Errors & Omissions: We're all sunk if we keep losing perfectly good verbs

Is "sank" doomed? I hope not, but when a scrupulous writer forgets a word things may be starting to look bad. This is Dominic Lawson, writing on Tuesday: "David Cameron made his provocative remark last Friday; but within hours the news began to leak of David Laws' catastrophically ill-judged expenses claims and the PM's speech sunk without trace."

That should be "sank", which is the proper past tense of "sink", but you often see the past participle "sunk" taking over. The same decay seems to be threatening the verb "ring". The proper paradigm is ring-rang-rung, but you do see things like "He rung the bell and went in."

However, the field workers of the English Irregular Verbs Preservation Society report some cheering successes. Stink-stank-stunk and sing-sang-sung appear to be holding their own. Of course, the forms of such verbs are not uniform anyway. The glory of irregular verbs is that they are irregular. A different pattern is displayed by sting-stung-stung, and fling-flung-flung; there is no "stang" or "flang".

Eyes have it Andrew Cosgrove writes in from Oxford to complain about this, from a news story last Saturday: "The five-ton dinosaur with two 4ft horns directly over each eye lived about 72 million years ago." Mr Cosgrove calls that "a classic example of something I find irritating".

So do I. Two horns over each eye means four horns in all. The writer was struggling to say either "a horn directly over each eye" or "two horns directly over its eyes".

Straight out I'll tell you something else that irritates me: people who write "straight" when they mean "strait".

"Straight" comes down from Middle English, originally the past participle of the verb strecchen, meaning "stretch". So a straight line is like a stretched string – without curves or wiggles. "Strait" comes by way of French from the Latin verb stringere, which means to draw tight. "Strait" means tight, narrow or restricted.

Here is the headline that appeared above a report about Danny Alexander in Monday's paper: "Strait-laced loyalist who played a key role in coalition negotiations." Nothing wrong there. Laces may be tight or restrictive, but they are rarely straight. Indeed, they usually zigzag about a good deal.

A pity then that the story below it was allowed to commit the drearily familiar error: "Even before his arrival at Westminster he earned a seriously straight-laced reputation."

Journalese If some unpleasant event is not being "sparked" then it is being "triggered". Like most journalese usages, these two verbs are employed to create an air of drama. Sometimes it is justified by the facts, sometimes not.

Take this news report from last Saturday: "An overnight passenger train derailed yesterday in eastern India, triggering a crash with an oncoming cargo train that killed at least 71 people and injured another 200."

Consider the trigger of a crossbow or a gun. It is a device for releasing stored energy. The bent bow and the explosive in the gun are ready – eager, you might even say – to produce a sudden kick of energy. Pressing the trigger merely removes the last barrier in the way of an event waiting to happen. Figuratively, then, it may be reasonable to say that a riot, for instance, was triggered by some heavy-handed police action which had the effect of releasing the pent-up anger of a ghetto. But the Indian train crash was not like that. The trains weren't ready and waiting to crash. Everybody expected them to proceed to their destinations. The derailment was the original cause of the accident, not a trigger.

Poor rate A news report on Thursday, about anonymity for rape defendants, said: "Women's groups had argued the proposals would stop other rape victims coming forward, setting back plans to improve low conviction rates in rape trials."

I think this sentence confuses the conviction rate – the proportion of trials that end in conviction – with the absolute number of people convicted. If more complainants come forward and more cases go to trial, the number of convictions is likely to rise, but so is the number of acquittals. The conviction rate could easily fall.