Hasil Pencarian  ::  Simpan CSV :: Kembali

"ABSTRAKTujuan penelitian ini adalah mengungkap hubungan antara penguasaan kosakata kognat bahasa Inggris-Indonesia pelajar Madrasah Aliyah MA dengan kemampuan membaca bahasa Inggris mereka serta untuk mengetahui sikap mereka terhadap kosakata kognat bahasa Inggris-Indonesia. Penelitian ini menggunakan metode campuran yaitu kuantitatif dan kualitatif. Sampel penelitian ini adalah 48 pelajar Madrasah Aliyah Negeri MAN Muara Labuh, Solok Selatan. Teori yang dipakai dalam penelitian ini adalah teori kosakata kognat Park dan Dronjic: 2013 . Tes kosakata dan tes membaca bahasa Inggris dilakukan untuk mengetahui hubungan antara tingkat penguasaan kosakata kognat bahasa Inggris-Indonesia dengan kemampuan membaca bahasa Inggris mereka. Untuk mengetahui sikap pelajar terhadap kosakata kognat diberikan instrumen berupa kuesioner dan stimulated recall. Dari hasil olah data penelitian ditemukan bahwa nilai kosakata kognat bahasa Inggris-Indonesia pelajar MA masih sangat rendah dengan nilai rata-rata 43 dari total nilai yaitu 100. Sementara itu, nilai rata-rata tes kemampuan membaca bahasa Inggris berada pada kategori kurang dengan rata-rata 47 dari total nilai 100.Setelah dianalisis, terdapat hubungan yang signifikan antara penguasaan kosakata kognat bahasa Inggris-Indonesia dengan kemampuan membaca untuk pemahaman bahasa Inggris. Terakhir, Berdasarkan hasil kuesioner, juga diperoleh bahwa para pelajar MA memiliki sikap yang positif terhadap kosakata kognat bahasa Inggris-Indonesia.

---

ABSTRACTThis research aimed at identifying correlation of English Indonesias Cognate rsquo s Mastery of Madrasah Aliyah Students and their reading comprehension skills in English and also knowing the attitude of the students toward English Indonesian cognate. The research used a mix method, quantitative and qualitative. The samples were 48 students of Madrasah Aliyah Negeri MAN Muara Labuh Solok Selatan, Sumatera Barat. The theory used in this research is the theory of cognates Park and Dronjic 2013 . Vocabulary test and reading comprehension test were conducted to identify student rsquo s mastery of English Indonesian cognates and English reading comprehension. A questionnaire was also distributed to know students rsquo attitude toward English Indonesian cognates. The finding showed that the students rsquo score on the cognate test was low with an average of 43 from total score which is 100. Meanwhile, the score of English reading comprehension test was also low with an average of 47. The analysis showed that there was a significant correlation between the mastery of English Indonesian cognates and English reading comprehension skills . Based on the questionnaire, it was found that the students have a positive attitude toward English Indonesian cognates."

"Machine Reading Comprehension (MRC) merupakan salah satu task di bidang natural language processing (NLP) dimana mesin memiliki tugas untuk membaca secara komprehensif dari sebuah bacaan (passage) yang diberikan agar dapat menjawab pertanyaan terkait. Metode terkini untuk mengautomasi MRC menggunakan deep learning dengan memanfaatkan pretrained language models (PLMs) berbasis BERT. Dalam menangani kasus MRC sumber daya rendah, digunakan PLM multilingual seperti XLM-R. Namun PLM multilingual memiliki masalah untuk bahasa sumber daya rendah yaitu: bahasa sumber daya rendah yang tidak terepresentasi dengan baik, imperfect cross-lingual embeddings alignment dan instabilitas ketika di fine-tuning pada data berukuran kecil. Penelitian ini mengusulkan beberapa strategi fine-tuning dan metode pembentukan data augmentasi untuk meningkatkan kinerja MRC dibahasa sumber daya rendah. Strategi fine-tuning yang diusulkan adalah 2-step fine-tuning dan mixed fine-tuning. Untuk metode pembentukan data augmentasi yaitu dengan penggunaan data asli, pengaplikasian model machine translation dan perturbasi code-switching. Hasil eksperimen menunjukkan, untuk dataset FacQA (Bahasa Indonesia) dan UIT-ViQuAD (Bahasa Vietnam) diperoleh strategi terbaik dengan kombinasi strategi penggunaan data asli dan metode 2-step finetuning dimana menghasilkan peningkatan kinerja sebesar 3.858%, 2.13% secara berurutan. Untuk dataset FQuAD (Bahasa Prancis), strategi terbaik diperoleh de- ngan kombinasi strategi pembentukan data perturbasi code-switching dan metode mixed fine-tuning dimana menghasilkan peningkatan kinerja sebesar 1.493%.

---

Machine Reading Comprehension (MRC) is one of the tasks in the field of natural language processing (NLP) where the machine has the task of reading comprehensively from a given passage in order to answer related questions. The latest method for automating MRC uses deep learning by utilizing pretrained language models (PLMs) based on BERT. For handling low-resource MRC, multilingual PLMs such as XLM-R are used. However, multilingual PLM has problems for low resource languages: low resource languages that are underrepresented, imperfect cross-lingual embeddings alignment and instability when finetuned on small data.This study proposes several fine-tuning strategies and data augmentation generation methods to improve lowresource languages MRC performance. The proposed fine-tuning strategies are 2-step fine-tuning and mixed fine-tuning. For the method of form- ing augmented data, namely by using data original model, application of machine translation and code-switching pertubation to optimize cross-lingual embeddings alignment in multilingual PLM. The experimental results show that for the FacQA (Indonesian) and UIT-ViQuAD (Vietnamese) datasets, the best strategy is obtained by combining the strategy of using original data and the 2-step fine-tuning method which results in an performance improvement of 3.858%, 2.13%, respectively. For the FQuAD dataset (French), the best strategy was obtained by a combination of code-switching perturbation strategy and mixed fine-tuning method which resulted in an performance improvement of 1.493%. "

"CRYSTAL?Alberta was established to research ways to improve students? understanding and reasoning in science and mathematics. To accomplish this goal, faculty members in Education, Science, and Engineering, as well as school teachers joined forces to produce a resource bank of innovative and tested instructional materials that are transforming teaching in the K-12 classroom. Many of the instructional materials cross traditional disciplinary boundaries and explore contemporary topics such as global climate change and the spread of the West Nile virus. Combined with an emphasis on the use of visualizations, the instructional materials improve students? engagement with science and mathematics. "

"Motivating children to read is essential to building a lifelong love of reading. Many parents cannot model engagement with books and instead, teachers, school media specialists, and children?s librarians must step into the breach to help children embrace a love of reading."

"Color words originate from natural world. In general different language have similar patterns of color hypernymy. Nevertheless, when it comes to color hyponymy, it is not the case. Specific color hyponyms have emerged because different countries. Besides, color words are also rich in cultural symbolism. In other words, same colors in different countries have different cultural symbolism. This paper analyzed six color words (red, white, yellow, blue, green, and black) between Chinese language and bahasa Indonesia to find similarities and differences both in hyponymy and in cultural symbolism. "