The spelling rules of English are by far the most challenging aspect of learning to read, of reading, and of writing. Anyone who has seen a child’s spontaneous writing has seen all kinds of creative misspellings. Adults, of course, are also not immune to spelling errors.
Spelling is so strange in the English language that the United States holds an annual spelling bee, a moderately high-stakes spelling contest, in which a word said out loud must be spelled letter-by-letter, getting increasingly tricky as the contest proceeds. Children who participate in spelling bees often memorize all kinds of unique rules for turning Greek, French, or Sanskrit words into English. For example, the Greek prefix <eu-> (meaning “good”, as in “euphoria”) sounds like “you”, or “cashmere” from Sanskrit is pronounced “cash – meer”, or knowing that an “o” sound at the end of a French word might be spelled <-eau>, as in the word “chapeau”, which is pronounced roughly like “sha po.”
Even adults find spelling challenging – for example, if you saw a written word like “pneumonia” without having heard it said out loud before, you might not know how to say it – similarly, if you’ve only heard it, that does not mean you will know how to spell it. The pervasiveness of autocorrect features—and their glitches—on smartphones and computers, and the difficulty in learning to type, also highlight just how difficult it can be to write a word right the first time. Some people’s phones even begin to learn how their users misspell words. Because our phones do not know the difference between a misspelling, an intentional misspelling, and a word it does not know, our phones memorize these words instead.
Thankfully, not all of English spelling and pronunciation is difficult, and even more gratifyingly, most languages of the world have much more systematic writing systems. Because of its idiosyncrasies, however, English is a great place to ask what factors influence people’s spelling.
There are a number of regularities in English spelling, since most of the time we know that <m> and <s> letters are said “m” and “s” out loud, or that the past tense of “stop” becomes “stopped” because we tend to double terminal letters when adding endings to them. One of the most famous rules, “<i> before <e> except after <c>”, is taught everywhere in English classes, and it is present in words like “receive” or “believe”—but note that “weird” does not follow this rule.
Spelling rules are supposed to allow us to translate between how a word sounds and the way it is written, so most researchers have focused on how the sound system – or the phonological properties of a word – can influence spelling. However, the reverse link also holds: In addition to the sound system of the language influencing spelling preferences, spelling regularities can also contribute to spelling. These regularities are known as graphotactic regularities.
For example, English is full of all kinds of unusual sequences of letters, such as the “ck” in the word “hammock”, or the state and river “Mississippi”. Nonetheless, people will typically avoid doubling consonants, as in one of my favorite words, co-occurrence, which people often spell as “co-occurence” or “co-ocurrence.” People will also prefer to write “bappic” rather than “bappick” when they are asked to spell a made-up word used in psycholinguistic experiments (known as nonce words). This is because people prefer spellings that conform to regularities rather than exceptions.
The vowels (a, e, i, o, and u) also matter for spelling. English also has the property of “short” and “long” vowels: In American English, “short” vowels include the vowels in “cat”, “bet”, “mitt”, “ostrich”, and “pun.” “Long” vowels are usually multiple sounds at once, often involving the letter “e” at the end of the word, like “mate”, “meat”, “mice”, “moat”, and “muse.”
But spelling rules can change based on the context, thereby representing a complex rule. One rule that does not get taught in school is that when we have written two vowels in a row, like in the word “moat”, the last letter (“t” in this case) does not get repeated when we add a new ending, like “-ing”. For example, making a moat would not be spelled “moatting”, but “moating”. Likewise, “moatted” looks odd while “moated” looks fine.
Disentangling the phonological and graphotactic contributions to spelling is difficult with familiar words. To circumvent this difficulty, researchers Rebecca Treiman and Sloane Wolter tested how people spell novel words in a study that was recently published in the Psychonomic Society journal Memory & Cognition. Treiman and Wolter tested 111 individuals from the surrounding community at their university, which is critical for understanding how spelling functions in the real world, not just in a sample of highly-educated college students. These participants also varied substantially in their spelling abilities, so this study can inform us about individual differences as well.
In the present experiment, an experimenter was present with the participants and pronounced the words next to the participants, so they would be able to use both visual and auditory cues to spelling. The nonce words that Treiman and Wolter created were combinations of different vowels (a, e, i, o, u, and sometimes y) and consonants (all the other letters, such as p, b, r, z). The consonant sounds were wedged in the middle of the word, which ended with sounds like “-er”, “-ed”, “-age”, “-ow”, and “-ee.” The researchers then tested whether the participants typed the middle sounds twice, or only once. For example, a word pronounced vaw-nid could be spelled in many different ways, ranging from “vannid”, “vaunid”, “vawnid”, to “vawnid”, “vawnnid”, “vaunid”, or “vanid.”
Previous research had found that spellers tended to avoid using two letters for a consonant (e.g. <nn>) after a long vowel (e.g. <aw>), and so Treiman and Wolter predicted that they would find the same pattern of results. The results of this experiment demonstrated exactly that – that despite this principle never being taught in schools, spellers did know not to double letters after long vowels.
An additional finding of interest was the strong correlation between spelling ability and a person’s spelling preferences. Treiman and Wolter found that the tendency to spell “vannid” or “vawnid” was greater, the greater a person’s spelling ability. Stated another way, people with less strong spelling skills tended to write “vawnnid” or “vanid” more often than “vawnid” or “vannid.” Surprisingly, only 26 of the 111 participants realized that only using one letter (<n>) after a long vowel (like <aw>) was a spelling principle in English, suggesting that these rules are implicitly learned, and rarely, if ever, explicitly taught in classes.
Why would there be such a strong relationship between a person’s spelling skills and their preferences for spelling? Treiman and Wolter hypothesized that statistical learning is behind these correlations. The more experience someone has with unusually spelled words, and the more often a person has to write unusual words, the more likely they are to learn these types of patterns. The fact that most participants did not explicitly know these rules suggests that this could be an easy win for educators. Instead of relying on learners to infer the tendency to double only one set of letters in a word, teaching them about these patterns could help people acquire English spelling more easily, with fewer headaches.
Psychonomic Society journal article featured in this post:
Treiman, R., & Wolter, S. (2018). Phonological and graphotactic influences on spellers’ decisions about consonant doubling. Memory & Cognition, 46, 614-624. DOI: 10.3758/s13421-018-0829-1.