Last week, I talked a bit about the analysis process, which is pretty subjective. Listening through recordings and then classifying songs based on their song type means introducing a ton of biases. These biases include my experience (or, more often, inexperience) listening to birds in the field and any classification tendencies I might have. For example, I might count a repeated, two-note vocalization as a single song type, while others might count each note in the vocalization as belonging to separate song types.
This week, I’ll discuss some examples of difficulties that have arisen in analysis so far. There’s really no quick fix for these difficulties: each one has to be sorted out in its own way, and this greatly lengthens the analysis process. It all depends on how the bird is singing. If it’s taking time between songs, and making them clear cut, it can take half a minute to breeze through a minute-long recording. Other times, a five second fragment might take five minutes to analyze, not counting the mandatory break at the end to clear my mind of all the green lines, dots, and curves on the screen.
First off, the subjectivity is hopefully not as big of an issue as it may seem. Though it’s true that a different researcher would classify songs differently, there’s only one person doing analysis for this project. If the work was split in half with someone else, comparing mimicry frequencies would be problematic. As long as I do my best to apply the same analysis standard to every single recording, the calculated mimicry frequencies should be statistically comparable.
The primary difficulty is identifying what bird the mockingbird is mimicking, if it’s even mimicking at all. Mockingbirds have such a diversity of original songs that many of them end up being acoustically similar to other birds’ songs and calls. Differentiating between mimicry and original songs hasn’t been as easy as I hoped it would be. Then, there’s also the potential problem of imperfect mimicry. Mockingbirds are known to introduce variation into other bird’s songs, making their vocalization a hybrid of their original song and another bird’s song. For example, in this fragment, the middle five vocalizations sound similar to California Scrub-Jay calls in their rise in pitch, but have a less raspy and more whistly tone than actual calls would. The question is, count them as a mimicry event? Give the mockingbird 0.5 of the event and the scrub-jay 0.5 of the event, as other researchers have done? Or count all the calls as mockingbird vocalizations that simply happen to sound like scrub-jay calls?
In addition to this, I’m not even close to knowing every single song and call out there. Resources that let me compare recordings to sonograms and other recordings, such as Xeno-canto and Cornell’s All About Birds, have been a great help. Still, even if it’s obvious that a song is not a mockingbird original, it can take a lot of work to pinpoint which exact bird is being imitated.
The next problem is a bit embarrassing to admit to having. It involves background singing – other species of birds singing during the recorded mockingbird’s song. This makes it possible to mistake background singing for mimicry. In some cases, background singing is easy to sort out since the frequency and beginning of the phrase is completely off from what is expected for the mockingbird song. For example, here is an Oak Titmouse calling in the background (top vocalizations) of a Northern Mockingbird song (bottom).
Other times, it’s much harder to tell which bird the sound belongs to. For example, when listening to the fragment below in the field, the third vocalization sounded like a Western Tanager’s call. It would be quite surprising, but not impossible, for a mockingbird to mimic a tanager (which is uncommon in suburban habitats), so at home I took a closer look at the sonogram.
It reveals that there are two separate vocalizations: one upslurred and the other downslurred. It would be impossible for a mockingbird to produce both sounds simultaneously, so the conclusion is that there was an actual, real-life Western Tanager calling somewhere in the background. Which is pretty strange, since tanagers usually don’t migrate into the Silicon Valley until April, though some do overwinter here. But strange is more likely than impossible, so tanager it is!
In addition to background singing, there’s also other ambient noise.
Cars, trains, planes, humans talking, dogs barking, you name it. All of these make a recording much harder to analyze, since their sounds essentially mask the lower frequencies of the mockingbird song. Large portions of some recordings pretty much had to be discarded because of this background noise.
That’s all for this week. Next week will consist of surveys in the morning, and more analysis in the afternoon. Fun!