International Postgraduate Research Conference (IPRC)
Permanent URI for this communityhttp://repository.kln.ac.lk/handle/123456789/155
Browse
3 results
Search Results
Item Applicability of Google Translate in Sinhalese Diglossic Contexts(International Postgraduate Research Conference 2019, Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2019) de Silva, J.Diglossia is the co-occurrence of two different varieties of a language, for distinct functions, throughout a speech community. Sinhalese is one of the languages which depict this phenomenon, with standard written Sinhalese and spoken Sinhalese as the two varieties. Nevertheless, the necessity of employing both varieties occur in certain contexts, for example, in the translation of prose work into Sinhalese, in which narrative is generally translated into standard written Sinhalese and dialogues are translated into spoken Sinhalese, unless the necessity of foreignizing or classicizing occurs. The aim of this study has been to examine the response of Google Translate in the translation of prose work from English into Sinhalese, in which the diglossic nature of Sinhalese language should be taken into consideration. Accordingly, the study is based on Sinhalese translations of selected parts of English prose texts, produced by Google Translate. The selected parts of source texts consisted of both narratives and dialogues, and pertained to different social and cultural backgrounds. The Sinhalese translations were compared with relevant source texts and an analysis was conducted in order to determine their appropriateness. The findings of this study indicate that the diglossic nature of Sinhalese language is not given consideration in Google Translate and both written and spoken varieties are employed inconsistently in producing a translation. This inconsistently is identified to occur in both sentence level and paragraph level, with a blend of morphological and syntactic attributes of standard written Sinhalese and spoken Sinhalese. Incompatibility with diglossic languages can be adjudged a significant weakness of Google Translate, which stands parallel to the failure of producing natural output consistently. Developing the option for the user to select the required variety is identified as the measure to solve this issue.Item Evaluating Spatiotemporal Dynamics of Snakebite in Sri Lanka(International Postgraduate Research Conference 2019, Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2019) Ediriweera, D.; Diggle, P.; Kasturiratne, A.; Pathmeswaran, A.; Gunawardena, N.; Jayamanne, S.; Lalloo, D.; de Silva, J.Snakebite data has shown spatial and temporal variations in many countries and regions. Yet, no study has evaluated spatiotemporal patterns of snakebites across a country in detail. We used data from the National Snakebite Survey (NSS), which sampled 0.8% of the national population (165665 people) living in 1118 clusters representing all the provinces. Explanatory variables of previously published spatial and temporal models for the NSS data were considered as candidate explanatory variables for our spatiotemporal models. Spatial prediction models for snakebite incidence was a geostatistical binomial logistic model and the temporal prediction model was a Poisson log-linear model, which predicted snakebite incidence at the national level. These spatial and temporal models could not explain locally varying temporal patterns in the country. Therefore, we constructed spatiotemporal models at the provincial levels. The NSS was conducted for 11 consecutive months, and different clusters were surveyed in each month. Therefore, the NSS can be considered as a set of 11 repeated cross-sectional surveys at different locations. NSS captured bite events that occurred in the survey month and in the 12 preceding months. Hence, each individual provided information regarding the number of bites experienced in each of 13 months. In the NSS data, the location of each sampled individual was fixed at the cluster centroid and the data contain the month of each recorded bite, if any, over a 13 month period covering the survey month and each of the preceding 12 months. We modelled the data from each cluster as an inhomogenous Poisson process with cluster-level explanatory variables and estimated the model parameters by maximising the pooled log-likelihood over all. The fitted cluster-level spatiotemporal models were aggregated so as to predict the province-level monthly bite incidence rates in Sri Lanka. Snakebite incidence showed complex spatiotemporal patterns in Sri Lanka. Models fitted for Southern, North Central, Uva and Sabaragamuwa provinces showed both spatial and temporal variation in snakebites. The geographical extent of the high-risk areas (i.e. hotspots) in these provinces dynamically changed over a period of a year. The remaining five models (i.e. Western, Central, North Western, Northern and Eastern) did not show any spatio-temporal interaction, in risk, i.e. the geographical extent of the hotspots persisted throughout the year. Southern, Sabaragamuwa and North Central provinces showed triannual seasonal trends. High snakebite incidences in Southern and Sabaragamuwa provinces were noticed in April followed by December and August to September. Peak incidences in North Central province were seen in November and another two smaller peaks were observed in April and July. Uva province showed a biannual trend with highest incidences in June followed by December. These findings can inform healthcare decision-making at local level, taking account of the seasonal variations in order to prevent and manage snakebites in Sri LankaItem Envenoming Snakebite Risk Map for Sri Lanka(Faculty of Graduate Studies, University of Kelaniya, Sri Lanka, 2016) Ediriweera, D.; Kasturiratne, A.; Pathmeswaran, A.; Pathmeswaran, A.; Gunawardena, N.; Jayamanne, S.; Wijayawickrama, B.; Isbister, G.; Giorgi, A.D.E.; Diggle, P.; Lalloo, D.; de Silva, J.Snakebite is a neglected tropical disease. Hospital based statistics often underestimate snakebite incidence because a significant proportion of victims seek traditional treatments. Since geospatial risk assessments of snakebite envenoming are rare, health care resources are distributed based on administrative boundaries rather than on a need analysis. The aim of the study was to develop a snakebite envenoming risk map for Sri Lanka. Epidemiological data was obtained from a community-based island-wide survey. The sample was distributed equally among the nine provinces. 165,665 participants living in 1118 Grama Niladhari divisions were surveyed. Model-based geostatistics was used to determine the geographical distribution of envenoming bite incidence. The Monte Carlo maximum likelihood method was used to obtain parameter estimates and plug-in spatial predictions of risk. A predictive model was developed with natural and social environmental variables to construct an estimated envenoming bite incidence map and a probability contour map (PCM) to demonstrate the spatial variation in the predictive probability that local incidence does or does not exceed national envenoming snakebite incidence (i.e. 151 per 100,000). Envenoming bite incidence had a positive association with elevation up to 195 meters above sea level, with incidence dropping at higher elevations. The incidence of envenoming was higher in the dry zone compared to intermediate and wet climatic zones and decreased with increasing population density. Developed risk maps showed substantial within-country spatial variation in envenoming bites. Conclusion: The risk maps provide useful information for healthcare decision makers to allocate resources to manage snakebite envenoming in Sri Lanka. We used replicable methods which can be adapted to other geographic regions after re-estimating spatial covariance parameters for each region of interest.