If you were looking for a ‘good’ night out with a ‘bad’ man or woman, would you prefer someone who was amoral, immoral, or non-moral?

That may seem a strange question, particularly in relation to computer security, but when writing my recent articles about Mac security, I was looking for the right word to describe a connection that is not secure. Insecure was clearly wrong because of its connotations of sitting clutching a blanket (however appropriate in this context). Unsecure didn’t sound particularly right to my ear, although it is not unheard of. Non-secure seemed the best compromise, and I hope that you understand what I mean by it.

A quick trip to the books has been interesting but somewhat disfinitive. Huddleston and Pullum’s Cambridge Grammar covers this in its chapter on Lexical Word-formation, at page 1687 ff, and starts by listing the possibilities, adding the inevitable a- and dis- to make me quite discertain. They make the point that the semantic contrast between amoral and immoral has become lexicalised, confirming my unconfidence about insecure.

They then state that “distinctions between un- and non- are semantically predictable”, explaining:
“The forms with non- are emotively neutral and non-gradable, while those with un- have a wider range of meaning, so that they may convey criticism and gradability.”

In that context, both secure and [not]secure are hardly absolute, seem eminently gradable, but the word itself should not convey criticism. So perhaps I should have used unsecure after all.

Noting that chapter was co-authored by Laurie Bauer, I then travelled from Cam to Cherwell with Bauer, Lieber and Plag (2013) “The Oxford Reference Guide to English Morphology”. Here is a whole chapter of the morphology of negation, which explores in-, un-, non-, dis-, a(n)-, and mal-. I feel that I am getting closer to what I want with a privative, which they explain with “at the core of a privative based on ‘X’, we have ‘no X’.” In other words it negates the existence of X, which seems just the meaning that I am after.

However using their Table 17.2 (containing the disfortunate misprint of o(n)- for a(n)-) gives me the possibilities of a-secure, security-less, and security-free. I am now feeling quite insecure about this.

Sure enough, when I look at usage, only insecure seems to have been seen much in British English up to 2000. According to Google Ngrams, unsecure and nonsecure only crop up once or twice, while insecure is growing (as you might have guessed).

The Shorter OED reckons that unsecure and insecure mean the same thing, and non(-)secure does not exist at all. The full OED (v 4.0) is also happy with insecure, but tells me that unsecure is obsolete, while non(-)secure is nonsense, or perhaps insense.

So for the moment, I will remain unsure/insure/nonsure/dissure/assure on the matter. Any pointers are always welcome.