Chickpea has gotten pretty good at reading CVC short vowel words, as well as CVCE words, and she’s gotten good at most consonant digraph/blends like ST, CH, PL, etc. But vowel digraphs (or what people sometimes call vowel teams) are still confusing to her. I was trying to figure out in what order it makes sense to introduce them, and I was curious about the absolute and relative frequency of various pairs, as well as how consistent their pronunciation is. I looked around for some statistics and found this website , which reports on an analysis of the top 3000 most frequent words in English. It lists how reliable (or consistent) the pronunciation of each digraph is. I’m sorting them here from most to least consistent. I’m also comparing to the sounds listed for the basic phonograms on the Logic of English page, which as I understand it are supposed to be sorted from most to least frequent.
Extremely consistent:
- oy -> /oy/ (boy) 100%
- oi -> /oy/ (join) 100%
- aw -> /aw/ (saw) 100%
- ee -> /E/ (feet) 96%. One common exception is been.
- ay -> /A/ (play) 96%. One common exception is says.
- oa -> /O/ (coat) 95%. One common exception is broad.
Pretty consistent, with one dominant pronunciation:
- ew -> /OO/ (flew) 88%; /U/ (few) 19%. LOE teaches both sounds. Other /U/ words include pew, skew, phew, view, curphew, and nephew. I think the percentages add up to more than 100% because some words (like “new” and “dew”) can have either pronunciation, depending on a dialect’s degree of yod-dropping. One common exception where ew says neither /OO/ or /U/ is sew.
- au -> /aw/ (cause) 79%, but I feel like in common words kids are likely to encounter it’s much more consistent than this percent implies. The LOE program only teaches the /aw/ sound for this digraph. Two common exceptions are laugh and because. But note that LOE teaches “augh” as a separate phonogram. I’m not sure why, given that it almost always makes the sound /aw/, with the exception of laugh and its derivative words.
- ey -> /E/ (monkey) 77%, but in short words it usually makes (/A/, they). And, indeed, the LOE program teaches both sounds, and actually lists the (/A/, they) sound as more frequent than /E/. One common exception where ey makes neither a long A nor a long E sound is the word eye. Also geyser.
- Are there patterns regarding when ey makes a long /A/ vs. a long /E/ when it appears at the end of a word?
- ey = /A/ in short (3- to 4-letter) 1-syllable words: hey, they, grey, prey, whey (exception: key). Note that if a new reader were to pronounce these with a long /E/ sound, they would often be confusing them with other words that they should most likely know (he, the, we…)
- ey = /A/ in 2-syllable words with stress on the 2nd syllable: obey, convey, survey (the verb)
- ey = /E/ in 2-syllable words with stress on the 1st syllable: money, honey, kidney, alley, journey, valley, barley, parsley, monkey, donkey, turkey (exception: survey, the noun)
- Are there patterns regarding when ey makes a long /A/ vs. a long /E/ when it appears at the end of a word?
- ai -> /A/ (rain) 75%, occasionally it makes a schwa sound (like in captain) but this usually only happens in a multiple-syllable word in an unstressed syllable. The LOE program only teaches the long /A/ sound for this digraph. Two common exceptions are said and again. That said, even if a new reader pronounces these words with a long A sound, most likely the context will still make it clear what the word is.
Less consistent, with at least two common pronunciations. (I’m not sure why the percentages don’t add up to 100%?)
- ow -> /O/ (snow) 68%; /ow/ (how) 32%. The LOE teaches these two sounds in the opposite order, oddly. Maybe because in short words kids encounter the /ow/ sound is more common? Let’s see, if we ignore plurals and -ed verbs, then looking at these word lists:
- For 3-letter words ending in ow the situation is pretty even. We have 6 /ow/ words (how, now, cow, wow, vow, bow [bending]) and 5 /O/ words (low, row, bow [ribbon], tow, sow).
- For 3-letter words not ending in ow we have one /ow/ word (owl) and two /O/ words (own and owe).
- For 4- to 5-letter words ending in an ow we have only 4 /ow/ words (brow, plow, meow, and allow) vs. 15 /O/ words, (know, show, grow, flow, slow, snow, blow, glow, crow, stow, throw, arrow, widow, elbow, and below).
- For four- and five-letter words not ending in an ow we have the reverse, with more /ow/ than /O/ words. We have 20 /ow/ words (down, town, gown, howl, fowl, power, brown, crowd, crown, tower, towel, clown, drown, frown, rowdy, vowel, dowry, growl, prowl, scowl) vs. only 3 /O/ words, two of which are past-tense verbs (bowl, flown, thrown).
- ea -> /E/ (seat) 64%; /e/ (head) 17%.
- The LOE program also teaches a third sound (/A/, steak) for this digraph. It’s pretty rare though. On a quick search (ignoring “ear” words) I only found great, steak, and break, all of which seemed to have somehow escaped the great vowel shift. I’d consider that an exception rather than a rule.
- Note that LOE teaches “ear” as a separate phonogram, with the sound /er/. But looking at some 3- to 5-word examples I see 15 short /ear/ words (ear, year, near, hear, fear, rear, gear, sear, tear–noun, clear, beard, shear, spear, smear, teary) but only 7 /er/ words (earl, earn, early, learn, earth, pearl, yearn), along with 5 short /air/ words (wear, bear, tear–verb, pear, swear), and 1 /ar/ word (heart).
- Are there patterns regarding when ea makes a short /e/ rather than a long /E/?
- “ead” words seem to always be /e/ words: head, dead, read, bread, spread, thread, instead, already, meadow
- other /e/ words it’s not as clear what the pattern is: deaf, sweat, threat, meant, dreamt, heavy, weather, feather, measure, pleasure, weapon, ocean, breakfast
- ia -> /E/a/ (piano) 54%; /u/ (Asia) 46%
- ui -> /i/ (build) 53%; /U/ (fruit) 24%. The LOE program only teaches the /OO/, fruit sound. I looked at the words containing ui on this page and on this page, and ignoring qu words the only relatively common short words I see are suit, ruin, fruit, juice, bruise, and cruise. Other than build/built the /i/ sound seems to occur almost entirely in qu words, in which case it’s not really a ui digram, it’s just a qu+i, so I wonder if these percentages are wrong. Other exceptional ui words: suite, guide, biscuit. Apparently gui, like qui, is considered a special case.
- ei -> /A/ (rein) 50%; /E/ (either) 25%. The LOE program actually teaches a third sound /I/, as in feisty. Not sure why it’s not shown here. Note that LOE teaches “cei” as a separate phonogram making the sound (/s/E/, receive), and “eigh” as a separate phonogram making the sound (/A/, eight) and /I/, height). Not sure why these graphemes get special treatment, given that their sounds are just a subset of the sounds of “ei”?
- oo -> /OO/ (boot) 50%; /oo/ (book) 40%. The LOE program teaches a third sound, (/O/,floor). But maybe that only occurs with an r? Two common exceptions are blood and flood.
- ie -> /E/ (field) 49%; /I/ (tied) 27%. The LOE program actually only teaches the (/E/,field) sound. Maybe the other sound only occurs with suffixes?
- oe -> /O/ (toe) 44%; /OO/ (shoe) 33%; /u/ (does) 22% [only 9 words in sample] The LOE program only teaches /O/,toe and /OO/,shoe.
- ou -> /ow/ (out) 43%; /u/ (touch) 18%; /U/ (your) 7%. LOE actually teaches 5 sounds for this digraph! /ow/, house; /O/, soul; /U/, group; /u/, country; /oo/, could; It also teaches ough as a separate phonogram. I look at some patterns for “ou” below.
But that’s just looking at consistency. What about frequency? The best resource I could find is this article by Edward Fry “Phonics: A Large Phoneme-Grapheme Frequency Count Revised.” They analyzed a corpus of 17k+ words in which each word was only counted once, even if it occurred much more frequently. Each word was divided into phonemes and the corresponding grapheme (spelling). I extracted the data for vowel digraphs and diphthongs (like aw, oy, ew, ey…). Each digraph can be classified based on its frequency in the corpus (in this case frequency is defined as the number of distinct words it occurred in). Here’s a list of the most frequent vowel/diphthong digraphs:
- ou 622: 366 (/u/, double), 227 (/ow/, out), 29 (/OO/, you)
- ea 398: 245 (/E/, eat), 139 (/e/, head), 14 (/A/, break)
- oo 287: 173 (/OO/, moon), 114 (/oo/, look)
- ee 249: (/EE/, bee) — an exception is been, but it’s not listed in the table?
- ow 236: 124 (/O/, own), 114 (/ow/, owl)
- ai 223: 208 (/A/, rain): 15 (/i/, captain)
- au 146: (/aw/, auto)
- ay 131: (/A/, say)
- oa 126: (/O/, goat)
- oi 92: (/oy/, oil)
- aw 75: (/aw/, awful)
- ew 60: (/OO/ or /U/ as in new/flew or few)
- ey 54: 40 (/E/, money), 14 (/A/, they).
- oy 48: (/oy/, toy)
Putting these two together, we can classify the digraphs based on how frequently each occurs and how consistent its pronunciation is. Here’s a rough breakdown, in the order I think it would make sense to teach them:
- High frequency (125-250 words), consistent (95-100%), long vowel sounds: ee, ai, ay, oa (I’m including ai in this list although technically it’s only 75% consistent because I think in one-syllable kid-friendly words it is much more consistent.)
- High frequency (125-250 words), fairly consistent (75-95%): au (Note that although au is only listed above as 79% consistent, in words kids encounter it seems to be much more consistent than that.)
- High frequency (125-250 words), somewhat consistent (67-75%): ow.
- Medium frequency (45-100 words), consistent (95-100%): oi, aw, oy
- Highest frequency (275-650 words), inconsistent (<65%): ou, ea, oo
- Medium frequency (45-150 words), fairly consistent (77-88%): ew, ey (Note that although ey is listed as only 77% consistent above, it’s more regular than it seems because in one-syllable words it usually says /A/ and in multi-syllable words it usually says /E/)
- Very infrequent (<45 words), inconsistent (<65%): ia, ui, ei, ie, oe
This page suggests teaching vowel teams that correspond to long vowel sounds first (ay, ai, oa, ee, etc.)
This page suggests teaching vowel teams in the following order. Note that they don’t usually recommend teaching different phonemes for the same grapheme back-to-back. Instead they break them up:
- ai (rain), ay (say), ee (bee), ea (eat), oa (goat), oe (toe), ow (snow), oo (food and book), ow (plow), ey (valley), ou (ouch)
- oi (boil), oy (boy), ie (piece), ie (pie), ea (bread), ea (steak), ou (soup), au (August), aw (saw)
- ew (few), ew (grew), ei (ceiling), ei (vein), eigh (eight), ue (rescue), ue (true), eu (Europe), eu (sleuth), ui (juice)
- ou /u/ (touch)
Finally, are there patterns regarding when ou makes which sound? The primary sound of “ou” is /ow/:
- ou = /ow/: loud, pout, about, found, round, sound, house, cloud, south, proud, couch, flour, doubt, count, mouth, bound, ounce
If we ignore “ough” words, which LOE considers a separate phonogram, then can we still find some predictable patterns where “ou” makes a sound other than /ow/?
- When does ou=/oo/?
- ould = /oo/: would, could, should
- When does ou=/O/?
- our = /or/: your, tour, four, pour, court, yours (exceptions: our, hour)
- ou=/O/: soul (why?)
- When does ou=/u/?
- ous at the end of a word = /us/: serious, nervous, famous
- ou = /u/: young, touch, country, trouble (I don’t see any pattern here?)
- When does ou=/OO/?
- ou = /OO/: you, soup, group, wound, route, youth (I don’t see much of a pattern here?)
Then there are the crazy “ough” words:
- ought = /o/: ought, thought, bought, sought, brought
- ough = /uf/: rough, tough, enough
- ough = /of/: cough
- ough = /O/: though, dough
- ough = /OO/: through
Pretentious Mama 