In Issue 135 (Volume 34, Number 3, September 2018) of English Today there was an article by Blasius Achiri-Taboh entitled ‘English spelling: Adding /ʃən/ (or /ʒən/) to base-words and changing from -tion to -sion.’ The author's stated aim was to provide help for deciding the forms of these words and whether -tion or -sion was the correct ending. All the words he cited were ones that can be traced back to Latin or French. That is to say, they are Latinate words that became part of the English language either in a French form as a result of the Norman invasion in the 11^th century or later as a modification of a classical form in the Renaissance period. This historical fact was not mentioned in the article. I acknowledge that the purpose of the article was to give guidelines for spelling, particularly perhaps for non-native speakers, and that it might be going too far to suggest that, if you wanted to be able to spell these words correctly, you could learn Latin and maybe French too, but it seemed to me that the author's assertions misrepresented the nature of those English words. I should like therefore to discuss certain points of the article, in the order they were presented.

Having explained that he prefers what he calls the ‘base-word’ model to the ‘word-base’ one, Achiri-Taboh proceeds to the application of his principle first with the noun persuasion, saying that the consonant at the end of the verb persuade, the base-word, ‘simply drops off’ when the suffix is added. The two English words come ultimately from the Latin verb persuadere, its supine form persuasum and the cognate noun persuasio (accusative case persuasionem). In Latin, the [d] of persuadere does not, in fact, ‘drop off’, but is assimilated to [s], with the long <a> in persuasum preventing the assimilated [d] from being orthographically represented as <s>. There is therefore only one <s> in the spelling. Philip Baldi (Reference Baldi1999) gives a similar example for the Latin verb mittere. There, for the perfect tense, a putative *mitsi becomes *missi and then misi, the single <s> in the spelling being forced by the preceding long vowel. The supine of that verb is missum, where the assimilated [t] can appear as <s> in the spelling, as the preceding vowel is short. The principle is also clear in other contexts: the English verb assign, for example, comes from the Latin verb assignare, which consists of the preposition ad plus signare. Those who know Latin and Italian will be well aware of the degree to which the modern language assimilates the consonants of its ancestor: otto from octo, oscuro from obscurus, assoluto from absolutus, estremo from extremus, for example. The explanation of the form of the English noun persuasion lies therefore in the phonetic system of Latin, not of English.

Achiri-Taboh continues his idea of nouns related to verbs ending in <d> with a claim about extend and pretend that I at first found strange. He says that, although the [d] is not represented in the spellings extension and pretension, it is retained in a weak, unvoiced form, under the influence of the preceding nasal, when the word is spoken. He gives this phonetically as /prɪtɛntʃən/. Now, I have never been aware of anyone pronouncing pretension in that way, with a faint [t] before the -sion ending. In the two dictionaries I possess and the five online ones I consulted, no sort of [t] sound in that position is given in the pronunciation guide. The author cites the third edition of the Cambridge Advanced Learner's Dictionary (CALD) for this pronunciation. I have not been able to consult that work, but I have looked at the Cambridge English Pronouncing Dictionary (CEPD), 18^th edition, and that does, indeed, offer the phonetic transcription given by Achiri-Taboh. It is contradicted, however, by the Cambridge Online Dictionary. I must hesitate before criticising the editors of CALD and CEPD, but I can't help feeling they are proposing a pronunciation that, though it may be found, is not the standard one. Whatever may be the right view about the pronunciation, it is still false to claim either that the [d] disappears or that it is transformed into a weak [t]. The usual Latin supines of extendere and praetendere are extentum and praetentum, though you also find extensum in several classical authors and, more rarely, praetensum. The reason the English spellings are not *extention and *pretention can be seen in the cognate Latin nouns extensio and praetensio. Even though there are normal Latin words ending -entio, such as intentio, a theoretical *praetentio would easily become praetensio because of the well-known effect that [ɪ] can have on a preceding [t]. In the French word intention the <t> becomes pronounced as [s], though the spelling is not changed to *intension. The Latin praetensio is medieval, rather than classical, and so pretension may come from that or just by analogy with words like ascension or, indeed, extension.

The English nouns ending in -tion or -sion are presented therefore by Achiri-Taboh as the result of a process applied to a shorter cognate word. Even if this is proposed as an abstract idea, without any claim that it actually happens, it still gives a false impression of the nature of these words. This becomes particularly clear in the discussion of the word reformation. Achiri-Taboh claims that in this word ‘epenthesis takes place between the base-word and the suffix.’ Epenthesis is the production, usually involuntary, of a supplementary sound that makes the word easier to pronounce. Sometimes it becomes represented in the spelling, sometimes not. In English, the [d] of thunder, for example, is epenthetic to facilitate the transition from [n] to [r] and appears in the spelling, whereas in dreamt there is, in many people's speech, an epenthetic [p] that is not orthographically represented. Here, Achiri-Taboh is saying that the <a> before the -tion of reformation is an epenthetic vowel. In other words, that the ‘real’ form is *reformtion, which is made more pronounceable by the insertion of a vowel that is not a true part of the word. The correct view is simpler: the word reformation comes directly from the Latin supine reformatum and the cognate noun reformatio. It is not the result of any transformation of the word reform. In the Latin words, the <a> is not epenthetic, nor is it in the English one. In fact, an epenthetic vowel in the middle of an English word is rare. You only find it in non-standard pronunciations, such as /fɪləm/, for the standard /fɪlm/.

Achiri-Taboh then presents two tables listing 18 different ways in which various types of words, a mixture of verbs, nouns and adjectives, become nouns ending in -tion or -sion. The first table consists of six words ending in double <s>, with cognate nouns ending in -sion. These are designated as being of a single type. An example is compress – compression. The proposal here is to treat them all in the same way, in a procedure whereby the final <s> of the base-word is said to be dropped and the -sion ending added, thus neatly preserving the double <s> of the shorter word. The fault here is that, even if the idea of such a transformation were valid, it is the wrong way round, as all six of these words, compress, express, impress, process, profess and success, are back-formations, based on the -sion form, which is derived from the Latin supine or, in the case of profession, the perfect participle. Furthermore, they include three different types of words, exemplified by procession, profession and impression, whose Latin infinitives are procedere, profiteri (a deponent verb) and imprimere. We can see here the English verb proceed, which is not a back-formation. There is no English word, having the sense of profess, that looks like profiteri. From imprimere we eventually get imprint and we can note the French infinitive imprimer, which looks more like imprimere, with both French and English having impression, from the Latin impressum or impressio. It should be clear, then, how these six words cannot be regarded as a single type.

The second table, listing 17 types of transformations, is commented on in detail by the author. I shall not deal with them all. The basic misconception seems to me to be exemplified by the following: ‘What determines the preferences for consonant change and the choice of consonant in cases like conjoin – conjunction as opposed to epenthesis and the choice of epenthetic vowel in cases like define – definition before the addition of the suffix is not clear’ (Achiri-Taboh, 2018: 39). The answer to that statement is that, on the contrary, it is perfectly clear: the consonants and vowels that appear are determined by the forms of the Latin or French words that have come into English. In the case of conjoin – conjunction, for example, the origin is the Latin verb coniungere, whose supine is coniunctum, where the voiced [g] of the infinitive becomes an unvoiced [k] in the supine under the influence of the following [t]. That is the origin of the <c> in conjunction. The verb conjoin comes from the French conjoindre, a development of the Latin coniungere. Although there may have been further changes over the centuries in the pronunciation of all these words in English, the explanation of their form lies in the languages from which they are derived. The etymological origins are given in the relevant works of reference, such as the Oxford English Dictionary.

It seems to me therefore that the author has created a non-existent problem beset with unnecessary difficulties. This is further revealed in the final section of the article, having the heading ‘Non-derived words’. The author defines these as words ending in -tion or -sion that are not based on some shorter English word. Yet these words have a derivation. Some, like auction, come from a Latin verb, others, like nation, come from a Latin noun (though that noun, natio meaning nation, is related to the Latin participle natus, meaning born). For those that are related to verbs, the fact that there is not a shorter cognate English word does not make them a different class. The word diction, for example, is of the same type as confusion in that both come from the supine of Latin verbs and neither is produced from a shorter English form. As with the words ending in -ess, it is confuse that comes from confusion, and not the other way round. The Latin origin is confundere – confusum/confusio, giving confondre in French and confound in English. Both French and English have confusion, but it is only English that has formed confuse from it: there is no French verb *confuser. To follow Achiri-Taboh's approach would, in my view, be to risk acquiring a false idea about the character of these so-called ‘non-derived’ words.

How, then, will what I have said help the uncertain writer to avoid having to consult a dictionary each time for these words ending in -tion and -sion? My answer may seem disheartening to start with, but I think it is the right one. Rather than trying to remember Achiri-Taboh's seven rules, it would be better simply to try to remember how the words are spelled each time you encounter them in a reliable source. This, though, could profitably be reinforced, not perhaps by going so far as learning to read Latin (though I would not discourage such an enterprise!), but by learning about Latin as the source of these -tion and -sion words. It could lead to a more correct understanding of the form of these words and a better feeling for them. Dipping into Latin should be interesting and that interest should back up the memorisation.