^{page 124 note 1} Christaller, J.G., A grammar of the Fante and Asante language called Tshi …, Basel, 1875.Google Scholar

^{page 124 note 2} cf. Twi nsεm nkor͛nkor͛ kyer͛wbea, Accra, undated.

^{page 124 note 3} cf. Christaller, J.G., Dictionary of the Asante and Fante language called Tshi (Twi). Second edition, Basel, 1933.Google Scholar Introduction by J. Schweizer.

^{page 125 note 1} From the dictionary under S we take

sa (v) ‘cut marksin’, se (v) ‘say’, a-se (n) ‘under part’, Sε (v) ‘equal’, si (v) ‘stand, be fixed upright’, sɔ (v) ‘drop, trickle’, so (v) ‘catch’, ε-so (n) ‘top part’, o-su (n) ‘rain’.

^{page 125 note 2} Historically the e-vowel at least appears to be the result of the monophthongization of the diphthong ie. ‘Accept’ in Nzema, for example, is [die].

^{page 125 note 3} The same problem exists with regard to radicals with ‘r’. Some are clearly r-infixed but others might reasonably be considered to be structurally of the order R + suffix ri, re …

^{page 127 note 1} To the A- or biri- type of radical belong those radicals with palatal or labio-palatal consonants as initial and a as vowel final: nya (v) ‘get’, twa (v) ‘cut’, etc. (they are listed in Christaller). Historically these derive from radicals with *i- and *u- diphthongs as finals

twa < ¢kwia

nya < * nia

Their type persists.

^{page 128 note 1} In respect of set, that is. They are frequently susceptible to further analysis for other reasons, cf. fera which ia to be analysed as

^{page 129 note 1} Its ‘normal’ value is IPA 4½.

^{page 129 note 2} Until this fact is better understood and accepted amongst Twi speakers there seems little hope of an end to the constant disputes about the representation of vowel harmony. It is certainly incompatible with the aim of a ‘phonetic’ orthography.

^{page 129 note 3} It is not suggested that the two systems are thereby integrated; they remain discrete. For special purposes (it so happens) they may be expressed by the one notation.