Regular and Exception Words

English text is best described as a writing system with a deep orthography. What that means is there is not very good consistency between the way words are pronounced and the way they are spelled. In an alphabetic writing system, like the one used in English, phonemes in speech are encoded as letters in writing. In a shallow orthography, there would be something approximating one-to-one correspondence between phonemes and letters, but in English, every phoneme can be legitimately represented by several different letters.

Consider the long vowel sound /o/, as in GO. If English were a shallow orthography, the long vowel sound /o/ would always be encoded with the letter O, as it is in the word GO. So, GOAT, would be spelled GOT, THOUGH would be spelled THO, VOTE would be spelled VOT, TOE would be spelled TO, and TOW would also be spelled TO. Obviously, this isn't the way English works.

One might think, because of all of this inconsistency in English spelling-sound relationships, that there are no conventions or rules to English spelling, but that also is not the case. If that were the case, you would have no notion how to pronounce words you've never seen before (such as SPLECK, VEET, or PORTIVE). There are regularities and consistencies that allow competent readers to "attack" novel words. There certainly are spelling-sound conventions and regularities, but there are also a lot of words that deviate from these conventions to varying degrees.

Some have attempted to account for these conventions, and to formalize them in generalizations that explain the English spelling-sound relationships, so rather than trying to map all of the symbols that can be used to represent phonemes, these generalizations attempt to describe the structure of the word, and how the letters within the word relate to each other. So, for example, vowels are typically long when the word ends in a silent E (VINE, HOPE, PALE). That generalization, if true, would be a handy clue to help people to decode words, but that generalization is only true about 65% of the time with monosyllabic words (SOME, DONE, ONCE), and becomes even more vague and unreliable when polysyllabic words are examined (DETERMINE, CONCRETE, EXAMPLE). No matter how convoluted and contrived these generalizations are, they never account for all words - there are exceptions to every rule.

A different approach is to examine letter clusters that commonly go together in English words, and make generalities about those clusters. Rather than attempting to create generalizations such as "when a vowel is followed by two consonants, that vowel is pronounced as a short vowel" (fine for DUST, PICK, and PAST, but not for PINT, MOST, and FIND), some have observed that the generalizations are more consistent when they are made about specific clusters of letters. A cluster of letters is typically pronounced the same when that cluster appears in different words. There are some exceptions, but not as many as with the generalized principles described earlier, and while generalizations are fairly abstract, this level of examination is quite concrete (so exceptions are easier to grasp). For example, most words, both monosyllabic and polysyllabic, that end in EAT are pronounced with a long /e/ and a /t/ -- FEAT, MEAT, HEAT, SEAT, BEAT, PEAT, NEAT, PLEAT, CHEAT, DEFEAT, REPEAT, etc. There are some exceptions (SWEAT, CAVEAT, GREAT), but they are comparatively rare. It is reasonably safe to assume that novel words, never seen before, such as ZEAT or VEAT will be pronounced with a long /e/ and a /t/.

Human beings (and children in particular) are very good at detecting patterns, but we are not very good at applying rules. We know implicitly that words that contain letter patterns like YNX and ONT do not occur often in our language. They do occur (LYNX and FONT), but because we are sensitive to the frequency of letter patterns in the English writing system, we know those letter combinations do not occur very often - certainly not as often as ING or ATE. If humans were simple "rule appliers," we would not be sensitive to the frequencies of letter combinations - rule appliers would not notice that the letter combination EDO almost never occurs in English, but the letter combination ARE is quite common.

While it is possible to create a set of rules that fairly completely describes English spelling-sound relationships (Gough and Ehri suggest that a set of somewhat more than 600 rules would suffice), it does not follow that humans use those rules when reading. Describing the rule set for a writing system is very different from describing how people read.

The implications for teaching children to read are quite clear - children must learn to "read by analogy" as some researchers put it. They must use the words they are familiar with to help them in sounding out the words they are not familiar with. The teacher can help children to develop this ability by encouraging them to think of other words they can read that have the same letters, and thinking about how those words are pronounced. For example, if the child is struggling with the first sound in an unfamiliar word such as BIRD, the teacher could remind the child of familiar words that start with the same letter (and the same sound).

Further, the teacher can help the child to see that some words are "exception" words, and the teacher can focus explicitly on those exception words, familiarizing the child with those words so the child learns that they are not pronounced like other words that are spelled similarly. In word family lessons, children can learn that most words that are spelled similarly are pronounced similarly, and they can become familiar with the relatively few words that are not pronounced like the other words in the family (Pat Cunningham has produced a series of books full of activities that help children to see consistencies in letter patterns - check out Making Words, Making Big Words, and Phonics They Use).

To help teachers to focus on words that could potentially confuse children, and which teachers should address explicitly when necessary, we've assembled a list of exception words (words that contain letter clusters that are pronounced differently in this word than they are typically pronounced in other words). In looking over this list, some words may appear to be missing - words that have traditionally been labeled as irregular like SCHOOL and DUMB. These words are words which violate one of the common generalizations (or possibly a phonics rule), but which are none the less regular (SCH, for example, is almost always pronounced /sk/, as in SCHIZOPHRENIA, SCHEDULE, SCHEMATIC, and SCHISM, although there are rare exceptions such as SCHNORKEL and SCHLEPP). Also, one and two letter words were omitted from this list - their regularity is unclear.

The words are also divided into two lists. The first list is a frequency sorted list of words commonly found in children's literature taken from J. Lloyd Eldridge's "Teaching Decoding in Holistic Classrooms." Some of the words in his list were not considered to be true exception words (although they may have violated a phonics rule, their pronunciation is consistent with other words sharing the same letter clusters), and were dropped from this list.

The second list is a supplementary list of exception words that are not ordered or sorted in any way. Some of these words are not necessarily words that young children would be familiar with, but are included here in the spirit of creating a broad database of exception words.

This list also continues to grow - if you think of an exception word that is not in this list, type it here, and it will be added.

List 1 of 2: Exception words commonly found in children's literature sorted by frequency.

List 2 of 2: Other exception words being compiled by SEDL.

Word	Letters	Syllables	Frequency (per million)
were	4	1	3284
school	6	1	492
often	5	2	368
building	8	2	160
hour	4	2	144
blood	5	1	121
touch	5	1	87
build	5	1	86
broad	5	1	84
campaign	8	2	84
lose	4	1	58
minute	6	2	53
choose	6	1	50
bureau	6	2	43
wear	4	1	36
height	6	1	35
tongue	6	2	35
exhibit	7	3	25
vein	4	1	25
sweat	5	1	23
thorough	8	2	21
mortgage	8	2	17
shoe	4	1	14
pint	4	1	13
deaf	4	1	12
dialogue	8	3	12
gauge	5	1	12
tomb	4	1	11
steak	5	1	10
colleague	9	3	9
bosom	5	2	8
choir	5	1	8
chord	5	1	7
cough	5	1	7
rendezvous	10	3	7
sword	5	1	7
aisle	5	1	6
benign	6	2	6
pear	4	1	6
sew	3	1	6
tread	5	1	5
gourd	5	1	2
naïve	5	2	1
succinct	8	2	1
voila	5	2	1
womb	4	1	1

More about this resource:
Click here to view the listing for this item in SEDL's list of Free Resources.

Related resources:
You may be interested in the 25 SEDL products or 12 articles from SEDL Letter (SEDL's magazine) that are related to this product's subject, Strengthening Teaching and Learning in the Content Area of Reading and Language: Early Reading.

Join Our Mailing List