴

Our research interest basically falls under the heading ‘overlap of musical and linguistic phenomena’. Since we are interested in how language and music interact, though not primarily in ‘song texts’, our research belongs to the fourth category: Language in music. It also relates to the first category: Music in language, since we are interested in the musical aspects of prosody, rhythm, phrasing, and lexical tones.

In an article on poetry, Giorgio Banti and Francesco Giannattasio discuss ‘Between speaking and singing’:

The vocal expressions they choose in order to exemplify the area between speaking and singing are prayers and magic spells for flat or monotone contour and Maori haka for heightened speech, with further examples from children’s counting rhymes, funerary lamentations, religious preaching, and political speeches.

In an attempt to organize forms of vocal expression in which music and language overlap, George List placed them in a continuum from ‘speech’ to ‘song’ on one axis and from ‘monotone’ to ‘sprechstimme’ on the other axis.¹⁶ He used English terms, well aware of the difficulties in translating them to terms in other languages, and his in-between forms include ‘recitation’, ‘intonational recitation’, ‘chant’, and ‘intonational chant’.

Anthony Seeger compared several genres of verbal forms of the Amazonian Suyá in a chapter named ‘Suyá vocal art: from speech to song’.¹⁷ Under ‘textual fixity’, he lists the genres on a continuum from ‘free text without parallelism’ to ‘entirely fixed text & parallelism through repetition’: everyday speech, plaza speech, myth performance, ceremonial recitative, invocation, and song (ngére and akua). Seeger summarizes them in a triangular fashion:

These are different ways of looking at the overlapping of song and speech, or what we have called the borderland between song and speech, seen as two poles of a continuum with no distinct border between the two phenomena. Consequently, this borderland, in itself, is seen as an area with no distinct borderlines. This is one way to approach the speech/music event as a whole, something that is strongly advocated by Anthony Seeger:

In research on the overlapping of language and music, it is necessary to combine linguistics and (ethno)musicology in order to obtain meaningful results. That this is a growing field of research is evident from the increasing numbers of conferences and workshops that bring researchers together by focusing on this field, as a whole or in certain aspects. Another indication of this is the appearance of special journal issues devoted to this theme.¹⁸

George List, Steven Feld, and Anthony Seeger developed this area of research as ethnomusicologists, with insights into linguistics and verbal art. But there are also examples of collaboration between ethnomusicologists and linguists. Linda Barwick points to an especially strong tradition of such collaboration in the study of aboriginal music in Australia.¹⁹ In studies of this kind, the expertise of ethnomusicologists and linguists of different specializations is brought together to give a fuller description and analysis of the objects studied. This is rewarding for both sides, partly because there are many interrelations between language and music in vocal expressions where the two overlap and partly because so little is generally known about both the language and the music.

As a result of the developments in combining ethnomusicological and linguistic approaches, new methodologies are being tried out. In a PhD thesis in linguistics titled ‘Musicolinguistics: New Methodologies for Integrating Musical and Linguistic Data’, Morgan Sleeper uses different technological resources in three case studies in order to integrate the two approaches. These are his starting points:

At Lund University, Sweden, Kristina Lindell (1928–2005), who was a field linguist specializing in the language, narratives, and general culture of the Kammu people in northern Laos, created a research group in the early 1970s. The ‘Kammu Language and Folklore’ project became a long-lived project under different names.²¹ Among the members were the linguist Jan-Olof Svantesson and the ethnomusicologist Håkan Lundström. Together with other researchers, we (Svantesson and Lundström) worked with the same material from different angles, and we collaborated over the documentation of vocal expressions in the material.²² This background permitted the publication of one person’s repertoire of orally transmitted poems used for vocal expressions, with glossing and translations.²³ This person was Kàm Ràw (Damrong Tayanin, 1938–2011), who grew up in a Kammu village and became a key co-worker in the research project. Collaboration with him led to a dissertation in musicology in 1999, with a focus on his technique of re-creating vocal expressions in the mono-melodic tradition of his home area.²⁴ These results were prerequisites for the ‘Borderland’ project. There was thus a long tradition of interdisciplinary research in ethnomusicology and linguistics at Lund University, pretty much in line with the general development of the field; and from the time the ‘Borderland’ project was conceived, around 2009/10, its researchers have played active roles in international developments.

The performance template

In our approach, the term vocal expression for those expressions in which speech and song overlap serves as a culturally neutral term. It is also chosen as a neutral term in our interdisciplinary research. ‘Song’, ‘recitation’, ‘narrative’, etc. all have a research history and praxis in (ethno)musicology and the various linguistic specializations. Focusing the research on vocal expressions is a way to avoid bias based on existing practices by approaching the object of research as something new that we know very little about.

As regards the composition of oral poetry, Ruth Finnegan discusses ‘re-creation and re-composition theories’ that explain variations in the transmission process.²⁵ Here we use the term ‘re-creation’ to denote the process of the performance of vocal expressions of material that existed prior to the performance occasion, but which has no normalized ‘original version’. In such cases, two or more different versions of the same basic material are thus seen as re-created differently. This kind of process is also described by Francesca Lawson in relation to narrative arts in Tiānjīn, China:

Our assumption is that re-creation, in this sense, can be analysed by isolating performance templates. These templates that ‘relate to the meter of the text or would directly relate to the text itself’ were used by Nigel Fabb in the metrical analysis of poetry.²⁷ It goes back to the ‘delivery instance’ of Roman Jakobson concerning vocal renderings of poetry, defined as ‘the verse line as it is actually performed’.²⁸ Here, the concept performance template will be widened so as to relate to other aspects of language than metre, and to include music as well.

The performance template is a tool used for analysis of a performer’s assumed mental image of a vocal expression, which makes the principles for the interaction of language and music visible. It is applied when re-created vocal expressions are performed and consists of analysis based on the following parameters: melody, rhythm, form, phrasing, initial/final formulae, word variations and lexical tones.

This kind of analysis has previously been applied to certain vocal expressions in Kammu culture.²⁹ One objective of the present research project has been to develop and deepen this approach and, in this process, test the extent to which it can be applied to the material. That material includes vocal expressions that lie close to recitation or rhythmicized speech, i.e. in the borderland between speech and song, though some expressions are more ‘song-like’. Some parts of the material are based on syllable-counting poetry, whereas some consist partly or totally of vocables. The material includes vocal expressions that easily fit the concept of re-creation and involve much variation, sometimes improvisation, while others are composed beforehand and are supposed to be performed anew in the same way.

The collaboration

The collaboration between the seven members of the research team has been accomplished at three levels: collectively, in smaller constellations depending on specific topics, and individually. The work has been coordinated throughout. In recurrent working meetings, theoretical and methodological perspectives have been discussed and developed and the results jointly examined. Most of the researchers involved in this collaboration live in different cities in Sweden, and one lives in Alaska. Physical meetings have therefore often been combined with the use of Skype. Tuttle had an opportunity to spend an academic year in Lund, which increased the possibilities of continuous meetings and in-depth communication.

It soon became obvious that there are many differences between linguistics and ethnomusicology, and also between different branches of linguistics. Sometimes we use different terms for the same thing. In other cases, the same term means different things. In addition, some terms are so specific to the field of research in question that they are not easily comprehensible in other fields. In some cases, we agreed on new terms: vocal expression, vocal genre, melody-centred, tone-centred, etc. Other terms were listed in a glossary explaining how they are used in this book (see Appendix 3). In some cases, it was necessary to adapt terms. For instance, the term ‘phrase’ has different meanings in musicology and linguistics – sometimes they coincide, but not always. When the meaning is not clear from the context, we use terms such as musical phrase or linguistic phrase. These adaptations were necessary in our communication in order to avoid misunderstandings; or, rather, it was often through misunderstandings that such problems were identified.

Most steps in the progression of the research have made it necessary to proceed from documented performances, which involves much basic work. In most cases, there is little or no previous research that tackles these matters in the performances studied here. Regarding language, most of our material is complex even at the most basic level of transcription and translation. In many cases, the performers themselves were not able to fully explain the meanings of words or expressions. Musically, the performances are often very open regarding form, as words are often added or subtracted by the performer and as melodic movement is also often varied in combination with speech intonation and, in some cases, lexical tones. As a result, many problems at a basic level of analysis had to be solved in order to achieve any results at a systematic or higher level.

One major challenge has been the combination of linguistic and musicological research. In this process, we have experimented with different formats for representation of the audio recordings that were studied. This has been done partly to facilitate communication between the researchers, and partly in order to present the working methods and results to the reader. Consequently, this book contains a great many graphs, tables, and notations. The software and their use are explained in Appendix 1. Transcription – whether musical notation or various kinds of graphs – and analysis usually make it necessary to break down the sounding totality into segments, sometimes very small ones. All the performances are, of course, totalities, with a flow of sounds and a direction, so one consequence of dividing them into segments is that the conclusions in each case relate to the whole performance. A choice of the actual recordings used is also made available to the reader; see p. xii in this volume.

Presentations at conferences and seminars have been of great importance in several ways. They have not only set useful deadlines and created opportunities to provide information about our research activities or findings, and to benefit from networking and feedback. They have also given us reasons to spend time together in various constellations, enabling us to collaborate on specific themes and to develop a stepwise progression of our research.

The layout of the book

The researchers have worked as a team when discussing methods and outcomes, and they have also worked on specific topics in various sub-groups. The layout of this book basically reflects the chronological order in which the research developed, even though, in reality, there has been considerable overlapping between the different branches of the research. Different sections form chapters, with their separate authors. Parts of these chapters have been published as articles or as presentations at conferences, but this is the first time that we are putting them together as a whole. In so doing, we have chosen to list 26 case studies as ‘Analysis 1–26’ across the chapters of the book. While the chapters reflect the separate sub-areas of our research, the analyses are intended to reflect totality and continuity. Thus, we attempt to present the results reached concerning each sub-area, as well as the overall results that are presented in the concluding discussion.

As mentioned above, a network of scholars at Lund University has been working on a long-term documentation of the Kammu, an Austroasiatic-speaking people in Laos, starting around 1970 with several research projects targeting different aspects of their language and culture. Chapter 2, ‘Kammu vocal genres (Laos)’, is based on the fact that the material contains a rather large number of vocal genres performed by a single person. This makes it possible to isolate a number of separate performance templates within one cultural context. The Kammu dialect we studied is a tone language with two lexical tones. The material also permits a detailed study of the role of the lexical tones, as well as their realization in the performance of the different vocal genres.

The research on the Native American Athabascan language and the culture of Interior Alaska carried out at the University of Alaska Fairbanks has likewise produced a substantial material over a number of years.³⁰ There is also a link to Lund University that dates back to fieldwork carried out in Alaska in the 1960s and 1970s.³¹ Since the available Athabascan material covers a time period of approximately half a century, it is to some extent possible to relate orally transmitted vocal expressions in newly collected material to older samples. However, there had been no actual collaborations between musicology and linguistics before the present studies were commenced. Therefore, Chapter 3, ‘Athabascan vocal genres in Interior Alaska’, also includes a basic description of vocal expressions. An important aspect of this culture is that of song-making, either by instant improvisation or by carefully planned composition.

These chapters report extensive research that was carried out in order to use the combined approach of musicology and linguistics as well as the performance template as an analytic tool. The three following chapters are case studies which aim to test the analysis method that utilizes performance templates. Chapter 4, ‘Seediq canonic imitation (Taiwan)’, focuses on two related, newly recorded performances. The music of the aboriginal groups in Taiwan has been studied by Taiwanese and Japanese researchers. Our own knowledge was in the area of language.³² It was therefore necessary to develop a musical understanding of the material before searching for performance templates. A special feature of this genre is imitation between performers.

The Akha minority in northern Thailand is geographically rather close to the Kammu in northern Laos, but the two belong to different language families. Chapter 5, ‘An Akha shaman performance (Thailand)’, is a study of a section of one performance chosen from an extensive documentation of the Akha language. Our competence was mainly in Akha language and texts, while no collaboration between linguistics and music preceded this study.³³ This performance is very long, and some sections from the beginning, middle, and end were chosen for analysis.

Chapter 6, ‘Waka and ryūka performances (Japan/Ryukyu)’, uses material from a culture that has had a writing system for a very long time, but where there are still contexts in which the vocal performance of poetry is orally transmitted. The main knowledge of the area was in the field of language, and there had been no preceding collaboration between linguistics and musicology.³⁴ The vocal expressions studied are vocal renditions of short poems: waka of Japan and ryūka of Ryukyu. While the two kinds of performance are very short, with a fixed number of syllables and similar progression of rhythm and pitches, there are differences in their relation to speech intonation.

Map 1 The geographical location of the languages and cultures included in the study.

The source material

The material used is basically of three kinds: field material in individual researchers’ collections or in archives; field material gathered in new fieldwork; and material from publications. As regards Kammu traditions there is an extensive material, stemming from the mountain village of Rmcùal in northern Laos, that has been collected and analysed in the long-term project at Lund University. The data contain a considerable amount of narratives and vocal expressions, including information on transmission and also general linguistic and cultural information. Most of this material exists in the form of recordings, in audio media ranging from open-reel tapes and cassette tapes to digital files, and in visual media ranging from drawings, black-and-white photos, colour photos, diapositives, VHS videos, Hi8 and DV to digital videos. During the project period, we have helped to sort out this material for digitization for a database in the ‘Rwaai: Digital Multimedia Archive of Austroasiatic Intangible Heritage’ project at Lund University, headed by Associate Professor Niclas Burenhult.³⁵

Interior Athabascan tales and myths, including vocal expressions, were recorded in Alaska in 1966 and published by Anna Birgitta Rooth. They were analysed in two studies that also described the context of storytelling in Alaska.³⁶ Descendants of the storytellers are still living there. The source material has been expanded as a result of new information from continuous fieldwork in Interior Alaska carried out by Siri G. Tuttle. Recordings and transcriptions are available at the Alaska Native Language Archive at the University of Alaska, and in publications.³⁷

An extensive material of narratives was collected from aboriginal groups in Taiwan in the 1930s by the Japanese scholars Takuji Ogawa and Susuma Asai; it has subsequently appeared in English translation.³⁸ Recent fieldwork data collected for other purposes include recordings of these and similar narratives from past decades for Seediq, Bunun, and Puyuma. Music of the aboriginal groups has been documented in one study and in recordings. There is a recent study about the present-day musical culture of the Amis minority.³⁹ New material has been gathered by Arthur Holmer as a result of fieldwork periods, particularly among the Seediq and the Puyuma.

The Akha material basically stems from extensive fieldwork in northern Thailand carried out by Inga-Lill Hansson and commenced in the 1970s. There is an extensive recorded material on partly digitized cassette tapes. All these recordings have been transcribed and translated, and there are also detailed and well-organized field notes. This material is all in Hansson’s private archive, which is in the process of being digitized.

For the Ryukyuan study, material in the database ‘Okinawa’s Old Chant’ (Okinawa Cultural Promotion Association) has been used. The information about the Ryukyuan language comes from Yasuko Nagano-Madsen’s own fieldwork there, and her subsequent analysis and publications.

In some of the cases, the ethical considerations have been solved naturally by long-term collaboration between researchers and ‘informants’. In the case of Kammu, Kàm Ràw (also known by his Thai name, Damrong Tayanin) became a co-worker over more than three decades of collaboration with researchers. The Athabascan elders in Minto, Alaska also took an active part in encouraging the researchers to work with the traditional vocal expressions and spread this knowledge. The work with Akha was conducted in collaboration with knowledgeable individuals over a long time period. Many of these co-workers have passed away, and we feel obliged to continue our commitment to their legacy. Part of the Athabascan materials stem from a currently active song-maker and performer, Norman Carlo, who collaborates closely with Tuttle. He has read the text, provided corrections, and approved our description of his work. The work with the Seediq people in Taiwan has taken place in a larger context of preserving and disseminating the Seediq language and culture, with the full support of the speaker community. In the cases where we have used existing recordings, we have, wherever possible, found copyright holders and have their permission to use the material. In the case of the commercial Melodyne software, permission to use the graphs has also been acquired.

The transcriptions

Various methods of transcription of words, pitches, and rhythms have been tested in this study. The choice of method depends on the purpose of the transcription: what it is used for, and what factors are to be communicated. Another factor that influences the choice is how well known the relevant performance is to the researcher(s). In our study, for instance, Western notation was used for the musical transcription of the material in the Athabascan and Seediq sections (Analyses 11–20), basically because we did not know enough about those performance styles to start with. They therefore had to be approached separately as music and as language, with the intention of exploring whether it would be relevant to use the concept of performance templates.

Hence, the use of performance templates as a method leads to different applications depending on the material that is being studied. In some cases, it has been sufficient to transcribe the words and poetic phrases with marks added for lexical tones (Analysis 8). In others, the addition of number notation has been found useful (Analyses 3, 10); see Appendix 2. These forms of transcription have proved practical for the communication of linguistic and musical aspects. We have also experimented with letter notation (see Appendix 1), which may be easier to combine with analysis in the free software Praat and ELAN, as it can fairly easily be written into a tier, like other parameters.

Whatever method is used, each transcription depends on the interpretation made by the transcriber. Musical transcription, whether in musical notation or graphs, tends to approximate pitches musically as fixed tones, i.e. what is interpreted as the main part of a pitch is transcribed, while initial or final movement within a vowel or syllable is normally considered as just that: a sliding motion up to or away from the tone. Thus, a certain pitch is defined as a tone that may be approached or left with downward or upward movements. Changes in pitch during one such tone may be described as embellishments. In sum, the pitch notated is the one that best fits the transcriber’s feeling of musical meaning. There is some degree of subjectivity and also ethnocentricity, since the notation system originated in the notation of Western classical music and since the transcriber’s cultural musical experience may influence the interpretation. However, the result will be a musically meaningful transcription that communicates the musical sound as experienced by the transcriber and can hence be received by a person with a similar background in musicology.⁴⁰ The accuracy increases with the transcriber’s knowledge of the music that is being transcribed.

An analysis based on phonetic expertise will approach the sounding material differently, with a stress on exact measurements. In principle, our experience has shown that music transcription adds the important perspective of transcribing motion while being less stringent with regard to factors that will be more precisely measured in the linguistic analysis. In interdisciplinary work based on musicology and linguistics, these approaches will be combined, and herein lies the strength of our interdisciplinary research. At the same time, this collaboration entails a need to communicate the results of the analyses between researchers of different backgrounds – in this case musicology and linguistics.

In interdisciplinary research involving musicology/ethnomusicology on the one hand and linguistics areas like phonetics or prosody on the other, both these approaches to transcription are necessary for the analysis of the material and for communication between the researchers.⁴¹ As a result, various forms of graphic representations have been found particularly efficient for the analytical aspects, not least because there is software that can produce the underlying graphs. This analysis and the graphs produced are presented in Appendix 1: Software used.

Mechanical transcription has been around for a long time. Around 1900, the kymograph, which had been adapted to linguistic studies and could produce graphs, was used for musical sound as well. Seeger’s melograph was an adaptation for the analysis of music. This was paralleled at Uppsala University, Sweden, where a machine for analysing monophonic music was developed in 1964.⁴² The output was a paper strip that showed frequency (in Hertz) and amplitude along the horizontal time axis. These graphs could be used for detailed measurements, but it was hardly possible to create notations from them, and while they were useful for measurements, they were too detailed for musical notation. Often the two methods were combined: a slightly simplified graph derived from the mechanical transcription would normally be used, with a parallel transcription in notation.

Since then there have been huge developments with respect to note-editing software, as well as to software that converts digital material like MIDI (Music Instrument Digital Interface) into notation within the tempered tuning system, rhythm notation, and harmony developed in the tradition of Western classical and popular music. When it comes to digital transcription of music belonging to oral traditions, very little has actually happened, and there is still no software that can do this properly. From the point of view of music transcription, the software used in this study has many advantages and disadvantages in common with the old technology, with the important difference that it is faster and much more flexible when it comes to adjusting the scales of the vertical and horizontal axes. Another major difference is the possibility of watching the analysis in real time.

Some of the programs tested can export graphs, whereas others cannot. In a case of the latter sort, a screen dump must be made. In either case, it has often been necessary to use additional signs to demonstrate certain factors. Though mechanical transcription of sound facilitates the analysis in many ways, it is still time-consuming and involves considerable manual work.

Transcription is also a matter of defining the degree of precision needed for particular ends. In this study, most of the musical transcriptions are more or less rudimentary, and even a rough mechanical transcription would suffice – it could, whenever necessary, be used as a starting point for a more detailed transcription for use in research that would require such accuracy. The Melodyne software captures pitch and movement in a manner that is intuitive for a researcher with a musical approach, but the musical notation generated by the framework does not work for the transcription needed for ethnomusicology analysis. Praat or ELAN adds a great number of factors for detailed analysis of texts, but no musical notation. In the end, then, we had to use different kinds of software and interpolate between them manually.

There is no perfect software for the research method tried out here; but the results could suggest changes or additions to existing software. The development of mechanical transcription that communicates intuitively with researchers in both linguistics and musicology is an important task for the future.⁴³ It is not so much a question of replacing human transcription with mechanical transcription – this may not even be possible – but more a matter of developing the existing technology as far as possible in a direction that minimizes time-consuming manual work.