What percentage of English words have predictable spellings?
AI Overview
Approximately 84% to 87% of English words have predictable spellings based on regular sound-letter patterns. While only about 50% follow strictly regular, phonetic spelling, another 34-37% are consistent except for one sound, usually the vowel. Truly irregular words, which cannot be predicted, constitute only about 4% of the language.
Key Findings on Spelling Regularity:
- Predictable (50%): Words that follow standard phonetic rules, such as “cat,” “pitch,” or “jump”.
- Mostly Predictable (37%): Words that are regular except for one feature, such as “knit” (where the ‘k’ is silent) or “boat” (where the ‘oa’ makes the ‘o’ sound).
- Irregular (4%): Words that do not follow established patterns.
These statistics suggest that, contrary to popular belief, English spelling is largely rule-based. When considering word meaning, origin, and spelling conventions (e.g., “ck” vs. “k”), the vast majority of words can be spelled accurately by applying established patterns.
~~~~~~~~~~~~~~~~~~~
Practice is essential. No one denies the importance of reading or writing, but their effects on spelling are minimal. Think how many children are good readers but cannot spell properly. Reading and writing do not *cause* good spelling, they only enhance it. Such indirect (incidental) learning is only temporary; without explicit attention to and practice with these words the long-term retention is lost. (Henderseon, 1981; Horn T., 1969)
~~~~~~~~~~~~~~~~~~~~
How Words Cast Their Spell: Spelling Is an Integral Part of Learning the Language, Not a Matter of Memorization by R. Malatesha Joshi, Rebecca Treiman, Suzanne Carreker, and Louisa C. Moats
https://education.ufl.edu/patterson/files/2019/04/JoshiTreimanCarrekerMoats2009.pdf
Quotes taken from the above:
- Spelling is not arbitrary
- Researchers have estimated that the spellings of nearly fifty percent of English words are predictable based on sound-letter correspondences that can be taught. And another 34 percent of words are predictable except for one sound.
- Research has found language-based spelling instruction (e.g., that focuses on sound-letter correspondences) to be more effective than instruction that relies heavily on visual memorization of words (e.g., that uses flashcards).
- Spelling, therefore, is a window on what a person knows about words.
- Paraphrased: Spelling ability correlates with reading comprehension (i.e. improving the former improves the latter)
- Students who spell poorly write fewer words and write compositions of lower quality. In addition, nonautomatic spelling drains attention needed for the conceptual challenges of planning, generating ideas, formulating sentences, and monitoring one’s progress.
- To the well-known linguists Noam Chomsky and Morris Halle, English is a “near optimal system for lexical representation.” Cited by Richard L. Venezky, “From Webster to Rice to Roosevelt,” in Cognitive Processes in Spelling, ed. Uta Frith (London: Academic Press, 1980), 9–30.
- Isn’t mastery of correct spelling within the reach of every computer user? Not really. Spell checkers do not eliminate the need to learn to spell accurately. When we used a computer spell checker for the sentence “The bevers bild tunls to get to their loj,” the spell checker gave correct spellings for “bevers” (beavers) and “bild” (build). However, the spell checker did not come up with the words needed to replace “tunls” (tunnels) or “loj” (lodge). Instead, for “tunls” it provided tuns, tunas, tunes, tongs, tens, tans, tons, tins, tense, teens, and towns. And for “loj,” it provided log, lot, lox, loge, look, lost, lorid, load, lock, lode, lout, lo, lob, lose, low, and logs. The fact is, computer spell checkers are mainly a tool for correcting typos. They are helpful for those who are reasonably good spellers, but they cannot compensate for poor spelling. Further, computer spell checkers cannot be relied on with homophones. For instance, a spell checker cannot correct the errors in the sentence “Your sure glad to no” for “You’re sure glad to know.” It also misses errors such as “meet” for “meat” and “week” for “weak.”