February 21, 2020

The sCreen's sPell

February hasn’t been an easy month for me, and it’s looking increasingly unlikely I’ll make a major post by the end of the month. But to avoid letting the month go by without any sign of life, I’d like to share some observations that probably aren’t worth a whole article but may still be interesting to read.

I’m a fan of the song “Mad About The Boy,” written by Noel Coward. While looking for a version sung by Coward himself, I came across this strange cover by Peter Sellers, which was part of the music “The Magic Christian.”

There’s a lot to unpack here - the unholy mix between some kind of Californian English and Received Pronunciation, the melodramatic performance (par for the course considering the poor reviews for this movie), the ridiculous but era-appropriate panning of the voice hard left and right - plus copious amounts of vocal fry.

But what really caught my ear was Sellers’s pronunciation of “screen” (0:19) and “spell” (0:23)

Normally in English, a consonant cluster of /s/ plus a stop /k/, /p/, or /t/ results in the stop being unaspirated. This is in contrast to the usual situation when those consonants are at the start of a syllable, where they are aspirated. I’ve heard some English speakers impressionistically describe the word “spin” as “s” + “bin” - because /b/ sounds are unaspirated but /p/ sounds are not!

But no English speaker could say that Peter Sellers is saying “s + green” or “s + bell” in this song. He is fully aspirating the stop in “sc[h]reen” and “sp[h]ell.” Impressionistically, I almost feel as if the word has been split into two - how often is an aspirated stop the second member of a consonant cluster in English? Never - the phonotactics of English only allow aspirated stops to be preceded by /s/.

While here I’m describing it as if the “s” were causing the stops to lose aspiration, historically it was the other way around. At some point in the development of Germanic languages, what were originally voiceless unaspirated stops because voiceless aspirated stops. They contrasted with unaspirated voiced (or sometimes even voiceless) stops. You can hear this system in English, German, Swedish, and Icelandic. Dutch is an example of a Germanic language that doesn’t have this distinction - no aspiration.

But this change didn’t affect every instance of voiceless stops. One of the places where this change didn’t occur was a consonant was preceded by /s/, which functioned as a blocking environment. You can hear an example of this in the word ‘stop’/‘stoppen’/‘stoppa’ (all meaning stop) in English, German, Swedish, and Icelandic - the /t/ is unaspirated in these.

I don’t have any examples of a different language that allows /s/ + aspirated voiceless stop consonant clusters, but they’re probably out there somewhere.

For some further reading, here's a Master's Thesis on the origin of aspiration contrast in Germanic languages.


  1. David MarjanovićFebruary 21, 2020 at 5:57 PM

    Ancient Greek and Sanskrit come to mind. I wonder about Georgian and other languages with large inventories of aspirates and consonant clusters.

    Another, funnily enough, is English with a southern German accent! We had aspiration, but then we exaggerated it till it broke – that's the famous High German consonant shift –, and we haven't regained it. The rule that plosives following /s/ aren't aspirated is not intuitive enough for us to guess, and while I was taught to aspirate in one of my first English lessons, I was never taught not to aspirate after /s/ and had to figure it out much later.

  2. David MarjanovićFebruary 21, 2020 at 5:58 PM

    Oh, and, thanks for the thesis – I've downloaded it and will try to read it tomorrow!

    1. David MarjanovićFebruary 23, 2020 at 5:42 PM

      I've read it now. It's fine, and I especially like how it takes Kortlandt down several notches; but it completely lacks any mention of the work of Kurt Goblirsch, who read phonetic descriptions of some 500 extant Germanic dialects and built his amply published conclusions on that knowledge. It turns out Dutch isn't the only aspiration-free "voice language" in Germanic – there's an aspiration-free belt straight across southern Low and northern Central German. The western end could be blamed on Romance, as it usually is; the eastern end could be blamed on Slavic; but that leaves a lot in the middle. In the end, the easiest explanation for the geographic distribution of aspiration in Germanic is that aspiration is an areal innovation of North and northern West Germanic, carried south beyond the aspiration-free belt in the Migration Period, where it was exaggerated till it broke in the High German consonant shift. Goblirsch further thinks that the High German consonant shift (the "Second Sound Shift" as Grimm called it) was basically a repetition of Grimm's law (the "First Sound Shift"), and I can't think of any argument whatsoever against that.

  3. David MarjanovićFebruary 21, 2020 at 6:07 PM

    ...It's late enough that it didn't really occur to me I could listen to the video.

    screen is just weird: it has a [ʃ] in it. There's no aspiration, though. Or perhaps that's the aspiration, devoicing the /ɹ/ which has somehow been exaggerated into a fricative.

    spell does contain an aspiration... but that's a separate [h]. First the [p] is released, then the [h] starts.

    The [k] and the [p] are fortes, however. That's unusual after /s/ in English (but usual in German).