Академический Документы
Профессиональный Документы
Культура Документы
Hernani Costa, Anna Zaretskaya, Gloria Corpas Pastor and Miriam Seghiri
University of Malaga
Malaga, Spain
{hercos,annazar,g.corpas,seghiri}@uma.es
d s.r.l.
Engine
s
it
te
Rainbow
Terminu
TermSu
SDL M
Transla
Simple
Sketch
JATE
Kea
Bilingual extraction 3 3 3
Source and target context comparison 3
Terms validation 3 3 3 3 3 3 3
Bilingual dictionaries compilation 3 3
Context extraction 3 3 3 3 3 3 3 3
Support various file formats 3 3 3 3 3 3 3 3
Rank terms by frequency 3 3 3 3 3 3 3 3
Support for many languages 3 3 3 3 3 3 3
Specify the minimal number of 3 3 3 3 3 3 3
occurrences
Show linguistic information 3 3 3
Specify the maximum number of 3
translations
Stopword list option 3 3 3 3 3 3
Choose the minimum and maximum 3 3 3 3 3
number of words per term
Term statistics 3 3 3 3 3 3 3 3 3
According to the survey findings, 27% of the harder to perform than only monolingual as
respondents preferred to have a TE feature it requires a good word alignment system, so
within their computer-assisted translation (CAT) not many existing tools offer this feature. In
tool instead of a separate TE software. Some particular, among the tools we considered in the
translators, however, preferred a web-based previous section only SDL Multiterm Extract
application (9%) or installing a standalone tool and Sketch Engine have bilingual extraction.
on their computer (8%). Nevertheless, the Similarly, TermSuite also offers translation
majority (56%) reported that they did not have candidates for the extracted monolingual terms,
any preference regarding the tool’s interface. The which is a different procedure, but still leads to
fact that translators prefer to have a TE system the same results: terms in two languages. The
integrated in their CAT tool is related to the second ranked feature was the possibility to
general tendency of CAT tools to include more compare the context of the term in the source
and more different features. Indeed, translators and the target language, which is another type
have to deal with a great number of tools that help of bilingual analysis suitable for the translation
them automatize different stages of the translation task. This feature is also quite rare, and of all the
process, so they prefer having one tool with considered tools, only SDL Multiterm Extract
multiple functions rather than having to look for allows such analysis. The possibility to validate
and in many cases pay for several tools. terms or, in other words, choose the terms that
should be extracted instead of extracting all
Regarding the importance of the TE’s
terms was ranked third and is also considered
features, the most useful feature according
useful for translators. This feature is offered by
to survey’s participants was bilingual term
almost all systems, except for TermSuite and
extraction. In fact, considering that within a
Translated. Compiling a bilingual dictionary
translation workflow, terminology extraction is
from parallel texts is another useful feature,
performed with the final objective to translate
which is offered only by SDL Multiterm Extract
the extracted terms, it is more convenient to
and by TermSuite. Finally, the respondents
have the terms extracted in the two languages
considered it useful to extract context together
simultaneously. Bilingual extraction is much
with terms or to see examples from the corpus. extraction, compiling dictionaries and comparing
This is a common feature for many of the studied context in different languages as essential features
tools, including SDL Multiterm Extract, Simple for translators’ work. In addition, it is also very
Extractor or Translated. useful for translators to see the terms in their
Other features that were considered included context in order to understand their meaning and
support for different file formats, possibility be able to find an adequate translation equivalent.
to sort terms by frequency, support for many Not all existing tools, however, provide these
languages, possibility to specify the minimal functionalities. We suggest that developing TET
number of occurrences of the words, show more suitable for the purpose of translation
linguistic information about the term, and select could help professionals in the industry take
the maximum number of translations for one term. better advantage of TE technology. This has
All of them were considered useful, but were not to be done, first of all, by taking into account
among the most useful features. the user requirements. As a step further in this
And finally, some features were not considered direction, it would be necessary to investigate
so important by the respondents. One of them in more detail translators’ attitudes towards TE
was the stopword list option: some of the tools. Especially, the reasons that prevent the
tools, like Simple Extractor, allow to choose vast majority of professional translators to adopt
whether to use a stopword list, and others use them. For instance, many translators might not
it by default. Choosing the minimum and the be aware of their existence or understand their
maximum number of words per term, which was purpose, do not have time to learn how to use
also among the least useful features, can be tuned another complicated interface, or simply have
by all the mentioned TE frameworks, for example. other established procedures for dealing with
And finally, term statistics, which to some extent terminology.
are provided by all tools, were not very important
for most translators either. Table 1 shows which Acknowledgements
of the aforementioned features are presented in Hernani Costa and Anna Zaretskaya contributed
the 9 selected tools. equally to this work and are both supported by
Availability Notes the People Programme (Marie Curie Actions)
SDL Multiterm $500 Free demo of the European Union’s Framework Programme
available (FP7/2007-2013) under REA grant agreement
Simple Extractor $140 60-days demo n.317471.
TermSuit Open Source
Sketch Engine $65/ year 30-days demo References
Translated s.r.l Free
Terminus $440/ year 15-days demo Zaretskaya, A., Corpas Pastor, G., and Seghiri,
Kea Open Source M. (2015). Translators’ requirements for
Raibow Open Source translation technologies: a user survey. In
JATE Open Source New Horizons in Translation and Interpreting
Studies (Full papers), pages 247–254,
Table 2: Depending on the purpose the quotes may
Genebra, Switzerland. Tradulex.
vary. This table only shows the prices for licenses paid
by individuals.
6 Conclusion
Although terminology extraction plays an
important role in several disciplines such as
linguistic research or language teaching, it is
in the field of translation, particularly in the
translation industry where its advantages are
fully exploited and integrated in the workflow.
An example of that is the use of bilingual term