Your quote seems unlikely to be correct.

After an initial interest in instrumental phonetics, Daniel Jones went off it completely. Besides that, he’s known for his emphasis on traditional ear training. The cardinal vowels are his, after all.

The version of the anecdote that I am familiar with involves Daniel Jones setting off on some fieldwork and a journalist asking him what instruments he would be taking.

Jones was Abercrombie’s teacher of course, so the version of the anecdote that you give seems unlikely.

As for instrumental phonetics, it’s essential these days if you want to do phonetics professionally (as your job, I mean) in an English-speaking country. As I found to my disappointment some years back! The fact of the matter is that university lecturers must publish to get and keep their jobs, and you can’t publish anything based on impressionistic analysis these days – it won’t wash. Although I consider myself an accomplished practical phonetician (especially by today’s standards) and an average instrumental phonetician, I agree that this is the way it must be.

1. They are different on each different person. So, if you do research with only your ears, that research has a lot less value when you are elsewhere, because no one else can really duplicate your ears.

2. They can hear rhythms, but they cannot tell you what rhythm *is*. Do it with a machine, and you will eventually find out that it is (perhaps) some particular patterns of loudness and duration (or something like that). Do it with your ears, and you can hear it, but you can only define it by saying that it has that certain je ne sais quoi.

3. They do not do statistics. You can only hear the rhythm of one person at a time: you cannot hear the average rhythm of a language.

That’s not to say that machines (alone) don’t have equally large problems. In fact, what we really want to do is to connect the subjective perceptions that we all have to the objective reality out there. That takes both humans and machines.

Some idle thoughts – having lived 9 years now in the US, I’m beginning to think that that US speech (esp young persons) is moving towards syllable timed. The rhythm certainly seems different than British English. (Also all the young females are adopted markedly creaky voice quality, but that’s another story)!

In phonetics, machines can just help us to consolidate our perceptual findings. We cannot always depend on experimental findings. It is unfortunate that, these days phoneticians rely solely on experimental evidence and they ignore other perceptual evidence. In fact many phoneticians these days do not know basic phonetic measures like Cardinal Vowels etc. When I asked the late Peter Ladefoged: “What do your students in American universities use if they want to refer to different vowel qualities in two or more regional dialects?” His answer was” “They compare formants, plotting on vowel diagrams!”