The promise of numbers
When I was a primary school child, numbers were a quotidian humiliation. I was bad at maths, and decades of anxiety followed. It probably started with the times tables, which we learned by rote, and had to line up to parrot or respond to our teacher’s interrogations. When we failed, we were roundly slapped. I dreaded the daily line-up. It was terrifying. I took to asking out to the toilet before this daily exercise in incompetence and failure played its course. My mind would freeze and even when I should have known the answer, I simply couldn’t access it. Long division was a mystery; fractions, an unsolvable puzzle. As for decimal points – well, placing them correctly was a lottery.
I know my experiences aren’t that unusual, not even for a biologist. Facility for numbers is rare enough even among biologists for it to remarked upon within the profession. Perhaps this derives from the origins of many of us as nature-lovers, biophiles, whose passion is the material, the animate, rather than the abstraction of numbers. Only through our training do we of necessity, become mathematicians of varying quality.
My days are now spent dealing with numbers, creating them, in the sense I’m doing what all science comes down to: measuring. At the moment, I’m essentially a glorified technician, although I love the technical part of my work, the sense of skill and specialised knowledge that it bestows. I like to glorify this work even further because that skill that I have honed over months of practice, is to my mind, a form of art. Art in the sense of knowing your subject so well, that deep sense of familiarity that it becomes a part of yourself, not second nature, but almost first nature. In essence, this work, this measuring, is artistry, in the sense that I’m drawing out some fundamental attribute, some essential quality, of birdsong. A musician can read notation, play an intrument, and can also hear an unfamiliar piece and transcribe it as notation. It’s not dissimilar to what I do. And the tool I do that with, the expression that emerges from this labour, is numerical. There are two major transformations: song, sound, becomes image, in the shape of the spectrogram; image gets measured and quantified, gets reformed as numbers.
The basic routine is this: I go through my recordings to identify the songs of suitable quality for further analysis. I number these. Then I randomly select a fraction of these numbers for the actual analysis. I bring up an image of the song. While I have preset the broad settings, each song usually requires some minor adjustment so that the facility the software has to automatically measure many of the song features can work effectively. In other words, the context has to be taken into account. There may be other birds singing in the background; a breeze may slightly blur the song’s syllables. I actually enjoy this bit. Above this spectrogram, a power spectrum shows the relative amplitude of each sound in the image. When the ‘automatic parameter’ tool shows, by a sequence of red marks, that it is perfectly aligned with the black bubbles of the power spectrum, I feel very satisfied. Now I have captured the number of syllables, their length, their intervals. Sometimes I have to use the eraser tool to wipe out extraneous noise that interferes with this alignment process, to isolate the sounds, the song, that I’m interested in. It’s a little like sculpture, so easy to get wrong. One slip of the hand, the mouse, and I’ve to start again, but finally I have it: a perfectly delineated species song, in black and white.
There is artistry in this process, where my own subjective skill and judgement comes in. The whole process, the tweaking, the isolating, reminds me of the preparation of a tissue sample for examination under a microscope. Without treatment, staining, without the foreknowledge, to an extent, that comes with deep familiarity with your material, you can’t make sense of the image. Your eye has to know what to look for, the shape, the form of it, that very physical form of learning. Things have to be made ready to see: made visible, measurable, quantifiable. That’s where the pleasure comes in: the software can’t do it alone. It has to be set up, and even then it still requires my input, my laboriously-acquired skill, to measure correctly what has to be measured. So I measure: length, various frequency attributes. I copy an image of the song into a worddoc, a before- and after- prep image if I’ve had to do a lot of work on it. And then onto the next one. Hundreds of songs. Scores of individual birds. The images of the songs are etched into my brain, as is their sound into my memory. With some, I even know the particular individual that was singing this particular version of the species song. It is a lot of work to acquire this knowledge. It’s also a privilege. I often think of our ancestors, who could never have accessed this level of detail, this different way of knowing, birdsong.
The results are scores of excel sheets, which are becoming, ultimately, the species master sheet, another form of extended phenotype. Columns and columns, rows and rows, of numbers. When I do some calculations to check them, and to derive some additional parameters, these new numbers cascade down the sheet. They grow and grow. My individual birds have vanished, are subsumed into the vast forest, this repertoire, of numbers. They sequester who knows yet what patterns. These data are the DNA of my findings. The shape of them is not yet emerged, will need the stats to transform them into meaning. That’s another thing I’m learning. Stats are not really about numbers. They’re about patterns, and patterns are something I understand. Stats iron out blips, so the shape of any pattern is made clear. They too make things visible. There is something entrancing about the process. Individuals, song, distilled to a huge matrix of numbers that still hide their secrets from me. I’ll be slogging at this acoustic analysis until Christmas, but already I’m excited about what lies ahead. It’s like an unopened present that’s just waiting for the right moment to be unwrapped.