January 13, 2022

DWEEB ALERT:

How to win at Wordle using linguistic theory (David Shariatmadari, 11 Jan 2022, The Guardian)

It's helpful to understand what Wordle is mainly testing, and I think there are a couple of things: first, your knowledge of the frequency of individual letters in the English language - that is, how common they are (think of the value of letters in scrabble - "q" is 10 because it's harder to find words that use it, whereas "e" is 1). So it would be unwise, for example, to use "hyrax" as your first guess.

More interestingly, though, it probes your instinct for how letters can be combined. Very often you find yourself thinking: I've got an "o" an "r" and a "t". Do many English words end in "o"? Is "t" a good bet to start the word with? Should it be followed by "r", or should there be a vowel in between them?

OK, so most of us will have a sense of how to answer these questions already, because we use words every day and have a gut feeling for which sequences are possible where, and which aren't allowed at all ("ng", for example, is pretty common in English, but never at the beginning of a word, and "lng" never appears anywhere).

But linguistics can help a bit. There is actually a whole branch of it that looks at the way sounds enter into sequences, and it's called phonotactics. Each language has its own phonotactic constraints - such as the rule that says "ng" can't start a word in English (it can, and does, in Māori and Swahili). Then there are rules that determine the possible order of consonants in a syllable: "tr" is fine at the beginning of a syllable but not the end, and the reverse is true for "rt". "Bl" and "lb" follow the same pattern.

That's not actually a coincidence. There's a mechanism behind it called the "sonority sequencing principle" or SSP. Certain sounds, often "hard" ones like "t", "b" or "g", are not very sonorous, or resonant. Softer sounds like "r", "l" and "w" are a bit more sonorous, and vowels are very sonorous. In a syllable that contains consonant clusters, the less sonorant sounds will tend to appear at the beginning, there will be a sonority peak in the form of a vowel, and then a gradual downward slope back to something less sonorous. "Blurb" is a lovely example of this, as is "twerk" or "plump". "Rbubl", an impossible word, violates SSP because "r" is more sonorous than "b", and "b" is less sonorous than "l". Within this framework, some combinations of sounds are seen way more often than others: "tr" and "pr" are everywhere, but "dw" is pretty unusual, appearing in "dwell" and "dweeb", with "dz" rarer still - "adzes" may be the only Wordle-able example.

Posted by at January 13, 2022 8:37 AM

  

« THERE'S NO SUCH THING AS QUALITY: | Main | IT'S A DEFLATIONARY EPOCH: »