Thesis (MEd)--University of Stellenbosch, 2003. / ENGLISH ABSTRACT: According to Prins(1995), readability factors in Mathematics text do not only influence the comprehension of questions, but also have a marked influence on learner achievement levels. Extending on Prins (op cit), this study sought to investigate whether there are any differences in the quality of interpretation and choice of algo-heuristic methods when isiXhosa-speaking learners respond to mathematical word problems set in English and isiXhosa. The study was located within an ethnographic framework, with all of the 109 participants speaking isiXhosa as LI. The participants were in grades 8 (44), Grade 9 (29) and grade 10(36) and all took Mathematics as one of their school subjects. Learners were divided into four groups based on achievement levels in English in June examinations. The study was cross-sectional, with each of the four groups comprising learners who were good achievers, average achievers and under achievers in English second language. A unique methodological and data collection design was undertaken in such a way that each of the participants responded to word problems set in both English and isiXhosa. Two questions were administered to all learners (one in each of the two languages). The two questions were written in two sessions. If a learner responded in English during the first session, s/he will respond in isiXhosa during the second session and vice versa. There was a 5-minute break in between the sessions. Data was analysed both quantitatively and qualitatively. The broad focus of the analyses was on learners' quality of interpretation of the given word problems and the choice of computational methods they employed when they responded to the word problems. Three categories were investigated under each of the focus areas. Categories investigated under the quality of interpretation were: [J Totally false interpretation [J Partially correct interpretation and [J Totally correct interpretation Categories investigated under the choice of computational methods were: [J Standard methods [J Non-standard methods [J Unidentifiable methods The evidence gathered suggested that isiXhosa-speaking learners interpret word problems better when they are set in isiXhosa rather than English. Another important finding was that isiXhosa-speaking learners prefer to use standard methods when they respond to word problems set in English and also prefer to use non-standard methods when they respond to word problems set in siXhosa. / AFRIKAANSE OPSOMMING: Volgens Prins (1995) beïnvloed leesbaarheidfaktore in Wiskundetekste nie net die verstaan van vrae nie, maar het dit ook 'n merkbare invloed op leerders se prestasievlakke. Ter uitbreiding op Prins (op cit) het hierdie studie gepoog om ondersoek in te stelof daar enige verskille in die kwaliteit van interpretasie en keuse van algo-heuristiese metodes is wanneer isiXhosasprekende leerders op wiskundige woordprobleme reageer wat in Engels en isiXhosa gestel is. Die studie is binne 'n etnografiese raamwerk geplaas. Al 109 deelnemers het isiXhosa as eerste taal gepraat. Die deelnemers was in Graad 8 (44), Graad 9 (29) en Graad 10 (36) en het Wiskunde as een van hul skoolvakke geneem. Leerders is in vier groepe verdeel, en die indeling is op prestasievlakke in Engels in die Junieeksamen gebaseer. In hierdie deursneestudie het elk van die vier groepe uit leerders bestaan wat goeie presteerders, gemiddelde presteerders en onderpresteerders in Engels tweede taal was. 'n Unieke metodologiese en data-insamelingsontwerp is op so 'n wyse toegepas dat elkeen van die deelnemers op woordprobleme gereageer het wat in beide Engels en isiXhosa gestel is. Twee vrae is aan elke leerder gestel, een in elk van die twee tale. Die twee vrae is in twee sessies beantwoord. As 'n leerder tydens die eerste sessie in Engels reageer het, sou sy/hy tydens die tweede sessie in isiXhosa beantwoord, en omgekeerd. Daar was 'n pouse van vyf minute tussen die sessies. Data is beide kwantitatief en kwalitatief ontleed. Die breë fokus van die analises was op die kwaliteit van die leerders se interpretasie van die woordprobleme en die keuse van bewerkingsmetodes wat hulle aangewend het wanneer hulle op die woordprobleme reageer het. Drie kategorieë is in elk van die fokusareas ondersoek. Die kategorieë wat onder die kwaliteit van interpretasie ondersoek is, was: Cl 'n Algeheel verkeerde interpretasie Cl 'n Gedeeltelik korrekte interpretasie en Cl 'n Algeheel korrekte interpretasie Kategorieë wat onder die keuse van bewerkingsmetodes ondersoek is, was: u Standaardmetodes u Nie-standaardmetodes en o Onidentifiseerbare metodes Die gegewens wat ingewin is, het daarop gedui dat isiXhosasprekende leerders woordprobleme beter interpreteer wanneer die probleme in isiXhosa eerder as in Engels gestel is. 'n Ander belangrike bevinding was dat isiXhosasprekende leerders verkies om standaardmetodes aan te wend wanneer hulle op woordprobleme wat in Engels gestel is, reageer, en dat hulle ook verkies om nie-standaardmetodes te gebruik wanneer hulle op woordprobleme reageer wat in isiXhosa gestel is.