1 Most of these footnotes illustrate some of the ‘questionable constructs’ that the program Correct Grammar (CG) found in this text. The international version was used – i.e. a British rather than American database but incomplete British style messages. The program is discussed more fully later in this article. The footnotes show how the product’s ‘mind’ works. As I had all its classes of rule turned on except the passive-voice trapper, CG found a lot to query in this article – much of it inappropriately. Here the message is, ‘Wordy expression. Consider “usually” instead. Try to express your meaning with direct terms.’

2 This illustrates CG’s parsing capacity. Here it says: ‘Colloquial modifier. Omit or replace with “very”. Real is a colloquial form of the adverb “really” which you should avoid in most writing’. ‘Real tennis’ above, however, gave: ‘Nonstandard modifier. Consider using an adverb instead of the adjective “real”.’

3 CG is having difficulty distinguishing between a verb and a noun: ‘The word “a” does not agree with “brushes”. Consider “brush” instead of “brushes”. ’

4 CG (international version) has no comment. The Fowler, Partridge, Longman and Webster usage books all have entries on this.

5 CG: ‘It is preferable to avoid beginning a sentence with “Or” ’ – predictable advice which is given for ‘And’ and ‘But’ too. To be fair, you can turn this (or any other) rule off and the tutorial does concede that beginning a sentence with a co-ordinating conjunction is partly a question of style. In the British release of the product the tutorial also adds, ‘Consider the differences among [sic] the following sentences.’

6 CG: ‘Word choice. Consider rephrasing with “through”, “by way of”, or “by”.’ Fair enough, but I decided that the ‘in transit’ image was what I wanted here.

7 CG allows you to choose whether to accept one, two or three words splitting an infinitive and acknowledges that split infinitives are necessary to avoid awkwardness and ambiguity. A human style checker (i.e. a colleague) suggested, more pragmatically, that I should alter this split to avoid giving the pedant the chance to criticise.

8 CG may be interpreting ‘what’s’ as a possessive because it is finding a mismatch here. Even so it is hard to see why it suggests changing the verb to ‘are’ instead of ‘is’. The parsing mechanism doesn’t appear to detect that the ‘problem’ here is that I have used a level of informality that may be inappropriate for this kind of journal.

9 CG creates a parse tree and then gives its best guess for each part of speech according to data assembled for the sentence. It gets into difficulties with long sentences with many subclauses. However, no other product comes near it, not even the revamped Grammatik 5 with its boasted ‘Mor•Proof technology’.

10 According to CG the abbreviation ‘e.g.’ should be followed by a comma. ODWE does not agree – preceded, yes; followed, no.

11 Not all of it available at the time of writing.

12 CG has succumbed to purist pressure here in advising: ‘Consider using “that” as the restrictive relative pronoun.’ The tutorial expands and those who disagree with the advice (as OUP does) can turn the rule off.

13 And may have been corrected by the time this is printed. One advantage software publishing has over book publishing is the ease of updating after the first release.