Although photo is still not nearly as widely used, video is being more and more used as a medium by linguists in general and language documenters in particular. (Probably this should change, since, as Mark Dingemanse pointed out to me “Good photo equipment is sometimes thought of secondary importance, but from an anthropological linguistic point of view essential”.)

So I too would like to do video recording for language documentation. While I was researching the required technical equipment for my work (a never ending process…) I was looking to optimize the use of our budget and the technical results of our documentation. While traditionally, people use camcorders for making video (which usually means prosumer or even just consumer devices as opposed to professional devices due to the small budgets in science in general and humanities in particular), I learned that an increasing number of people (mostly independent film makers, who also happen to be on a short budget) have been using video enabled Digital Single Reflex Lens photocameras (DSLR) to record video at quality levels comparable to professional video cameras as used in the film making industry. Ever since I have become more and more infatuated with the idea and have been researching whether it’s feasible to do the same for Language Documentation work. Mind you, that this is a theoretical discussion based on research on the internet and discussions with many people, not actual experience using or operating those devices.

Discussions on differences between DSLRs and camcorders/videocameras can be found in numerous places, such as:

• http://www.jiscdigitalmedia.ac.uk/stillimages/advice/digital-cameras/
• http://www.luminous-landscape.com/reviews/camcorders/cam-cam.shtml
• http://www.aibal.com/video-dslr-vs-camcorder/
• http://www.videomaker.com/community/forums/topic/dslr-vs-prosumer-camcorder
• http://www.hdwarrior.co.uk/2011/04/04/dslr-v-video-where-are-we-now/

Simply put, the advantages of DSLR over camcorders are basically:

• DSLR cameras have exchangable lense systems, while camcorders have a fixed lense. Therefore, dedicated lenses with a variety of optical characteristics are available for different recording situations.
• DSLR cameras have much larger and therefore higher quality optical sensors, resulting in better general quality, specifically a claimed better performance in low light conditions and a more shallow depth of field. There are numerous examples of video recorded on DLSRs and nowadays even entire Hollywood films are being made using DLSRs.

So using a DSLR we could do both Full HD video recording, in better quality than any sub-professional camcorder (start counting your money for that) and professional quality still photography. So instead of running around “the field” (whatever that is supposed to be anyways) with two devices, you get to carry only one device – Less devices might mean less luggage and therefore weight. Of course there are camcorders nowadays which do still photography, but supposedly at a quality worse than even low end point-and-shoot cameras.

But, there might also be disadvantages to this approach:

• DSLR cameras might require more accessories, such as a preamp (see below) or different lenses.
• Most DSLR cameras were not made in the first place to record video but audio, so the handling is very different.
• Having only one device, and being on a restricted budget could mean that if your equipment brakes, you have no funds left to buy a replacement. Of course the same can happen with a camcorder, and DSLRs have more modular components which can be replaced, such as the lenses, therefore again making replacement or repair simpler and cheaper.

Some people have discussed pros and cons of DSLR usage from a general perspective, but Language Documentation has specific requirements of course, differing from general film making:

• From a technical point of view, high quality audio recording is expected since the foremost medium of language being usually documented is speech. Unless we are documenting non vocal modes of communication, such as gesture only, I guess there is no point in having good quality video without (good quality) audio. Especially for analysis of particular phonetic nuances, it is extremely important to make high quality audio recordings (for a detailed discussion, see below). Supposedly, audio recording quality of DSLRs as opposed to prosumer (and consumer?) camcorders is inferior.
• From a practical point of view, usually language documentation projects lack manpower and/or technical skills to handle multiple complex technical devices, while conducting interviews. That is, while asking questions with a specific goal or discussing one or the other linguistic or cultural practice, we need to pay attention to the content of the discussion, and don’t really have that much time, to adapt audio/video recording settings in a changing environment or handle such equipment, e.g. to shift focus from one speaker to another. Of course with a non-participatory approach, this is much less of an issue. This is supposedly another factor speaking against DSLR usage as opposed to a camcorder, since the former, due to their physical shape and due to their lack of some features such as (video) image stabilization or autofocus, are said to be more difficult to use.

Obviously, at some point technical aspects become negligible, especially taking into consideration that in field recording situations, we usually are unable to control the environment. The surroundings such as lighting, noise and other factors of recordings are uncontrollable unlike in a studio-recording situation. There are differently balanced views to what level of technical quality is needed for Language Documentation: While David Nathan from ELDP seems to argue for the best possible quality of audio recordings and devices used, Paul Trilsbeek from DoBeS seems to argue, that technical differences stop making a difference at some point (albeit talking about software, not hardware) from the perspective of Language Documentation Archiving. I expect either would agree though that even in a field recording situation a certain level of quality should be aimed for. While software aspects such as e.g. container, codec, compression, bitrate, and sampling rate do play a role in recording quality, hardware aspects such as e.g. noise, directionality, and practical aspects such as a knowledge of technical factors and skills in operating devices are even more important – to put it simple:

1. You can get good quality recording from mediocre quality devices, due to technical skills of operator, resulting in optimal settings
2. You can NOT get good quality recordings from bad quality devices, due to a lack of technical abilities of devices used
3. BUT, you can also get bad quality recordings from great quality devices, due to a lack of technical skills of the operator, resulting in pessimal settings AND/OR due to the recording environment

Specific technical factors possibly speaking against the use of a DSLR for Language Documentation purposes include:

• Supposedly low quality preamps – This is disputable in my eyes: Germany’s most renowned computer magazine c’t just recently did a review comparing different types of cameras for video recording and the Panasonic DMC-GH2 (very similar to a DSLR – a Micro Four Thirds (MFT) if you want to become technical) achieved better test results than all devices but a Sony HDR-CX 690 prosumer camcorder, to which it achieved comparable results (criteria included audio frequency response). There are different work-arounds even it it was the case: There are external preamps , such as the Juiced Link CX231, which is a small box (which can be attached to the bottom of the camera) enabling a high quality pass through of audio.
• Low quality built-in microphones and no connector for external microphones. Low quality mics might be the case, but several models such as the Panasonic DMC-GH2 or the Canon EOS 600D do have external microphone connectors. When it comes to low quality built-in microphones – well also when it come to audio recorders, quality external microphones are usually better than integrated ones, so you should just attach an external microphone to it.
• In either case you could always, record audio separately on a dedicated audio recording device and synchronize it to the video after recording (which would have the benefit of having two audio recordings of the same event, in case something goes wrong). There is a review of audio recording using combinations of DSLRs, external preamps, audio recorders and mics here.
• DSLRs are heavier and bulkier than camcorders and are not made for handheld video operation. But in any case you shouldn’t do handheld video recording with a camcorder either and already do require a tripod. And I expect speech events to have less movement – after all we are usually not documenting events with fast movements such as sports.
• Mostly due to European tax regulations, DSLR cameras are limited to 29:59 mins of continuous video recording. I did a brief survey among some 5-10 colleagues who have conducted video recordings in Language Documentation, and most said they almost never have had to video-record speech events lasting more than 1-10 mins. I figure this might be dependent on the field situation and the specifics of the research conducted, but it seems generally questionable whether we need more time. Also, there are firmware modifications (Should you make use of any of these, I take no responsibility. Theoretically, these modifications could harm your devices or void the guarantee, although I am not aware of any cases where this actually has happened), which, besides yet raising the quality of the recordings of such cameras, sometimes override this limit. These include:
  • PTool for Panasonic, Examples of results here
  • Firmware modifications for Pentax
  • Magic Lantern for Canon
  • CHDK for Canon
• Supposedly DSLRs are more difficult to operate. Well, the more dedicated the equipment, the more complicated the handling gets. This should be a personal choice, depending on how tech-savvy the user is.
• Also, it has been claimed that DLSRs are prone to overheating, when recording video. Looking at what users are saying, it seems that newer models do not really suffer anymore from such issues, but then again, some of us are working in climatically more extreme conditions.

Taking everything into consideration, Micro Four Thirds cameras (due to their reduced size and weight) such as the Panasonic DMC-GH2 (due to it being particularly designed and marketed for video recording), combined with either a dedicated microphone for video recording (such as the RØDE VideoMic) or a more general purpose microphone (such as the RØDE NTG-2) seem to be a quite good choice, offering superior quality and comparable pricing to a video set-up with prosumer camcorder, given that you do not intend to film considerably long scenes or scenes with fast movement. But other (true) DLSRs would also work well. If audio quality is an issue, either an external audio recorder, or even a preamp should be used. That way, such a setup could produce significantly better recordings than any camcorder, given the operator knows how to handle it. But it is very much dependent on your actual aims and circumstances of research. For a more definite albeit general verdict, I would need to test such a set-up myself (which I might end up doing). Of course, any comments and particular experiences with a comparable setup will be most welcome.

I received an calendar invitation for Google Calendar from Facebook in my Google Mail inbox. Whenever I tried to import the invitation to the attached Google Calendar, I always received a 414 error message:

Request-URI Too Large
The requested URL … is too large to process.

I don’t know and care about the technicalities. Only work around was a Google Chrome Browser extension, called Google 414 Fixer by zeroasterisk:

https://chrome.google.com/extensions/detail/kpkldbhofkekfkngljfbknkigpediafn/

Thanks to zeroasterisk for fixing other peoples mistakes

If anybody is using OneNote 2007 in a Microsoft Windows Vista 64 bit enviroment (possibly also Microsoft Windows XP 64 bit) they will notice that the virtual printer is missing, which enables users to send print-outs from any program with printer capability to OneNote. This issue has been discussed by the OneNote developers and widely criticized by OneNote users. This is due to the lack of 64bit support of the Microsoft Office Document Imaging (MODI) component.

This feature is based on OCR’ing from a .tiff file if I understand it right. So basically you run OneNote with a command line switch pointing it to any .tiff file, and OneNote will automatically OCR this file and add it to the unfiled notes section. So we need to create a .tiff file and automate the process of opening OneNote with the correct command line switch. A long time ago I saw a solution with a commercial virtual printer, similar to PDF creation solutions, by Niels Kaiser. So I thought the same should be possible with the open source and more importantly free PDF Creator. I have been using this solution for maybe a year now but never got around to documenting it. So here it goes:

1) Install PDF creator with the command line switch /UseINI, so you can use different .ini configuration files.

2) Create a new printer using any printer driver installed.

2.1) Launch the properties dialog of the newly created printer. (If you got UAC running, you’ll need to do it as Administrator)

2.1) Go to the drivers section and select PDF Creator as driver.

2.2) Go to the ports section and select PDF Creator.

2.3) Click configure below the ports selection and under command line switches put the following

-PPDFCREATORPRINTER -OptionsFile”E:\Eigene Dateien\shared\PDFCreator\PDFCreator_onennote.ini”

Of course, you can change the path to point to wherever you want to put your .ini file containing the configuration for PDF Creator.

3) Create an .ini file in the path specified. I suggest copying the default “PDFCreator.ini” file, found in your user folder (“C:\Users\SOMEONE\AppData\Roaming\PDFCreator”) to another folder.

3.1) Open the .ini file and change the following properties:

AutosaveFilename=onenotetemp
AutosaveFormat=5

…

RunProgramAfterSaving=1
RunProgramAfterSavingProgramname=C:\Program Files (x86)\Microsoft Office\Office12\ONENOTE.EXE
RunProgramAfterSavingProgramParameters=/insertdoc “<OutputFilename>”

…

AutosaveDirectory=E:\Eigene Dateien\Dokumente\
LastsaveDirectory=E:\Eigene Dateien\Dokumente\

Of course you should change the paths specified as needed. Alternatively you can copy and paste my config (I could only attach as .doc) into your .ini file, provided you have the same paths as me.

If you are looking for a different solution to this particular problem, you could go stick to a 32 bit OS or try the XPS solution put forward by David Rasmussen, one of the OneNote developers long after the release. I for one have never tried his solution since I already had a working one.

Of course you can always try different products, and the biggest competitor to my knowledge is Evernote. I fiddled with it some time ago, but prefered OneNote for the tight integration with other Microsoft products, namely Internet Explorer and Outlook, though Evernote does have some other advantages over OneNote. You can find a more detailed comparison here. But recently I’ve been looking into Evernote again, mostly because of their impressive iPhone application. The only comparable solution to accessing and using my notes on the go I found for OneNote was MobileNoter, which, for me, simply didn’t work – but maybe it will work in your system.

PDFCreator_onennote

Another inspiring talk was the keynote speech by Gratien Atindogbé, who was talking about language documentation. I remember him saying that he himself, as an African, was sometimes striving to see the practical value and effects of his work. I enjoyed very much his pragmatic and realistic approach, where he advocated basically doing more P&R for the field of African linguistics, to the extent of even going into politics oneself, which caused the whole audience to laugh. I for one, thought that this formed part as well of what Atindogbé called not being an “arm-chair linguist” – We do need more researchers, which take an active stand representing their field of studies, explaining its practical value for society and implementing realistic solutions, which are based on their research. But since all this is dependent on financial and political support, it seems to me that we should accept that reality and deal with it, which I saw as the main idea behind Atindogbé’s suggestions.

The most entertaining talk was certainly by Philip W. Rudd, on the status of Sheng, a youth language of Kenya. This seems to be based on his PhD dissertation. He already had my attention as he introduced himself as “My name is Philip W. Rudd, please DO call me Phil”. It was also very interesting to see how much he relied on historical data to supplement his linguistic theories on the status of Sheng as a mixed language, the results of which he commented as “Folks…I think we’ve got a language going on here”. I have never seen one person combine historical and linguistic data the way he did, though interdisciplinary work is wide spread in African studies.

The best presentation I attended from a didactic and graphical point of view was the one by Mark Dingemanse about Ideophones. It was so good that I had to go over there and ask him how he did some of the things he did – I never knew that Powerpoint could do that.

But to me the most interesting talk was by Justus Ch. Roux of Stellenbosch University, South Africa, titled “Implementing high level Xhosa text-to-speech synthesis in mobile learning systems“. He demonstrated a mobile phone application for usage in South Africa. Since South Africa has 11 official languages, he explained that there are lots of everyday problems, since no one ever speaks all 11 languages. Accordingly, translation is a daily process for everyday and particularly administrative purposes. Together with a group of researchers he is developing a mobile phone application to deal with this language issue.

What the application does is basically offer a graphical user interface (GUI), which will provide automated text based translation and text to speech synthesis. He mentioned that lots of people use SMS text message and that these, for sake of simplicity, are usually written in English, independent of the readers languages. If the recipient is not fluid in English, he can always pass his cell phone to someone asking that person to translate. So what the software does is take any given text and translate it into another languages. Now what if a person is illiterate as well? Well, in comes the second level of optional speech output, where the user clicks on the written text on the touch screen of his cell phone and the text is read aloud. Also I think they consider (it was unclear or I can’t remember) implementing a third level using the camera features of cell phones (which they nearly all have nowadays) to make a photo of any kind of text, which in turn is optically recognized for translation. So we are talking about a combination of three technologies here: Optical character recognition (OCR), Machine translation (MT), and Text-to-speech synthesis (TTS). Fascinating, isn’t it?

Mr. Roux, being a phonologist, mostly discussed the phonetical aspects of doing a proper text-to-speech realization of a tone language like Xhosa, which was his share of the project. He mentioned how difficult it was getting a phonetical text-to-speech realization to sound right or natural to a hearer. I immediately thought of Auto-Tune, despite it’s death being proclaimed by Jay-Z. I wonder if phoneticians ever thought about using it in speech synthesis of tone languages, since Auto-Tune does exactly that: Alter pitch levels in real time in such a natural way, that even (most) human hearers are unable to detect it. Of course its proprietary code and software, so once again science might have ‘issues’ with copyright.

Now you might ask why doing a mobile application. Well – in Africa mobile phones are everywhere. You might not have running water or electricity in all places, but your chances of finding a person without a cell phone in Somalia for example are quite low (In fact, calls to Europe are often free of cost from Somalia – so much for failed states). Since land lines have been notoriously unreliable and under developed, mobile phones were quickly accepted even in the most remote regions. And it seems to me that the cell phone markets are quickly evolving in Africa and that cell phone products are more innovative because of higher competition in the market and great potentials and growth rates. For example Zain has introduced it’s One Network, which offers free roaming with pre-paid phones & tariffs across 12 African countries. Or the other day I read about a project to increase cell phone coverage across Lake Victoria to increase emergency response times and security. A second step in that project will be enabling SMS text message services so that fishermen can check local market prices on the go to know where to take their fish to make the best of profits, thereby actively raising market competition through telecommunication services in a (mostly) traditional economical system. On the other hand, I know how complicated and expensive (!) roaming can be in Europe or South America from my own experience. On an anecdotal note, it is not uncommon to see a traditional person using a mobile phone. In Tanzania, I remember being surprised at seeing a young Maasai with a traditional dress and an iPod (A number of other people besides me have seen traditional people, such as the Maasai, use mobile phones). Also the Internet is growing quickly – In July 2009 Seacom finished its new sea-cable glass fibre connection, which drastically expands East African Internet connectivity: Another important factor since software products are quickly developing to be a combination of software and services through Web 2.0 technologies.  So, technology is spreading quickly throughout Africa, and mobile phones, the Internet and telecommunication technology are on the forefront.

Besides this MT and TTS on demand application, Mr. Roux also explained various other applications. For example he demonstrated the platform using a touchscreen interface on a cell phone. Suppose you are in the mayors office and need to fill some form. So you click through a menu, where typical phrases are translated enabling a user to understand and fill a form. He was mentioning how this could be used in HIV education or even for extending adult literacy for mother tongue speakers and how one could download these context-driven packages of phrases directly from one’s cell phone on the go via Internet.

I see a huge potential for this software, because it is the right time to launch such a product: Cell phones are widely spread, in Africa in particular and the third world in general, and communication services and Internet access are quickly advancing. The potential range of usages is very wide, since both multi-lingualism and illiteracy form part of the daily life and culture of Africa for years to come. Further, we have reached the technological point, where such things become more and more feasible. Cell phone applications are quickly advancing since the release of the IPhone: 5 years ago it was difficult finding a mobile platform with significant technological development (from both a hard- and software level point of view) to support mobile applications. How many pieces of software, which were not installed out of the box, did you have on your cell phone 5 years ago? But nowadays people are fusing all kinds of applications in cell phones: E-book readers, music players, street navigation, email and Internet services,… And it’s all possible since the operating systems for mobile devices are revolutionizing themselves ever since Apple dropped the IPhone. Nowadays, there is Google’s Android and more and more Smartphones are coming along. For the moment, Smartphones are still expensive, but prices will eventually fall.

I find this application and the concept fascinating, demonstrating the practical applications and impact of connecting (speech) technology and linguistics in the African context. Africa is after all linguistically the most dense region of the world and in dire need of progress. If I understood Neville Alexander right, he sees linguistic independence as a key factor of African independence. This independence should lead to socio-economic progress. And this small piece of software seems to do just that – further linguistic independence.

There have been some more recent advances on e-learning such as ITunes-U, where Apples ITunes store offers free university level course material from international universities such as Stanford or MIT. This is of course more of a side-effect of the currently biggest sales platform for music downloads. Whatever the case, this side-effect seems to overshadow every other e-learning platform or initiative I am aware of.

I was taking a look the other day, and besides the usual suspects I was surprised to find course material for minor languages (considering teaching at least), such as Twi, Tibetan, Kimeru or Amharic (mostly by Emory University so far – Hands up for Emory!) on ITunes-U. To be honest, only the Arabic material left a real impression on me. But go ahead and try to find just one university in Europe teaching Kimeru, which Wikipedia still believes to be a J-Pop band.

As always, the German universities seem to be a little conservative about technological developments. There are very few German universities present in ITunes-U yet. To my knowledge, e-learning has not gained much ground in Germany so far, despite the very well established FernUniversität Hagen, the only “state-maintained distance teaching university in Germany”. Hagen offers degrees to students from all over Germany by a mixture of studying at home and on-site, i.e. campus courses.

Therefore, there was remote learning before there was e-learning in Germany. A little strange to see Germany being so reluctant about e-learning therefore. Then again there might be other similar universities such as Hagen outside of Germany, so it might not have been that innovative in the first place, though I am not familiar with them.

I have observed the same conservatism about e-learning at the University of Cologne, which recently launched a self-developed open-source e-learning platform called Ilias. This is a good initiative of course and inline with other initiatives for the development of open-source software by the university such as the library system OpenBIB. But Ilias went rather unnoticed. Most teaching staff at the university seem to ignore it, while the students seem rather scared to work with and even fed up with new Internet platforms, being deployed every other semester. Course material are mostly being distributed as xerox copies or are simply being emailed to participants. A quick look yields virtually no publicly accessible teaching material by single institutions.

Whatever the case, UOP is an interesting initiative and I hope it will provide some new input to e-learning and consequently increase open access to quality teaching materials.

Unfortunately this app is not being developed anymore and likes to crash all the time on Vista 64, despite fiddling with the compatibility settings and data execution protection. Therefore I was looking for a replacement. Unfortunately, there is no serious competition. The best thing I found was RapidCRC, but it’s slooooow. I checked every program at Wikipedias listing, but they are all no good. Bad interface all over the place, no further development, commercial apps, no GUI,… I was really surprised. I mean its a really small task. If people can put out complete office suites such as OOO for free, why not simple apps as file validators in a decent quality?