A new version of ELAN, the widely used tool for time-aligned annotation of linguistic data, was released today by the developers, Han Sloetjes and Aarthy Somasundaram. One of its major features is a whole new user interface for high-speed transcription. This interface is the outcome of a process of user consultation and usability testing at the MPI for Psycholinguistics led by Mark Dingemanse, Jeremy Hammond, and Simeon Floyd in close collaboration with the ELAN developers Han Sloetjes and Aarthy Somasundaram. In this post we outline the most important features of Transcription mode.

Transcription mode

Transcription Mode is a mode designed to increase the speed and efficiency of transcription work. The interface is keyboard-driven and minimizes UI actions. All annotations of a certain tier type are displayed in a vertical list for easy visual access. Transcription mode brings down the transcription work to the bare essentials: listen, type, listen, type, listen, type.

Note. Transcription mode presupposes that the initial segmentation of the recording is already done. The rationale for this is that the most efficient workflow for transcribing large amounts of linguistic data is a two-step process: first segmenting the recording into turns —also attributing turns to the appropriate speaker— (this can be done in Annotation mode or in the special purpose Segmentation mode), and then transcribing and translating these turns.

1. Setting it up

You can reach Transcription mode via the Options > Transcription mode menu. If you go to Transcription mode for the first time, a Settings dialog will come up. Here you can select the tier types to be used for up to three columns. Note that you select tier types, not individual tiers. This is because Transcription mode displays all annotations on all tiers of a certain type in a vertical column.

For the purposes of this description we will asssume that the user is working with a file that has four main tier types: po (practical orthography), tl (literal translation), tf (free translation), and vb (visible behaviour). Our example file contains tiers of these types for two participants, and the overall tier structure looks like this (tier names in bold, tier types in courier):

• A_po po (practical orthography)
  • A_tl tl (literal translation)
  • A_tf tf (free translation)
• A_vb vb (visible behaviour)
• B_po
  • B_tl
  • B_tf
• B_vb

In our example, we choose the type po (practical orthography) as the first column. We can leave it at that if we just want to work on the transcript. Or we can display up to two additional columns next to the primary one. In our example, we’ll add the literal translation type.

For the second and third columns we can only select tier types that are time-aligned with the first using the stereotype “Symbolic Assocation”. In our example, we can select tl (literal translation) and/or tf (free translation) as second and third columns. We cannot choose the tier type vb (visible behaviour) here, because it is not time-aligned with our primary column.

Having selected the tier types we want, we click “Apply”. Now the chosen tier types are displayed in vertical columns, and the two largest differences from the default Annotation mode become visible: (i) all annotations are displayed vertically (top to bottom) rather than horizontally (left to right), and (ii) columns display all annotations of a certain type. For instance, the po (practical orthography) column displays turns from both speakers A and B.

Note. Transcription mode presupposes that you use linguistic types to differentiate the types of information in your tiers. Thus the linguistic type of your free translation tier should be different from the linguistic type of your main transcription. This is necessary for any serious corpus work anyway — for instance, ELAN’s multiple layer search also relies on this. If you haven’t been using linguistic types yet, consider investing the time to bring your files up to speed. This will not only let you use Transcription mode, it will also allow complex corpus searches and in general make your data more structured. The best way to enforce proper use of linguistic types across your files is to use a template.

2. Using Transcription mode

Transcription mode is built for high-speed transcription work. It plays automatically so that you can start typing right away. You can hit TAB to replay, and if you finish typing you hit ENTER, which brings you to the next annotation, which is played automatically so that you can start typing right away… and so on. Transcription mode boils down the transcription process to the two most essential actions: listening and typing. Once you’ve set it up, you don’t need to worry about anything else.

You can use Transcription mode to do initial transcription of a segmented recording. For this you would use the simplest, one-column setup. You can also use it to work on translations if you already have transcriptions. For this you would display both tier types side by side. And of course you can do the transcription and translation work in one go. For this you would use the two-column setup and check the option to “Navigate across columns”.

The basic philosophy of Transcription mode is to make things as easy as possible for the transcriber. This is why it displays annotations in a table rather than on a timeline, plays automatically on selection, and moves to the next annotation without requiring additional clicks or key presses. It will also silently create child annotations if they don’t exist yet — merely clicking an empty cell (or moving there using the keyboard) creates an annotation and opens it for immediate editing. The user just has to make sure the relevant tiers exist for all participants, and Transcription mode takes care of the rest.

3. Essential shortcuts

Typing and playing back

• ENTER saves the current annotation, moves to the next annotation, and plays this new annotation if the automatic playback option is selected. ^{[Three for the price of one!]}
• TAB plays the current annotation. It acts as a play/pause key, so you can press it again to pause playback, and again to continue playing.
• SHIFT+TAB plays back the current annotation from the start.

Moving around

• UP and DOWN arrows move up and down within a column.
• ALT+LEFT and ALT+RIGHT arrows move left and right across columns (and just because we know you’d try this, ALT+UP and ALT+DOWN also move up and down within a column)
• Remember that ENTER automatically moves to the next annotation. The Navigate across columns setting controls whether you go down within a column or you move across columns (from left to right).

Using the mouse

• Clicking on any annotation activates it for editing. The cursor will be placed close to where you clicked and you edit right away.
• You can also use the mouse to select part of the waveform for playback. TAB will play/pause the selection.
• Right click on annotation will give you an option to jump to the Annotation mode. This will allow you to finely manipulate annotation boundaries and then return to Transcription mode.

4. The settings explained

Below the video signal and above the waveform you find the normal playback buttons (Play, Play selection, Clear selection). (Though recall that you can simply use Tab for quick playback of the full annotation or the selection.) In this area there are two further options:

• screen layout. This option determines whether the media and settings are displayed on the left side or on the right side of the screen. Clicking it flips the screen layout. Default: video, audio and settings on the left.
• loop mode. This option is found to the right of the play/pause buttons. When checked it means that a selected annotation while constantly loop until a new annotation is selected. Default: unchecked.

Below the waveform your find a couple further options that you can use to customize the Transcription mode experience.

• automatic playback of media. This controls whether the annotation is automatically played or not when you arrive at it. Default: checked.
• show tier names. This controls whether tier names are shown within the list or not. If unchecked, colour coding distinguishes different tiers/participants, and hovering over an annotation will tell you the tier/participant name. (There is an additional choice to show colours in the cells themselves or only in the line number column.) Default: checked.
• navigate across columns. This controls what annotation you move to after hitting Enter. If unchecked, you move only within a column (from top to bottom). If checked, you move across columns (from left to right). Default: unchecked.
• always scroll the current annotation to the center. This mode keeps your current annotation always in the middle of the screen. Default: unchecked.

5. Layout and viewing

The layout of the transcription mode is designed to replicate the best aspects of a word processor and a spreadsheet – all the while allowing you access to the time-aligned video and audio signals.

• The video column, which also includes the options and wav form, can be placed on the right or left. Press the screen layout button to toggle the video/settings column from left to right. The video can also be detached for viewing independent of the main window, for instance on a secondary monitor.
• All of the columns are resizable: just mouse click and drag the boundaries to fit your desired widths.
• You can order columns as you please. So once you have established your types you can then reorder them on the screen simply by dragging them to the desired location.
• You can zoom in on the video signal in order to focus a particular part (also available in annotation mode). This works best with HD footage.
• Font size of the columns is variable: you can change this in the settings dialog box.

6. Have (quick and easy) fun

We hope you enjoy this addition to the Elan toolset. It is designed to cut down on the many hours it takes to do detailed transcriptions and we feel that you will find it an indispensable part of your workflow.

We welcome your comments and feedback!

(This post was co-written by Mark Dingemanse & Jeremy Hammond and appears in slightly different form in the help function of ELAN.)

Gérard Diffloth, writing about the paradox of catching ideophones in the wild, notes the following:

Il faut donc guetter les expressifs et les attraper au vol ; mais dans le feu de l’action et de la discussion animée où ils naissent, qui aurait le culot d’interrompre tout le monde afin de pouvoir vérifier une voyelle, un sens, une intention?
— Gérard Diffloth, 2001

Or in English:

We must therefore watch for the expressive and catch it in full flight; but in the heat of the action and animated discussion in which they are born, who would have the gall to interrupt everybody to be able to check a vowel, a meaning, an intention?

Who indeed, we may ask. Interrupting an animated conversation anytime an ideophone flies by is a surefire way to kill any spark of spontaneity.

In good French academic style, Diffloth is highly skeptical about ways to subvert this problem: “Quant au magnétophone, n’en parlons pas, il gâche la spontanéité et il supprime le contexte non-sonore … La caméra est plus balourde encore.” (In English: “As for the tape recorder, don’t get me started, it ruins the spontaneity and removes the non-auditory context. … The camera is even more clumsy.”)

I share a lot of Diffloth’s skepticism, but as the Mawu say (using ideophones), àsi kpoo, ilo kpoo: if you do nothing, nothing happens. Interrupting a videotaped conversation is a lot easier than interrupting “tout le monde”. And using the sequential structure of recorded conversations to get access to participants’ own interpretations of each others’ talk is a extremely valuable complement to post-hoc reflections on meaning or intention. So for all its clumsiness, I prefer the camera.

Building a multimodal corpus of everyday interaction is a job fraught with difficulties (even if cameras have become a lot less clumsy since 2001). But if it is the closest we can currently get to a faithful (if not complete) representation of everyday social interaction, it is infinitely better than nothing.

Note. At the request of Gérard Diffloth, I changed my slightly literalist translation of “tout le monde” as “all the world” to the more common “everybody”.

References

1. Diffloth, Gérard. 2001. Les expressifs de Surin, et où cela conduit. Bulletin de l’Ecole française d’Extrême-Orient 88(1). 261-269. doi:10.3406/befeo.2001.3516.
2. Enfield, N.J., Stephen C. Levinson, J.P. de Ruiter & Tanya Stivers. 2007. Building a corpus of multimodal interaction in your field site. In Asifa Majid (ed.), Field Manual Volume 10, 96-99. Nijmegen: Max Planck Institute for Psycholinguistics. (online)

We’ve been working on this for quite some time, and we’re excited to go live now: the L&C Field Manuals and Stimulus Materials. This is a website providing access to many of the field manuals produced over the years by the Language and Cognition Group at the Max Planck Institute for Psycholinguistics. As the front page explains:

This site contains a bonanza of material for the field elicitation of semantics and and the field collection of verbal behaviour. These are unique resources that have been compiled over nearly twenty years of investigation of under-studied languages by the Language & Cognition Group at the Max Planck Institute for Psycholinguistics. During this period we collectively pioneered the field of semantic typology.

Many entries from these manuals have been circulating informally for years and they have been used by field workers all over the globe. With this archive we offer a centralized, easy to use resource. We’ve started by making available the most recent couple of years. Over the coming months, we will be uploading older manuals and materials, but you can start by checking out the wealth of materials already there — from guidelines on Building a Corpus of Multimodal Interaction in your Field Site to our cross-cultural Synaesthesia Pilot, and from the recent Language of Perception project to the classic Put Project: The Cross-Linguistic Encoding of Placement Events.

Transient Languages & Cultures published a nice post by Peter Austin last month on the question of how much time it takes to transcribe linguistic data. Working under tight time constraints during some recent fieldtrips, I found one way to speed the process up. It still takes an awful lot of time, but here goes.

In my experience, two very important bottlenecks in transcription, especially of conversational material, are (1) initial recognition (what exactly was said?), and (2) writing it down (how quickly can this be written down in the orthography you have chosen?). In my field situation (a Siwu-speaking village in eastern Ghana with few literate and even fewer computer-literate people), I don’t have someone (yet) who could do the actual transcribing, which is usually done directly in ELAN, so I am responsible for bottleneck #2 (getting it into the computer).

As regards the first bottleneck, for a native speaker it is much, much easier to ‘hear’ a fuller form of conversational speech than for a non-native speaker, so it makes sense to get that kind of help for bottleneck #1. Initially therefore, I would sit down with a consultant, play a conversation utterance by utterance (I would have done the segmentation in ELAN beforehand), and have the consultant repeat the speech while I wrote it down. For bottleneck #2 reasons this often meant replaying the same utterance multiple times. I soon realized that this was a waste of time for my consultant, who would patiently repeat two to four times what he already got right the very first time. Essentially I was using him as a tape recorder, rewinding and replaying his careful repetitions to make up for my deficiencies in short-term memory and typing speed! So here is what I did to speed up the process:

1. Presegment the conversation in utterances in ELAN.
2. Go through it a first time, utterance by utterance, and record the consultant’s carefully pronounced repetition of each utterance.
3. Get the results of 2 (tens to hundreds of small audio files) onto the computer, batch rename them if needed (I always include at least date and speaker ID in the filename), and put them in a media library that allows quick and easy file selection and playback (I used WinAMP).
4. With that material in hand (essentially slurred and non-slurred versions of utterances) do a first rough transcription. No need to bother the consultant during this time-consuming process.
5. Mark all the problematic points and get back to the consultant for a final check.

During my last fieldtrip, my main consultant had to study a lot for his final exams. The above modus operandi turned out to be a win-win situation. He provided me with carefully pronounced utterances (we would get so good at this that we nearly achieved real-time speeds), and I provided him with a place to study in my improvised office on a shadowy compound. So after an initial intense recording session, he would spend the afternoon learning while I would put my headphones on to do the first transcription, which was a breeze with his crystal-clear repetions (which included whistled tone patterns if needed!). Hours later, the final check would be a matter of minutes and we would be ready to enjoy our well-earned fufu in palmnut soup with goat meat — surely the best part of the whole transcription process!

Bulbul posted an interesting anecdote in a comment on one of my earlier posts:

On my way home today, I took the scenic route, through the old town, where the Weinachtsmarkt is in full swing with Christmas lights glowing, Glühwein flowing and all that jazz. As I was trying to get through the crowds, I noticed a black gentleman standing next to one of the stalls obviously admiring something and talking on the phone in a language I could not immediately identify.

And just as I passed him, he said “You know” and then something I would transcribe as “ŋɛrɛrɛrɛ” followed by a laugh. “I bet this ŋɛrɛrɛrɛ is an ideophone” I said to myself and immediately started wondering whether the person on the other end truly understood what was being conveyed – in other words, whether that “ŋɛrɛrɛrɛ” was a word with a shared meaning. Now I know better – assuming I was right in identifying the word as an ideophone, of course.

I still don’t know what language that was (I’m guessing Yoruba based on a few words I might have heard), so do ideophones really stand out that much that even a non-speaker can identify them as such?

Decide for yourself

So that’s today’s question: do ideophones really stand out that much? This is something you can only decide for yourself. Here are three examples from Siwu. They come from my corpus of everyday discourse and represent the three most common ideophone constructions. These three constructions account for 88% of 230 ideophone tokens in the corpus; the examples thus can be said to be typical of ideophone usage in day to day conversations in Siwu.

I will not transcribe them at this point; I just want you to listen.

Example 1

[See post to listen to audio]

Example 2

[See post to listen to audio]

Example 3

[See post to listen to audio]

Well do they?

Now you’re in the position to answer bulbul’s question: do ideophones really stand out that much that even a non-speaker can identify them? The answer —mine at least— is yes. If you are a hearing person, I’m willing to bet you had no trouble at all identifying the ideophones in the above three sound samples.

Before I give you the transcriptions, it is worthwile to ponder for a moment why ideophones stand out like this. I’ve hinted at this on other occasions, for example yesterday’s ditty on vivid suggestion, a post on Feedburner’s Zap! Pow! Kraaakkkk!, and the last ideophone proeverij; and also in a recent paper, where I wrote:

As marked words, ideophones set themselves apart from the surrounding linguistic material; as a likely locus of performative foregrounding, they stimulate emotional engagement; as depictions, they supply vivid imagery and recreate sensory events in sound, inviting the listener onto the scene as it were.

So ideophones stand out for a reason: to attract attention to themselves as words qua words, to mark themselves as depictions in a stream of descriptive material. Let’s suppose the gentleman overheard by Bulbul was indeed using an ideophone. Standing at the Weinachtsmarkt, he was attempting to share a vivid image of something he had in mind with the person on the other end; to do so, he needed to signal that what followed ‘You know’ was different somehow from bland referential prose; and this he did (unwittingly for sure) by performatively foregrounding ‘ŋɛrɛrɛrɛ‘.

Of course it’s a bit flaky to draw conclusions on the basis of a couple of syllables overheard on a Weinachtsmarkt. Was it Nigerian Pidgin, which we know has lots of ideophones (Faraclas 1996)? Was he codeswitching? Was he perhaps simply stuttering? There’s no way of knowing. That’s why I gave the Siwu examples, which come from an extensive corpus of everyday social interaction. Want to know what those mean? Click ‘Show’ below to check it out.

[[Visit blog to check out this spoiler]]

References

1. Dingemanse, Mark. 2009. Ideophones in unexpected places. In Proceedings of Conference on Language Documentation and Linguistic Theory 2, ed. Peter K. Austin, Oliver Bond, Monik Charette, David Nathan, and Peter Sells, 83-97. London: SOAS, November 14.
2. Faraclas, Nicholas. 1996. Nigerian Pidgin. New York: Routledge.

We’ve just launched the web page for a project we’ve been working on in the Language & Cognition group: Synaesthesia across cultures. The most exciting part of the project is the second iteration of a pilot we’ve developed for cross-cultural field research on the forms and prevalence of synaesthesia. In contrast to online tests (e.g. synesthete.org), our pilot uses low tech methods so that it can be used even in non-literate communities and in remote fieldsites where there is no electricity (let alone internet access).

The pilot is currently being taken to L&C fieldsites across the globe, but we’re also welcoming external collaborators. If you’re interested, check out Synaesthesia, a cross-cultural pilot or our project page.

References

1. Majid, Asifa, Tessa van Leeuwen, and Mark Dingemanse. 2009. Synaesthesia: a cross-cultural pilot. In Field Manual Volume 12, 8-13. Nijmegen: MPI for Psycholinguistics.  
2. Majid, Asifa, Tessa van Leeuwen, and Mark Dingemanse. 2008. Synaesthesia: a cross-cultural pilot (first version). In Field Manual Volume 11, 37-41. Nijmegen: MPI for Psycholinguistics.  

Morning clouds in Kawu

It is no news that some humans say that man is an animal, especially not this year. But wouldn’t it be rather more interesting if another member of the animal kingdom would weigh in on the matter?

It happens in Kawu, where I am right now for fieldwork (hence the silence on this blog). The call of the ìsakpòlò bird, singing in the early morning, perfectly resembles the tonal contour of the following Siwu phrase:

màturi bra màbɔi

people make animals

‘People are just animals’

Recording of the call:

[See post to listen to audio]

Whistled imitation and pronunciation of the Siwu sentence:

[See post to listen to audio]

The first time I became aware of this bird was when one of my assistants said, ‘That bird is insulting us.’ Next time I’ll try to provide a picture of this wonderful bird. Meanwhile, there you go. Man is an animal. You didn’t hear this from me. You heard it from the ìsakpòlò bird.

Illustration © Imai Lab 2006

A 2005 study suggests that Japanese ideophones of laughter activate striatal reward centers in the brain, but I think the results should be treated with caution. And Japanese gargle with salt water regularly as a prevention against the common cold; they even have an ideophone for it (but so do we, don’t we?). That’s giggles and gargles today. Let’s tackle the giggles first.

Ideophones that make you feel good

A 2005 brain imaging study suggests that ideophones for laughter, but not nonsense syllables, activate reward areas in the brain. Here is the abstract:

The neurobiological reward components of laughter induced by words were investigated. A functional magnetic resonance imaging-based brain imaging study demonstrated that visualization of mimic words and emotional facial expressionwords, highly suggestive of laughter, heard by the ear, significantly activate striatal reward centers, including the putamen/caudate/nucleus accumbens, prefrontal cortices, dorsal anterior cingulate cortex and the supplementary motor area, while non-mimic words under the same task that did not imply laughter do not activate these areas in humans. We tested a specific hypothesis that implicit laughter modulates the striatal dopaminergic reward centers by image formation of onomatopoeic words implying laughter and successfully confirmed the hypothesis. [Osaka & Osaka 2005]

Since ideophones have been claimed to somehow establish a more direct link between sounds and sensations than other words, brain imaging of ideophone production and comprehension is an exciting research area. Basically, the finding of Osaka & Osaka is that Japanese ideophones for laughter activate striatal reward areas, just like real laughter and other pleasurable activities do. The ideophones used are ‘ghera-ghera’ (strong laughter), ‘nikoh-nikoh’ (strong), ‘kusu-kusu’ (medium), ‘niyah-niyah’ (medium), ‘herahherah’ (weak) and ‘nitah-nitah’ (weak) (p. 1622, romanization by the authors).

But is it ideophony?

That is an interesting result, but I wonder: does the effect really occur because the words are ideophones, evoking the experience of laughter through their sound-symbolic form and imagistic meaning? Or could it simply be due to the fact that the words have to do with laughter? We can’t tell, because the baseline comparison is not with non-ideophonic real words but with nonsense words (called ‘nonsense phonemes’ by the authors). Since non-ideophonic laughter-related words have been kept out of the comparison, we cannot be sure that ideophony (onomatopoeia/mimesis) is causing the effect, although this is what the authors would like to claim.

There is some reason to think that embodied semantics might be enough to induce such effects; think for example of the brain imaging studies showing that certain sensori-motor cortex areas not only upon tactile stimulation of the body part in question (e.g. the hand), but also during the processing of body part terms and verbs implying them (e.g. hand, grasp; Rohrer 2001). So the question is: would the effect found by Osaka & Osaka also occur with non-ideophonic laughter-related words in Japanese? For comparison, it would also be good to have a not so heavily ideophonic language thrown in. Would the English verbs ‘giggle’ and ‘laugh’ also trigger the effect? Sound-symbolic ‘giggle’ moreso than ‘laughter’? Then things start to be really interesting.

A related problem is the claim that ‘image formation of onomatopoeic words’ plays a role in the effect. Once again this would be an interesting claim to test; native speakers of ideophonic languages often report that ideophones evoke vivid images. But in this study it remains an untested background assumption. The way the experiment is set up doesn’t seem to allow for any inferences about it. For all we know the effect might just be due to an association between the sound and the experience of laughter; it is not at all obvious why image formation would come in. One way to approach this issue would be to do imaging studies of ideophones that don’t imitate sounds, but other sensory events.

[Update: Kimi Akita notes that the stimuli, described by the authors as 'laughter onomatopoeic words' (p. 1622), actually mix sound-imitating ideophones (geragera and kusukusu) and movement/visual pattern-imitating ideophones (nikoniko, niyaniya, herahera, and nitanita). It doesn't really help that all of the results are averaged. I might add that Japanese itself does distinguish the two groups by the terms giongo and gitaigo, even though to a non-native speaker the actual categorization in this case isn't obvious (I would've grouped herahera with geragera, and I wonder what kusukusu laughter sounds like...).]

Gargles

So much for the giggles. What about the gargles? The gorgeous gargling girl above is one of the stimuli used by Prof. Mutsumi Imai in a study of child-directed speech in Japanese. One of her findings is that when describing scenes like this to their child, mothers will tend to use more mimetics (ideophones) than when they are describing the same scene to an adult.

I’m planning to do a pilot in Kawu using prof. Imai’s stimuli, and one question is to what extent the original material would be usable in a West-African context. The idea is that the stimuli can be described using ideophones. Since most of the illustrations are simple events (jumping down, jumping across, throwing, rolling sth. up) I think they should be usable by and large. Perhaps the skin color will have to be changed — I prefer stimuli to be as culturally inconspicuous as possible — though the question is whether that really would affect what we’re after.

However, the one stimulus that I think won’t be familiar is the gargling one above. In the Japanese context, it is meant to elicit the ideophone garagara, probably in the light verb construction X suru ‘do X’. But in Kawu, the scene isn’t very recognizable. People usually drink from calebashes (or their hands), though whites are known to prefer cups — so my guess is that the girl would simply be seen as drinking. Since gargling is not a culturally salient event in Mawu society, I don’t think people would readily think of it, even if there happens to be an ideophonic word for it.

The Japanese ideophone for gargling is garagara. Interestingly, Kimi Akita tells me that “Japanese mothers tell their kids to pronounce “garagara” while gargling. This is because the articulation (especially, that of the velar consonant) of the mimetic is believed to help kids gargle successfully.” Now that’s an interesting intermingling of habitus and embodied meaning. I tried this (without any appreciable gargling experience) and nearly choked. This gives a whole new meaning to the phrase “Embodied semantics is a killer idea”!

References

1. Imai, Mutsumi, Sotaro Kita, Miho Nagumo, and Hiroyuki Okada. 2008. Sound symbolism facilitates early verb learning. Cognition 109, no. 1 (October): 54-65. doi:10.1016/j.cognition.2008.07.015.
2. Osaka, Naoyuki, and Mariko Osaka. 2005. Striatal reward areas activated by implicit laughter induced by mimic words in humans: a functional magnetic resonance imaging study. Neuroreport 16, no. 15 (October 17): 1621-1624.  
3. Rohrer, Tim. 2001. Understanding through the body: fMRI and ERP investigations into the neurophysiology of cognitive semantics. Talk presented at the 2001 International Cognitive Linguistics Association, Santa Barbara: University of California.

P.S. Check out the wonderful bibliographies compiled by Kimi Akita:

1. Akita, Kimi. 2009-02. A Bibliography of Sound-Symbolic Phenomena in Japanese. Electronic ms, Kobe University.
2. Akita, Kimi. 2009-02. A Bibliography of Sound-Symbolic Phenomena in Other Languages. Electronic ms, Kobe University.

The Winners and Finalists of the 2008 AAA Photo Contest are now available in a Flickr gallery. The photos are really beautiful — I’m honoured that one of my submissions is featured among them (and happy that Siwu ideophones are getting some press!).

Click on a photo in the slideshow below to show the author and the caption; or go directly to the slideshow on Flickr.

Edit: The semifinalists are now online, too: Flickr gallery.

Women performing a funeral dirge in Akpafu-Mempeasem

Funeral dirges (sìnɔ in Siwu) are a special genre of songs to be sung during the period of public mourning preceding a burial. The musical structures of these dirges and their place in the larger context of the funeral have been described in some detail by Agawu (1988) and before him by the German missionary Friedrich Kruse (1911); however, the linguistic aspects of the genre have not received any attention so far.

The funeral dirge below was recorded August 17, 2007 in Akpafu-Mempeasem, Volta Region, Ghana (along with six other dirges). It was transcribed and translated with the gentle help of Reverend A.Y. Wurapa.

[See post to listen to audio]

The Siwu is beautifully economic in expression. It contains only two verbs: pia ‘be (with)’ and ɖe ‘be (existential)’. The sɔ that is translated as ‘said’ is actually a quotative complementizer. An English translation cannot do without marking tense, but in Siwu, the poem does not contain any tense or aspect markers, being set in an aorist-like default that can be interpreted as recent past or present.

Some of the poetic devices at work here are lost in translation. One is the focus construction which emphasizes the content words in the last four Siwu lines (‘mourning it is; tears it is; sadness it is; graves it is’). Another is the fact that these content words belong to four different grammatical genders in Siwu: the first is an noun in KA with locative connotation, the second a liquid/mass noun in MI, the third a singular noun in I, the fourth a plural noun in A. I’m not sure whether this pattern is as striking to native speakers as it is to me, but note that the gender is reinforced by the agreement morphology on the ‘be’-verb (ka-, mi-, i-, a-). One could think of it as a case of ‘subliminal verbal patterning in poetry’ (Jakobson 1980).

By fronting the content words and by presenting all four of them in the exact same frame, the dirge forces the reader to meditate on the inevitable consequences of being surrounded by mortality. We may think we’re lucky to have company, but in the end it turns out to be mourning, tears, sadness, graves. The enumeration of closely related tropes is a common technique in the funeral dirges of the Mawu.

A Dutch translation

It is difficult to approximate the beauty and subtlety of this piece of poetry in another language. Nevertheless I have tried my hand at composing a translation in Dutch, my native language, if only because I am intrigued by the subtle interplay of words and grammar in this poem. This translation, then, is a modest attempt to translate not just the words but the terse form-feel of the original. I realize it will be difficult for non-native speakers of both Siwu and Dutch to judge whether the attempt has been successful, but I do provide some explanatory words below.

The first line stays quite close to the Siwu (“I said (to myself), ‘people are with me’”). A less literal translation would have been Ik was blij dat ik mensen had ‘I was glad I had people’, but in the original the expression of contentness remains only implicit, so I felt making this more explicit was not necessary. The inflected verb form dacht ‘thought’ (past) in the first sentence also establishes the temporal setting.

Contributing to the terse style, the Dutch verb blijken ‘turn out (to be)’ nicely packages the Siwu sêgbe ‘not knowing’ together with the be-verb. Dutch verbs inflect not just for tense but also for number; hence the bleken/bleek alternation in the last four lines. To put this fact of grammar to poetic use, I have exchanged ‘crying’ for another word: afscheid. This makes it possible to couple the singular/plural alternation with two different rhyme patterns: Droefheid and afscheid are end rhymes, while tranen and graven are linked by assonant rhyme. The effect is an aesthetically pleasing ABAB structure in which the members of each pair agree in rhyme type and number. In the original, àsekpe ‘graves’ is the final word, but in the Dutch version it isn’t; it somehow doesn’t sound quite right to end with the plural graven; besides, afscheid (parting) is a very appropriate word to end the poem.

The future of dirges in Kawu

Speaking of parting, it is only rarely that dirges are heard in Kawu nowadays. Two factors are contributing to their decline: firstly the fact that many churches discourage their use, preferring edifying hymns instead. The reason behind this, I am told, is that the dirges reflect a pre-Christian worldview and as such are to be eschewed by true Christians. A second factor has been the coming of electricity to the villages halfway the nineties, which has led to loud music taking the place of the dirges during the wakekeepings. Elsewhere I wrote that “culture is a moving target, always renewing and reshaping itself”, yet at the same time I can’t help but lament the imminent loss of such a rich vein of Mawu culture.

However, during my last fieldtrip there were some signs of a renewed interest in the genre. For example, one pastor told me that he had been reconsidering the rash dismissal of the dirges by his church. Realizing how important the dirges had been in containing, orienting, and canalizing the feelings of loss and pathos surrounding death, he felt that the Christian hymns did not always offer an appropriate replacement. Another hopeful event was that I was approached with the request to help record a great number of dirges in Akpafu-Todzi in August 2008. This was not just to record them for posterity (although this was part of the motivation), but also very practically so that they could be played at wakekeepings. I gladly complied with this wish of course. The result is a beautiful collection of 42 dirges, sung by eight ladies between 57 and 87 years of age. The first time the dirges were played at a funeral they sparked a wave of interest.

Next in this series: a discussion of parallelisms in Siwu funeral dirges.

References

1. Agawu, Kofi. 1988. Music in the funeral traditions of the Akpafu. Ethnomusicology 32, no. 1: 75-105.
2. Fox, James J. 1991. Our ancestors spoke in pairs. In Explorations in the ethnography of speaking, ed. Richard Bauman and Joel Sherzer, 65-85. 2nd ed. Studies in the social and cultural foundations of language 8. Cambridge: Cambridge University Press.
3. Jakobson, Roman. 1980. Subliminal Verbal Pattering in Poetry. Poetics Today 2, no. 1a, Roman Jakobson: Language and Poetry (Autumn): 127-136.
4. Kruse, F. W. 1911. Krankheit und Tod in Akpafu. Der Anscharbote, October 29.