KPEERI • FOUNDATIONAL CONCEPTS

Identifying Irregular Words — 2.b. identify phonetically irregular words

How to recognize words whose spellings defy predictable letter-sound correspondences and why they matter for structured literacy instruction.

Historical Context & Motivation

The Knowledge and Practice Examination for Effective Reading Instruction (KPEERI) asks educators to demonstrate mastery over the structure of English orthography, and one recurring competency is the ability to isolate phonetically irregular words—lexical items whose spellings cannot be fully decoded through the standard grapheme-phoneme correspondences a student has been taught. The historical need for this skill arises from the deeply layered nature of English, which borrows orthographic conventions from Old English, Norman French, Latin, and Greek, producing a writing system that is only partially transparent. Because roughly 50 percent of English words are decodable through regular patterns while a substantial minority resist those patterns, teachers must be able to flag the exceptions so they can be taught through targeted strategies rather than sounding-out alone.

1066
Norman Conquest
French scribes reshaped English spelling, introducing conventions (e.g., the ‘ou’ in ‘could’) that conflicted with native sound patterns and seeded many of today's irregular words.
1476
Caxton's Printing Press
Early printing froze spellings in place even as pronunciation continued shifting during the Great Vowel Shift, cementing mismatches like ‘one’ and ‘have’.
1955
Why Johnny Can't Read
Rudolf Flesch reignited the phonics debate, prompting educators to distinguish systematically between decodable words and true exceptions.
2000
National Reading Panel
The panel's findings codified systematic phonics and explicit instruction of high-frequency irregular words, shaping today's KPEERI competencies.

The central question this competency addresses is deceptively simple: given a word, can the instructor determine whether it can be sounded out using taught correspondences, or whether one or more of its parts must be learned as an exception? Answering this precisely—rather than intuitively—is what separates a structured-literacy practitioner from a casual reader.

Core Principles & Definitions

A word is classified as phonetically irregular when at least one grapheme within it maps to a phoneme in a way that violates the predictable, high-utility correspondences a reader has been taught. Crucially, irregularity is relative to the scope of instruction: a word like ‘chef’ is irregular for a beginner who knows only /ch/ as in ‘chip,’ but regular for an advanced student who has learned the French ‘ch’ = /ʃ/ pattern. This scope-dependence is the principle most frequently tested on the KPEERI.

1

Grapheme-Phoneme Correspondence

A regular word is one in which every letter or letter-team represents its most common, taught sound. Irregularity begins where this mapping breaks down.
2

Partial vs. Total Irregularity

Most ‘irregular’ words are only partially irregular. In ‘said,’ the /s/ and /d/ are regular; only ‘ai’ = /ɛ/ is the exception.
3

High-Frequency Load

Many irregular words (the, of, was, have) appear with enormous frequency, so mastering them yields disproportionate reading fluency gains.
4

Instructional Scope Dependence

A word's regularity status shifts as a student's repertoire of correspondences grows. Classification must always specify the assumed knowledge base.
KEY TAKEAWAY
Think of decoding like a compression codec that reconstructs sound from a stream of bytes. Regular words decompress cleanly using the standard lookup table. An irregular word contains a corrupted byte the codec cannot resolve—so the reader must store that specific segment as an explicit exception in memory, much as an engineer hard-codes an edge case the general algorithm cannot handle.

Visual Explanation

The diagram below maps two words onto their component graphemes and phonemes. The word ‘lamp’ decodes cleanly, with each grapheme resolving to its taught sound. The word ‘said’ contains a single irregular grapheme, ‘ai,’ which is expected to produce /eɪ/ but instead yields /ɛ/. Notice how the irregularity is localized rather than global.

The Classification Mechanism

Although this competency is not numerical, it operates on a rule-governed decision procedure that can be expressed almost algorithmically. For each word, the instructor segments it into graphemes, retrieves the expected phoneme for each, and compares that prediction against the word's actual pronunciation. If every prediction matches, the word is regular; if any grapheme fails, the word is irregular.

REGULARITY CONDITION
Regular(w) ⟺ ∀ gᵢ ∈ w : P(gᵢ) = φᵢ
Here w is the word, gᵢ is the i-th grapheme, P(gᵢ) is the taught (predicted) phoneme for that grapheme, and φᵢ is the actual phoneme in the word. A word is regular if and only if the predicted phoneme equals the actual phoneme for every grapheme.
IRREGULARITY INDEX
I(w) = ( count of gᵢ where P(gᵢ) ≠ φᵢ ) / ( total graphemes in w )
The irregularity index quantifies how much of a word deviates. For ‘said,’ one of three graphemes is irregular, giving I = 1/3 ≈ 0.33. A fully regular word has I = 0, and a word like ‘of’ (where both graphemes deviate from taught values) approaches I = 1.
Watch the Scope Variable
Because P(gᵢ) depends on what has been taught, the same word can have a different irregularity index for different learners. Always state the assumed phonics scope before classifying.

A Taxonomy of Irregular Words

Irregular words are not a monolithic category. Sorting them by the source and locus of their irregularity helps instructors choose the right teaching strategy. The table below organizes common exception types, each of which appears on structured-literacy assessments.

Common categories of phonetically irregular words.
TypeExampleIrregular Element
Vowel exceptionsaid, been, friendVowel grapheme yields an unexpected sound
Silent-consonantcould, knight, combA consonant grapheme has no phoneme
Loan-word patternchef, ballet, genreForeign correspondence not yet taught
Wholly irregularof, one, wasMost or all graphemes deviate
Schwa-driventhe, a, toVowel reduces to /ə/ in unstressed use

Worked Example

Classifying the word ‘could’
1
Step 1 — Segment into graphemesBreak ‘could’ into its constituent graphemes: c – ou – l – d. Note that ‘ou’ functions as a single vowel-team grapheme, and the silent ‘l’ is retained as a grapheme placeholder.
c / ou / l / d
2
Step 2 — Retrieve taught phonemesPredict each grapheme's most common taught sound: c = /k/, ou = /aʊ/ (as in ‘out’), l = /l/, d = /d/.
P = /k/ /aʊ/ /l/ /d/
3
Step 3 — Compare to actual pronunciationThe actual pronunciation of ‘could’ is /kʊd/. Comparing unit by unit: c = /k/ matches; ou is realized as /ʊ/, not /aʊ/ — a mismatch; l is silent (no phoneme), a mismatch; d = /d/ matches.
φ = /k/ /ʊ/ (∅) /d/
4
Step 4 — Apply the regularity conditionBecause at least one grapheme fails the match (in fact two do), the regularity condition ∀gᵢ : P(gᵢ) = φᵢ is violated. ‘Could’ is therefore phonetically irregular.
Irregular(could) = TRUE
5
Step 5 — Compute the irregularity indexTwo of four graphemes deviate, so I(could) = 2/4 = 0.50. This moderately high index tells the instructor that ‘could’ needs targeted exception instruction rather than pure decoding, and it belongs to both the vowel-exception and silent-consonant categories.
I(could) = 0.50

Strengths & Limitations of the Approach

The grapheme-by-grapheme classification method is powerful but bounded. Understanding where it excels and where it falters keeps instructors from mislabeling words and misdirecting instruction.

Trade-offs of grapheme-level classification.
StrengthLimitation
Yields a reproducible, defensible classification rather than a gut judgment.Requires an explicit, agreed-upon scope of taught correspondences to be meaningful.
Localizes irregularity to specific units, enabling precise instruction.Dialect and accent variation can change which phonemes count as ‘actual.’
The irregularity index prioritizes which words need the most support.Morphologically complex words (e.g., ‘signature’ vs. ‘sign’) may look irregular but are explained by roots.
IN CONTEXT
Classifying irregular words is analogous to static code analysis in software engineering: a linter flags patterns that violate the language's conventions, but a truly rigorous review still requires knowing which conventions the team has adopted and which apparent violations are legitimate idioms. Likewise, morphology and etymology often ‘explain away’ surface irregularity, so the skilled instructor treats the index as a triage tool, not a final verdict.

Connection to Advanced Theory

Simple binary classification is a doorway to richer models of orthographic depth and lexical processing. The table contrasts the foundational skill with the more sophisticated frameworks that researchers and advanced practitioners employ.

From foundational classification to research-grade models.
Foundational SkillAdvanced Framework
Binary regular/irregular labelContinuous consistency modeling (probability a grapheme maps to a given phoneme across the lexicon)
Grapheme mismatch countingDual-route and connectionist reading models (lexical vs. sublexical pathways)
Scope-dependent classificationOrthographic depth hypothesis comparing English to shallow orthographies like Finnish
Treating exceptions as memorized wholesMorphophonemic analysis revealing hidden regularity through roots and affixes

As you progress, the crisp yes/no verdict softens into a probabilistic view in which every grapheme carries a distribution over possible phonemes, and ‘irregularity’ becomes a matter of low consistency rather than outright violation. This continuous perspective ultimately connects the KPEERI competency to computational models of reading and to cross-linguistic research on how orthographic transparency shapes literacy acquisition.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain why the statement ‘a word is either regular or irregular, full stop’ is imprecise from a structured-literacy standpoint.
PROBLEM 2BASIC
Classify the word ‘cat’ using the regularity condition, assuming standard beginning-phonics correspondences.
PROBLEM 3INTERMEDIATE
Compute the irregularity index for the word ‘knight’ and identify which category or categories it belongs to.
PROBLEM 4APPLIED
An instructor is sequencing a lesson and must decide which of these three words to teach first as a memorized exception: ‘of,’ ‘been,’ or ‘comb.’ Use irregularity indices and frequency to justify a choice.
PROBLEM 5CRITICAL THINKING
The word ‘sign’ appears irregular (the ‘g’ seems silent), yet a morphophonemic analysis argues it is not truly an exception. Construct that argument and discuss what it implies about relying solely on the irregularity index.

Summary

A phonetically irregular word contains at least one grapheme that fails to map to its taught phoneme. Classification proceeds by segmenting the word, retrieving each grapheme's predicted sound, and comparing it to the actual pronunciation; the regularity condition holds only when every prediction matches. Because most exceptions are only partially irregular, the irregularity index I(w) locates and quantifies deviation, guiding targeted instruction.

Regularity is always scope-dependent, shifting as a learner's repertoire grows, and surface irregularity can dissolve under morphophonemic analysis. Combined with word frequency, these tools let instructors prioritize which exceptions to teach first, connecting a foundational KPEERI competency to advanced, probabilistic models of reading.

Varsity Tutors • KPEERI • Identifying Irregular Words — 2.b. identify phonetically irregular words