linguistic gibberish (ha!)

I think that a lot of interesting results about our phonology could be obtained by generating strings of gibberish (using only sounds that occur in the speakers’ language, such as English, but that have no meaning when put together) and forcing speakers to guess at interpreting them.

For example, a crazy old man on the bus today spouted off a string of what was more or less gibberish to me but for some reason I at first thought I heard, "There’s a tornado out there."

Really, the only segments I distinctly heard were t r n r th (th as in bathe). Yet I filled in the rest and came up with a meaningful sentence. It’s already accepted that we don’t have to hear every single sound of a word to process its meaning, but what portion do we need to hear? 50%? I definitely heard less than that in my example. I think it’d be very interesting to see, using these gibberish sentences, what portion of the sounds speakers actually use to come up with meaning. We could then infer what portion of sounds they actually need to process when hearing speech to determine words. And then also, how do native speakers, say of English, differ from second language learners? I feel like in English I can generally guess at a meaning only hearing a small part of the word but in Spanish I need to hear most of the sounds to not be confused even though I speak Spanish quite well. And I bet that at the beginning stages of learning, second language learners think that they need to hear every single sound to interpret a word but in reality they could hear not quite all (but would still need most) sounds and guess the word if they were pushed to do so.

People also do this a lot in songs– when they can’t make out the words because of instrumentals or whatnot, they fill in the meaning, which is why there are so many misheard song lyrics ("Revved up like a douche, another runner in the night…") So once again, what percentage of a word do they need to distinctly hear to determine which word it is?

I also think it’s interesting how much variation we allow. For instance, I thought I heard crazy man say the vowel sound in hEr, but I credited him with saying the sound in Or to get tornado instead of ternado. In fact, I only distinctly heard the consonants, so I was basically allowing full variation for all the vowels. That is a lot of variation. Would I have similarly assumed he meant /t/ in tornado if I’d heard /d/ instead? Possibly, because /t/ and /d/ are phonemic partners they often pattern together. But would I have accepted /k/ for /t/, since /k/ is also extremely similar? Probably not. So analyzing interpretations of gibberish could tell us how much deviation we allow from the "ideal" target pronunciation while still keeping the meaning.

I think it’s interesting that in this case what I distinctly heard were the consonants, because I had previously hypothesized vowels to be the most salient type of sound when trying to comprehend song lyrics that were hard to understand. However, it should be considered that vowels are often held for a very long time in music compared to actual speech. And also, maybe it’s just that the most distinctive vowels /i, u, a/ are extremely salient but not the ones that are more similar. Those middle vowels tend to get muddled a lot.

I wonder if anyone has done gibberish studies like these. I don’t think I’ll be doing them anytime soon because I’m really not that into phonology compared to other linguistic fields. I also wonder if this is the kind of thing that would be appropriate to go into my professor’s office and just start chatting about or if she is too busy for that.

And I also know that probably no one is going to read this all and fewer people are going to care about what percentage of sounds are salient to speakers but, well, I just wanted to get it all down while I was thinking about it.