What a student knowledge profile contains Each profile records four kinds of knowledge:
Known graphemes , including multigraphs like ch, sh, ny and ng'. Known consonant clusters , which are sequences of two or more consonants that the student has learned to blend, like mb and nd. Known consonant cluster patterns , each pattern representing the knowledge of a whole family of consonant clusters, like consonant + "w". Known whole words , which are typically words with untaught graphemes or clusters introduced to move the curriculum along. Other factors affect how hard a word is to decode. See Out of scope .
Profiles The preset profiles are set according to Kenya's Tusome (2016) curriculum and are cumulative: each week's profile builds on the previous week's.
Decodability rules The checker looks at one word* (footnote: by "word" we mean anything between two spaces)
If it is a known whole word, it is decodable. If there is any letter that's not a valid Kiswahili grapheme, it is not decodable (e.g. q). If there is any grapheme in the word the student doesn't know, it is not decodable. If there is any consonant cluster in the word the student doesn't know (neither as a known cluster nor as its cluster pattern), then the word is not decodable. Otherwise, it is decodable. Kiswahili decodability rule, applied to one word at a time. Out of scope This version ignores several factors that affect how hard a word is to decode:
Sentence context. Syllabic nasals, where a word like mtoto can be decodable even without the mt cluster, because the m forms its own syllable. The difficulty of blending certain loan sounds. The difficulty of blending certain consonant clusters. The position of consonant clusters within a word. The student's oral vocabulary. Word length and morphological complexity. Letter case, where a student who reads a may not yet recognise A. Decodability package This web application uses our open source decodability Python package to assess each word against the rules above.