Browsing by Subject "translation"

Now showing 1 - 5 of 5
  • No Thumbnail Available
    Item
    Open Access
    Investigating subword tokenization strategies for the translation of English to low-resource South African languages
    (University of Cape Town, 2026) Tyobeka, Manala; Nyirenda, Juwa Chiza
    This thesis investigates how adaptations made to standard subword tokenization algorithms impact the translation performance of Transformer-based Neural Machine Translation (NMT) models that are trained for the translation of text from English into low-resource South African languages. While NMT models are data-hungry; limited research exists to determine how the vocabulary representation, architecture and training strategies of these models can be adapted to better suit South African languages, particularly in conditions with limited time and compute resources. This thesis sources parallel corpora from the Autshumato Project, using roughly 20 000 to 149 000 source-target pairs for model training, that contain translations from English into Northern Sesotho, isiNdebele and Xitsonga. NMT models are then trained on each of the parallel corpora to perform a comparative analysis of standard subword tokenization algorithms with subword regularization, and a task-specific tokenizer for target languages. The task-specific tokenizer narrows the gap between tokenization and translation by leveraging the loss function of a model, that has the same architecture as the NMT model used for translation, to constrain its parameter search. While standard subword tokenization algorithms generally outperform their corresponding subword regularization methods; subword regularization displays superior generalisation capabilities in the face noisy training examples. Relative to standard subword tokenization, the performance of the task-specific tokenizer is comparable; however, it has the capacity to improve the model's predictive accuracy with enough training examples.
  • Thumbnail Image
    Item
    Open Access
    Mobile Zulu
    (2014-09-17) Moolla, Saadiq
    Mobile Zulu is a resource for South African Medical Professionals - free and right on their cellphones - that will assist them when communicating with Zulu-speaking patients. It is intended to facilitate a basic level of communication with Zulu-speaking patients and should not be seen as a replacement to learning the Zulu language, which would vastly improve communication. This resource is valuable to physicians and other medical practitioners who are not yet fluent in Zulu but would like to be able to transmit basic medical concepts and ask questions in their patients' home language.
  • Thumbnail Image
    Item
    Open Access
    A Physiologically-Based Pharmacokinetic Framework for Prediction of Drug Exposure in Malnourished Children
    (2021-02-02) Sjögren, Erik; Tarning, Joel; Barnes, Karen I; Jonsson, E Niclas
    Malnutrition in children is a global health problem, particularly in developing countries. The effects of an insufficient supply of nutrients on body composition and physiological functions may have implications for drug disposition and ultimately affect the clinical outcome in this vulnerable population. Physiologically-based pharmacokinetic (PBPK) modeling can be used to predict the effect of malnutrition as it links physiological changes to pharmacokinetic (PK) consequences. However, the absence of detailed information on body composition and the limited availability of controlled clinical trials in malnourished children complicates the establishment and evaluation of a generic PBPK model in this population. In this manuscript we describe the creation of physiologically-based bridge to a malnourished pediatric population, by combining information on (a) the differences in body composition between healthy and malnourished adults and (b) the differences in physiology between healthy adults and children. Model performance was confirmed using clinical reference data. This study presents a physiologically-based translational framework for prediction of drug disposition in malnourished children. The model is readily applicable for dose recommendation strategies to address the urgent medicinal needs of this vulnerable population.
  • No Thumbnail Available
    Item
    Open Access
    The translation and validation of the LittlEARS auditory questionnaire in isiZulu
    (2025) Nefolovhodwe, Thendo Faith; Petersen, Lucretia
    Hearing loss is a significant contributor to years lived with disability worldwide, affecting approximately 34 million children whose daily lives are substantially impaired. To mitigate the detrimental effects of untreated hearing impairment, paediatric audiologists aim to identify hearing loss as early as possible. This proactive approach enables them to implement timely interventions, ensuring that infants and young children receive the support they need to develop communication skills and overall well-being. Early detection is crucial for optimizing developmental outcomes and enhancing quality of life for these individuals. However, most existing audiological diagnostic tools were developed for older children and are not suitable for assessing auditory development of infants and young children who are not yet verbal. In South Africa, the lack of contextually relevant assessment tools hinders the provision comprehensive audiological rehabilitation. The LittlEARS Auditory Questionnaire (LEAQ) is a parent/caregiver questionnaire that evaluates age-dependent auditory behaviours in the prelingual stage (0-2 years). It screens the auditory development of children with normal hearing as well as those with hearing loss who are fitted with hearing aids or cochlear implants. The responses obtained from the questionnaire are derived from observations made by caregivers or parents since assessing auditory behaviours in paediatric populations is often more complex than in adults. This reliance on caregiver insights is crucial, as formal testing methods may not capture the nuances of a child's auditory experiences. The LEAQ contains questions tailored to specific age groups with the auditory response complexity increasing as the child progresses through different developmental stages. Originally developed in Germany, the LEAQ has been translated into at least 21 languages. Therefore, translating and validating this tool for South Africa's multilingual and multicultural context is essential. This study aimed to (1) translate the English version of the LEAQ into isiZulu, and (2) validate the isiZulu translation. Methods: The study aimed to (1) translate the English version of the LEAQ into isiXhosa and (2) validate the isiZulu translation. Translation Process: (1) Forward translation from English to isiZulu, (2) back translation from isiZulu to English, and (3) an expert committeereview. Validation process: An expert panel (n=6) used acontent validity questionnaire to evaluate the appropriateness and relevance of LEAQ items. 57 isiZulu-speaking parent/caregiver participants completed both the isiZulu version of the LEAQ and a face validity questionnaire. The reliability of the translated LEAQ was assessed through item and scale analysis. Results: Item Content Validity Index (I-CVI) results (0.8-1) indicated relevance across all items of the LEAQ. ScaleContent Validity Index (S-CVI) (0.914) indicated a high content validity of the isiZulu LEAQ. Face validity analysis indicated that all items on the LEAQ achieved an agreement score of ≥80%. The indices of difficulty (0.71-1), signifying that all items on the isiZulu LEAQ were easily comprehensible. All items presented with strong discrimination power ≥0.2. The Cronbach's alpha results indicated good internal consistency with α = 0.86. Conclusion: The findings indicated that following a rigorous translation procedure is useful, as it allows for the identification of any inaccuracies or discrepancies during the translation phases to attain the best quality translation. The statistical analysis indicated that the translated isiZulu LEAQ is valid. Therefore, the questionnaire is appropriate for the use amongst isiZulu speakers.
  • Thumbnail Image
    Item
    Open Access
    Translation and Adaptation of the Self Assessment of Communication (SAC) and Significant Other Assessment of Communication (SOAC) into Afrikaans
    (2022) Liu, Chiawen Juoe; Petersen, Lucretia; Rogers, Christine
    Abstract The Self Assessment of Communication (SAC) and Significant Other Assessment of Communication (SOAC) are self-report outcome measures that are based on the World Health Organization's (WHO) International Classification of Functioning in Disability and Health (ICF) framework. The SAC and SOAC focus on hearing-related activity limitations, participation restrictions, quality of life, hearing aid use, and satisfaction from the perspective of the hearingimpaired individual and the communication partner or significant other (SO). The SAC is administered to the hearing-impaired individual, while the SOAC is the companion questionnaire that is administered to the SO. The SOAC focuses on the hearing disability and handicap of the hearing-impaired individual from the perspective of the SO. The SAC and SOAC were initially developed in English and standardised for a developed country population, the United States of America (USA). Thus, to make the outcome measures more relevant to a multicultural and multilingual country like South Africa (SA), the most practical solution is to translate and adapt the measure for the SA population. In SA, few audiological outcome measures have been translated and adapted. Therefore, to add value to the limited research available, the study aimed 1) to adapt and translate the SAC and SOAC into Afrikaans using the American Association of Orthopaedic Surgeons (AAOS) guidelines; and 2) to describe the usability of the Afrikaans SAC and SOAC by South African audiologists through surveys and verbal feedback interviews. The AAOS translation and adaptation guideline is a multi-step translation method that involves five stages: (i) Forward Translation from the source language (English) to the target language (Afrikaans); (ii) Synthesis of the forward translations; (iii) Backward Translation from the target language (Afrikaans) back to the source language (English); (iv) Expert Committee Review by analysing all the translations to create a pre-final Afrikaans SAC and SOAC measure; and (v) Field testing the pre-final Afrikaans SAC and SOAC. The field testing stage involved the target population (Afrikaans speaking hearing-impaired individuals and SOs) who analysed the pre-final Afrikaans SAC and SOAC. The participants in the field testing stage were sampled using purposive, convenience, and snowball sampling methods within the Western Cape Province in private practice. For the second aim of the study, 16 bilingual Afrikaans and English audiologists (referred to as audiologist participants) were sampled using purposive and snowball sampling. The audiologist participants were invited to use the Afrikaans SAC and SOAC in their practice for one to two months. Following implementation, the audiologist participants were required to comment on the usability of the Afrikaans SAC and SOAC by completing a survey. Three audiologists agreed to do a verbal feedback interview reporting their experience with the Afrikaans SAC and SOAC. Overall, audiologist participants' responses (survey and verbal feedback interview) indicated positive feedback regarding the translated Afrikaans SAC and SOAC. Most felt that the measure was a useful counselling tool, was user friendly and provides valuable information on the perception of hearing loss. However, some implementation challenges were reported such as (i) limited time available, which caused a delay in appointments with other patients, and (ii) patients' SO were not always present in consultation, as such, the SOAC could not be administered. In conclusion, findings revealed that following a multi-step approach when translating and adapting a measure is valuable, as it allows researchers to identify inaccuracies and discrepancies during the translation process to achieve a good quality translation. Furthermore, based on audiologists' perspectives, the Afrikaans SAC and SOAC can be used in clinical practice by South African audiologists. However, it is recommended that further research is necessary to explore psychometric properties such as criterion and concurrent validity.