Has anyone considered the application of MPEG-7 (multimedia content
description standard) to QDA? See...
http://en.wikipedia.org/wiki/MPEG-7, or
http://www.chiariglione.org/MPEG/standards/mpeg-7/mpeg-7.htm
for MPEG-7 info.
It would seem that QDA coding against MPEG-7 descriptions (which in turn
identify and label multimedia elements) could be part of a coding solution
for both XML-encoded and multimedia materials.
I think this would point out that a concern about marking overlapping
segments of XML with markup in the original document is erroneous. The XML
is intended first to demark content. It can be extended to mark format (but
gets messy), and annotations can be applied within the structure as an
attribute to elements. However markup from a QDA coding perspective should
be viewed as *separate* from the original content, more like an "overlay"
that demarks part(s) of the content and adds commentary as an annotation. It
would never structurally "overlap" with the original content. This is nicely
recursive, because the annotation overlay documents can be combined and
analyzed, then annotated with another overlay (keeping the source of each
separate annotation distinguishable, and related to the demarked part of the
first original document). Viewing tools should allow parallel display of the
original multimedia and XML document content and the associated one or more
levels of annotation to support the visualization needed for analysis.
Now we would just need this capability to be built into Atlas and other QDA
tools.
Regards,
John Hanna
----- Original Message -----
From: "Alan Stockdale" <[log in to unmask]>
To: <[log in to unmask]>
Sent: Friday, June 16, 2006 1:15 PM
Subject: Re: To be or not to be... XML compatible.
Normand,
I'm not familiar with QDA Miner and wasn't being critical of your program or
your earlier comments or suggesting that QDA minor in particular should
support this particular feature. I was just jumping off on a couple of
comments you made to advocate on behalf of better support for time-aligned
transcripts in QDA programs that are frequently used to analyze recorded
interview and focus group data.
It is true that not everyone works with audio data (and many fewer with
video) but there are an awful lot of people working with audio data. I'd
hazard that a large percentage of people who use programs like Atlas-ti,
NVivo etc. are analyzing transcriptions of audio recordings of interviews
and focus groups. Look around on the various qualitative discussion lists
and a question that is posted over and over is "What recorder should I buy?"
If you work with audio and you are transcribing that audio there are
enormous advantages to having the transcript time-aligned. It's a huge
time-saver, cost-saver, and greatly facilitates quality checking. Example:
We do research on medical topics so there are a lot of technical terms that
the transcriptionsist misspell or have trouble "hearing" because they are
unfamiliar with the word. Inevitably there are also sections that the
transcriptionist as difficulty making out. We have the transcriptionist mark
all those with square brackets. We also have them mark real names and
potentially identifying information. This makes checking, correcting, and
cleaning the transcript very fast. Search for "[", play, type in correction
or remove/change identifier, search next... During analysis if we want to
hear the tone of voice, pauses, etc. or just make sure that the transcript
was accurate, we can do that relatively quickly because we have time-aligned
transcripts. We can't do it in the QDA software itself, but I can find and
listen to the exact audio section within any of the hundreds of interviews
on my PC within a minute or so by loading the transcript into Transcriber. I
just can't imagine living without any of those things and I certainly can't
imagine trying to find particular portions of audio manually in either a
digital file or an analog tape. QDA programs could support this directly.
That would greatly improve the easy of use and functionality.
I'd also suggest that a huge benefit of supporting time-alignment within QDA
programs is that it would facilitate selective transcription. You'd always
be able to listen to the audio but maybe you'd only transcribe certain parts
as needed. For someone on a limited budget that would be a very useful
feature.
I think there are already quite a few people on this list and other QDA
lists that using Transcriber and other time-aligned transcription tools.
Transcriber seems to be consistently running 600-1000 downloads a month at
the moment. I don't know who is doing the downloading and for what purposes
but I do know there are quite a few qualitative researchers using it.
My larger point was that the discussion seemed to be heavily focused on data
*exchange* via XML betwen QDA and other data analysis programs when I think
many end-users might realize greater benefits form data *import* via XML
from initial data processing tools such as Transcriber that produce
time-aligned transcripts.
(Note that I don't use Onzeminer. I merely cited it as example of a research
project--very different from the sort of work I do--that benefited from
import of time-aligned transcripts created in XML.)
Final note: You wrote "I would also prefer to deal with one file format (for
example an XML standard for all transcription tools), rather than creating 5
different importation routines for 5 different transcription tools." Yes, I
agree that would be good. But isn't the whole point of XML that the data
file can be easily transformed? One of the nice things about Transcriber for
me is that I could readily transform the XML data file (it comes with
numerous pre-built export filters and the facilite to write your own). It's
not that the basic structure of a Transcriber XML file is particularly
complex. I don't see that importing different time-aligned formats should be
a big deal. It doesn't seem to bother the linguists. There are lots of
linguistic annotation tools out there are there are lots of them exchange
data. They even have a program to exchange between different time-aligned
formats. See http://medien.informatik.fh-fulda.de/tasxforce. In any case how
many transcription tools suitable for use by qualitative researchers are
there that generate time-aligned transcripts in XML? Not many at the moment.
Transcriber is the only one I can think of. I'm not sure how Transana stores
it's data. That's seems to be the only other one that is commonly used by
qualitative researchers. And if you wanted to come up with something
different than what's already utilized by Transcriber, the program even
allows you to use your own DTD superset. It's open source! All the QDA
developers could get together, agree on an XML format and just build on top
of Transcriber. It's been done by others before (e.g.
http://www.icsi.berkeley.edu/Speech/mr/channeltrans.html.
Cheers, Alan.
|