On May 15th, 16th the German-Czech workshop “Automated Context? Practices,
Opportunities and Risks of AI-Driven Translation in Bohemistics
and the Humanities.” will take place in Prague. The workshop is organized by
the Collegium Carolinum, Ufal, BAdW and .dhmuc.
See the announcement here: https://dhmuc.hypotheses.org/files/2025/05/collegiumcarolinum20250428_letak_A4_nahled.pdf
and the full programme here: https://dhmuc.hypotheses.org/files/2025/05/Automated-Context-and-Machine-Translation-in-the-Humanities_program-print.pdf
In anticipation of this workshop I’d like to put forward a theses on the
potential impact of widely available cheap and reliable
translation-technology on the use of languages in science.
The technology of automated translations has become so mature that one
can translate scientific research literature safely, i.e. with little
risk of unnoticed misrepresentation. Furthermore, this technology
opens great perspectives for dealing with historical sources in
different languages, in particular languages that the researcher does
not know closely – although, the automated translation of sources raises
more issues than of research literature.
He are some theses regarding the changes that we may have to expect and that
we can either help to bring about or try to mitigate, if not prevent,
depending on our own opinions and preferences:
1. National languages or, rather, mother tongues, will experience a
renaissance as languages of scientific discourse. And this time it will
not only be the “Kultursprachen”, i.e. languages of big nations, but
those of any nation or people no matter how small.
2. At the same time, paradoxically, scientific discourses will become
denationalized even more as the use of a national language does
not prevent the reception of the scientific literature by non-speakers
of that language, anymore.
3. Open-Access-Publishing is if not a necessary, then at least a promoting
condition of both the previously mentioned processes. (The same goes for
freely available, i.e. Open-Source language-models and software at least
for small languages.)
4. The greater ease with which historical sources in different languages can
be “read” will encourage cross-border-studies, comparative history and,
more generally, the study of regions with many different languages.
5. An enabling condition for the latter will be the digital provision of sources
according to the FAIR-principles by archives and libraries.
6. In both areas, reception of secondary literature and study of sources,
best practices (or rules for good scientific practice) for the use and
reliance on potentially buggy translations need to evolve. But this
should not put anybody off from using the available technologies for
these purposes, now.
added, May the 12th:
Theses regarding the “business” of scientific translations:
A. Independently of technology one can always ask the questions whether having
a single language of scientific discourse (e.g. Latin in medieval times,
English today) is better than having multiple languages in combination with
a vivid translation-culture.
B. However, with the availability of cheap and
reliable translation-technology two central arguments against
multilinguality + translation culture are falling away, in particular:
a) high cost and effort, b) (as a consequence of a) ) necessity of making
an (unavoidably) arbitrary selection.
C. Given that the potential costs are not prohibitive, any more, there are
many good reasons for moving away from the “English as scientific
world-language”-culture to a “multiple languages and translations”-culture:
a) There are no disadvantages for non-native English-speakers, any more.
(like: not being as able to express oneself as in one’s mother’s tongue;
being treated unfairly by referees on because of un-ideomatic use
of English; more time spent on writing and revising papers in a foreign
language)
b) Material that are published in languages other than English will not
as easily be overlooked, any more.
One example: Regional Studies on Biodiversity
are naturally produced in national languages. But this means that global
studies on diversity would in theory need to systematically take
into account studies in many different languages, which, as matter of
fact, does not happen – with detrimental consequences for the global
research on biodiversity.
(https://www.abc.net.au/news/science/2023-09-13/language-barriers-science-diversity-pandemic-conservation/102793488,
https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002184)
c) National languages might benefit from their use as scientific languages.
It might ease communicating scientific results to the general public.
(This is, of course, merely a conjecture.)
d) Further arguments are possible: Certain academic disciplines,
“Geisteswissenschaften” in particular, may be more sensitive to the use
of particular languages than others. Also, cultural conservatives will
certainly embrace the use of national languages in science, but it is
important that the gist of the arguments presented here,
does not at all depend on a conservative stance.
D. Will less (human) scientific translators be needed, in the future? The answer
is “yes and no”:
Yes: Machine translators make many human translators expendable and those who
are still needed will be able to be much more productive, i.e. produce more
translations of the same quality in a shorter period of time.
No: If a culture of multilingualism and translating evolves, many more
human translators will be needed to oversee, quality-check and proof-read
the masses of machine translations that will be produced (at least for the
time being…)
In particular, when it comes to translations of large corpora of
historical source materials (which are more difficult to translated
automatically and, at the same time, require a much more careful
translation) human translators will be needed. (This effect will be
particularly strong, if my conjecture is true that the availability
of machine-translation technology encourages cross-(language)-border
studies.)
I do not dare to speculate which of the two effects will be greater.
OpenEdition schlägt Ihnen vor, diesen Beitrag wie folgt zu zitieren:
Eckhart Arnold (11. Mai 2025). How will computer translations affect science? dhmuc. Abgerufen am 12. Juni 2025 von https://doi.org/10.58079/13wka