This paper discusses the problem of sparse translation of English into Sundanese and Javanese that were found in Translator-Gator. Translator-Gator is a language game created by the United Nation Global Pulse, to support the research initiatives in Indonesia. Thousands of keyword were generated and translated from English into some Indonesian local languages using the crowd resource. Unfortunately, many English words are still has no translation in Javanese as well as Sundanese. To overcome this problem we propose a technique to fill the un-translated English words in Javanese and Sundanese using Indonesian translation as a pivot language. Evaluation was made by manually investigated whether each phrase results a proper translation. Experiment shows that our technique results relatively low translation accuracy. Limited coverage of phrase translation list and ambiguous words are identified as causes of translations errors in our technique.

Japanese language has complex writing systems, Kanji and Kana (Katakana and Hiragana). Each one has different style of writing. One simple way to differentiate is Kanji have more strokes than Kana. Meanwhile, it needs a lot of effort to remember characters of Katakana and Hiragana, thus it will be very difficult to distinguish handwritten Katakana and Hiragana, since there are a lot of similar characters. This is the reason why we need pattern recognition. Handwriting recognition has been the object of intensive research during the last decades, including Japanese characters recognition. In this research, we build an application to recognize Kana's character handwriting. Pattern recognition on Kana is started based on Optical Character Recognition (OCR) consist of scan image, preprocessing, feature extraction and post-processing. Classification process will be done in post-processing using neural network back propagation. The processing of input images, involves training and testing image. Training image uses 3.956 characters done in three phases of training. Testing image uses 552 characters of some images inputted in training phase and image which has not been used before. The test result achieve maximum accuracy of recognition system at 91.88% from train image and 31.03% from test image.

Sundanese language is the second biggest local language used in Indonesia. Currently, Sundanese language is rarely used since we have the Indonesian language in everyday conversation and as the national language. We built a Sundanese lexical database based on WordNet and Indonesian WordNet as an alternative way to preserve the language as one of local culture. WordNet was chosen because of Sundanese language has three levels of word delivery, called language code of conduct. Web user participant involved in this research for specifying Sundanese semantic relations, and an expert linguistic for validating the relations. The merge methodology was implemented in this experiment. Some words are equivalent with WordNet while another does not have its equivalence since some words are not exist in another culture.

Phrase table combination in pivot approaches can be an effective method to deal with low-resource language pairs. The common practice to generate phrase tables in pivot approaches is to use standard symmetrization, i.e., grow-diag-final-and. Although some researchers found that the use of non-standard symmetrization could improve bilingual evaluation understudy (BLEU) scores, the use of non-standard symmetrization has not been commonly employed in pivot approaches. In this study, we propose a strategy that uses the non-standard symmetrization of word alignment in phrase table combination. The appropriate symmetrization is selected based on the highest BLEU scores in each direct translation of source–target, source–pivot, and pivot–target of Kazakh–English (Kk–En) and Japanese–Indonesian (Ja–Id). Our experiments show that our proposed strategy outperforms the direct translation in Kk–En with absolute improvements of 0.35 (a 11.3% relative improvement) and 0.22 (a 6.4% relative improvement) BLEU points for 3-gram and 5-gram, respectively. The proposed strategy shows an absolute gain of up to 0.11 (a 0.9% relative improvement) BLEU points compared to direct translation for 3-gram in Ja–Id. Our proposed strategy using a small phrase table obtains better BLEU scores than a strategy using a large phrase table. The size of the target monolingual and feature function weight of the language model (LM) could reduce perplexity scores.