Off-Axis Effects

A peculiar possible complication to the analysis of sperm whales coda is that their constituent clicks are not produced directly by percussion, but are instead amplified sonic signals sent down a very long sound gun. This they can focus into a tight beam. Such a mechanism opens the possibility that the coda might display variation by recording angle from the centre of that beam.

Although this problem has been discussed in peer reviewed articles, the only published direct evidence for it is addition of a very faint extra click between main clicks. This appears at high angles. To my knowledge, no distortion of the main interclick interval (ICI) has ever been shown. If this is the current state of knowledge, then it differs from many interviews and casual articles on this subject.

I was interested to find a popular article that claimed all sperm whale coda other than the 5R coda type remained undistorted by off-axis effects. What worries me is that scientific endeavour is directed by more that the established facts, research is also steered by the current common assumptions of those in the field.

To properly claim such distortion (as would appear in peer reviewed literature) you would need to compare sound from two microphones. However, if you were simply making your best guess, you could conclude this by visually inspecting the data set from a single microphone and finding that 5R clustering is much tighter than for other coda. The reason for this difference has remained quite a mystery to them, but several research groups hypothesised that 5R is analogous to the signature whistles of dolphins. As such they have measured the 5R variation in each individual within a pod, and found that this sometimes produces evidence in support, and at other times against, this proposition.

If recording angle interfered with the ICI, it would do so in a continuous fashion, not a discrete one. As such, it could not be responsible for the Duda-Hart breakdown of coda into dozens of subcoda, but it does have potential to interfere with the analysis that would tell us if these coda were produced with the distinctive pattern of a human-like language.

Let me place an alternative hypothesis before you for testing. I propose that off axis coda distortion is negligible, and that all two-microphone data sets will be found to support this. This will be my working assumption, and I have never found any evidence to contradict it. If so, we can return from this diversion, and continue to analyse the data in the fashion I have suggested in earlier pages under the ‘language’ tag. If anyone knows otherwise, (perhaps for a very recent publication) I would ask you to please contact me with urgency.