This article is about a small group of words in English, ones beginning with the prefix <de->, followed by <s> and a vowel. If we omit rare words and obvious derivations, we end up with the following listFootnote ¹:

One of the things that prompted me to think of this topic is a peculiarity of French orthographyFootnote ². There, you find the prefix <dé->, often in a privative sense, as in déneiger, meaning ‘to clear the snow from’. This <dé-> is transformed into <des->, however, without the acute accent but retaining the closed pronunciation of the <e>, in words such as desserrer, meaning ‘to loosen’, where the root is serrer. In French, a single intervocalic <s> is normally voiced, with the sound of the <z> of zip, and it seems that the doubling of the <s> in words like desserrer was adopted in order to guarantee the unvoiced pronunciation, as in the <s> of sip. So, whereas the English word desiccation has only one <s>, in French we have dessiccation, with two.

One consideration for this group of English words is how the <s> is pronounced, voiced or unvoiced. You could also group them according to the pronunciation of the <de-> or, again, according to its sense. Yet another permutation would be based on the independence or not of the root. For example, that of deselect is independent, whereas that of desist is not, although it is in the Latin verb desistere ², from which it comes. You could go further and consider whether the root had the same sense independently as it had in the compound. You might also consider the history of the words, seeing how and when they entered the language. If you tried to integrate all those criteria, it looks as if you would end up with a great number of permutations, with many overlapping groups. For this article, it seemed best to group the words according to one criterion only and to look at the similarities and differences within and across the groups. I have therefore split the words up according to the pronunciation of the prefix. The three possibilities are: [dɪ] as in deserve, [di] as in desensitise, and [dε] as in designate. Here are the groups. Note that the spellings desert and desuetude appear in two of the groups.

If we consider the stress pattern, the words of the third group, the [dε] one, with the exception of desuetude, have the stress on the first syllable, the prefix, whereas those of the other two have not. It is perhaps because desuetude is a rather Latinate word that dictionaries are undecided about the pronunciation of the prefix. The [dε] version may, then, be a false nod in the direction of Latin. Perhaps we should remove desuetude from the [dε] group therefore and stick to the pronunciation that seems more natural for English. For the single spelling desert, we have two different stress-patterns. In the [dɪ] group, the stress is on the second syllable, and there are two possible words, with different origins, one meaning ‘to abandon’ (Lat. deserere) and the other ‘that which is merited’ (Lat. deservire), the latter usually now found in the plural in the expression ‘just deserts’. In the [dε] group, the stress is on the first syllable, yet this desert, meaning ‘region lacking life’, has the same etymological origin as the one meaning ‘to abandon’.

How does the pronunciation of the prefix match that of the following <s>? In the [di] group, the <s> is always unvoiced, but in the other two there is a mixture. For the words of the [dɪ] group, we have voicing for desert, deserve, design and desire. For desist, dictionaries offer both possibilities for the first <s>. I myself pronounce it unvoiced, probably because I am more influenced by the classical Latin desistere (unvoiced) than by the modern French désister (voiced). Of the [dε] group, three words, desert, designate and desultory, have a voiced <s>.

If we now think of the meaning of the prefix, it implies separation or undoing for all the words in the [di] group, but only for some in the other two. In the [dɪ] group, it is privative for desist, desuetude and desert (in the sense of ‘abandon’), and in the [dε] group only for desecrate (if we exclude desuetude). Etymologically, the noun desert from that group is ‘an abandoned place’, but we do not now think of a desert in that way. The prefix, then, has a privative sense only by association with the idea of ‘lacking life’.

When it is not obviously privative, the sense of the prefix can be hard to distinguish. In desiccate, for example, the root already has the sense of ‘to dry’ and it is hard to see how the prefix functions other than, perhaps, to reinforce that sense. As for desultory, we do not now associate it with the idea of jumping from something, which is its Latin origin (Lat. desilire, desultum). With desire, you hardly sense the <de-> as a prefix at all. The word goes back to the Latin verb desiderare, which means to long for something, feeling incomplete without it. Therefore, the English word has retained the positive aspect of that, but has lost the privative one. The grammatical term desiderative is really cognate with the form desire, which is one of the two reasons I did not include it in the list. The other is that, although some grammarians may talk easily of ‘desiderative verbs’, the expression does not feature much in mainstream communication. The prefix in some cases, as in deserve or designate, seems to have an idea of application. With deserve, this becomes clearer if we consider the French verb desservir (that double <s> again), which is the equivalent of the English word in form, but not in meaning, having more the sense of the English serve. So, in French, you find, for example, the railway announcement ‘Ce train dessert Beaune et Dijon’ ‒ in English ‘This train calls at (serves) Beaune and Dijon.’

If we separate the root from the prefix, all those in the [di] group are independent. You might say salinate was a specialist word, but it is not unfamiliar. In the [dε] group, the only independent root is siccate, meaning ‘to dry’. I had not met this word before and such research as I was able to do leads me to conclude it is a rare word, restricted to specialized scientific contexts. Of the [dɪ] group, only deserve and design have independent roots. Yet, there is a difference: in the compound, the s is voiced, but unvoiced when it is a separate word. The meaning is not exactly the same either. For we do not normally think of the root of deserve as having the sense of ‘merit’, except perhaps in the expression ‘it serves you right’. The same applies to design, and we can note that design and designate can both be traced back to the Latin designare, meaning ‘to describe’. So, we have two words from the same origin, which, as a result of their different historical paths, differ in stress-pattern and the pronunciation of the prefix.

It is a good idea to finish with dessert, and you may have noticed that the verb in my French sentence about the railway stations looks remarkably like it. The word in its culinary sense is spelled the same in English and in French. A French speaker may sense it as something served or perhaps as something eaten when the table has been cleared. A non-francophone English-speaker is less likely to hear it that way. It is another of those French words, like desserrer and dessiccation, where the <s> has been doubled to ensure the consonant remains unvoiced and not voiced as a <z>. The English version should, in principle, be spelled desert and belong to the set of words I have been discussing here, but it came directly from French, which is why we anglophones are condemned to spelling it with a double <s>, when it only needs a single one. We have had the last laugh, though. For, how do we pronounce that fussy double <s>? As a <z>!