[UgaBYTES] Slow translation affecting product uptake in local languages

[Mwathi Francis]

Lack of agreed language terms, the slow pace of translation and lack
of technical expertise have greatly stunted the development of
software in local African languages.

Many African languages are not documented and most technology terms
such as mouse, phishing and malware, among others, do not exist in
African languages, which has hampered the development process.

"There is insufficient investment in product translation by major
global brands; the pace is too slow and even when done, it takes a
long time to have the final products for use. Meanwhile, languages are
developing and changing, meaning that the product may look "old" when
developed because of the slow nature of implementation," said Hezron
Mogambi, a linguistics lecturer and Swahili translator at the
University of Nairobi.

Microsoft, Google and Facebook are some of the global brands that have
translated their products into local languages for use by communities
that may not be conversant with English or prefer to use their local
languages.

Google has launched Amharic, Oromo, Tigrinya (Ethiopia), Somali
(Somalia), Kinyarwanda (Rwanda), Kirundi (Burundi), Lingala
(Democratic Republic of Congo), Hausa, Yeruba (Nigeria), Zulu, Xhosa,
SeSotho, Africaans (South Africa), Shona (Zimbabwe), Malagasy
(Madagascar), and Akan (Ghana and Ivory Coast), said Julie Taylor,
communications and public affairs officer at Google UK.

In its quest to reach more people, Microsoft is developing products in
more than 60 African languages under the Local Language Program.

Microsoft has probably faced the biggest challenge in its translation
projects -- its products are available in eight African languages, but
the products are never translated in a timely way for use by the
community. For instance, Windows Vista translation into Swahili is yet
to be unveiled for the 100 million Swahili speakers in East and
Central Africa. The company is in the process of testing Windows 7.

"Lack of local localization experience has affected Microsoft. Even if
partners have translation experience, they are not always familiar
with the languages, localization practices and tools. It takes time
for Microsoft to train and manage these skills," said Mark Matunga,
citizenship manager, Microsoft Eastern and Southern Africa.

Lack of agreed language tables may be a perfect excuse for lack of
swift translation, but there is also the tug-of-war over which
community has the authority to translate what language, and that
affects the product's acceptability in the market.

Tanzania has always held the higher ground in Swahili translation
projects as Tanzanians speak the more authentic and pure Swahili
compared to Kenya. So, if a product is translated in Kenya, Tanzanians
may not accept it, citing the difference in terminologies. There is
also the question of whether it is translated at a university or by a
commercial vendor.

A commercial translation company tends to have more experience, with
less need for training or troubleshooting. Such companies also are
familiar with the rigors of commercial development and tend to stay on
schedule and work within Microsoft expectations more easily, said
Matunga.

"But working with local community groups offers Microsoft the
opportunity to build deeper relationships, promote local businesses,
build greater opportunity to ensure local relevance, which takes more
time because we may need to communicate and coordinate among more
stakeholders," added Matunga.

Google has managed to escape the headache of slow translation projects
by using machine translation for searches and Suggest, which is
integrated within other Google global projects. Google recently
launched Ethiopic transliteration in Suggest.

"Most state-of-the-art, commercial machine-translation systems in use
today have been developed using a rules-based approach and require a
lot of work by linguists to define vocabularies and grammars," said
Google UK's Taylor.

However, automatic translation is considered very difficult because
the meaning of words depends on the context in which they are used, as
well as the structure and rules of a language. For instance, some
words are similar but different in meaning.

"While machine translation is not yet perfect, we believe that the
current quality of machine translation is sufficient for many users to
obtain a gist understanding of content in other languages -- and over
time, our automatic translations will get better," added Taylor.

While more and more Africans are expected to get online, the literacy
levels are still slow and access to local content in local languages
will be vital. Most of the content online, whether via PCs or mobiles,
is still in English and French, which are the dominant Western
languages in Africa.

Compared to other regions, the translation projects are taking a
similar slow pace, but the lack of documentation and local leadership
in volunteer translation has affected the pace. The glossary of words
is the toughest as meanings of words differ between language dialects,
but once a glossary is done, more people will access online resources.

"Use of local languages over the Internet will definitely help in the
translation process because once standard equivalent terminology has
been established, then the Internet would serve as a channel to the
efforts of African states to use modern media and technology for
development and civic participation," said Mogambi, who also lectures
in Swahili at the university.

Machine translation may become an easier tool, but the linguists and
cultural experts need to develop acceptable terminologies for
communities to take up.

Source: www.computerworld.co.ke

-- 
Enter our blog contest for your chance to win recognition and prizes
from telecentre.org
For more info, please check www.telecentre.org


Francis Mwathi
Support Community Facilitator
UgaBYTES Initiatives (www.ugabytes.org)
Telecentr.org (www.telecentrecommunity.ning.com)
Tel: +256 414 370163
Mob: +256 783 010269
Skype: francis.mwathi
E-Mail: mfrancis at ugabytes.org