Allomorphy occurs when a single morpheme has multiple phonological realizations conditioned by phonological, morphological, or lexical context, as in English plurals (/s/, /z/, /əz/) or past-tense -ed. Understanding allomorphy integrates phonological rules, morphological structure, and lexical exceptions.
Catalog allomorphs in a language and determine conditioning factors (environment, morpheme class, lexical listing); distinguish regular rule-governed allomorphy from suppletive forms.
Allomorphy is not disorder but systematic variation; even exceptions often follow phonological principles and are learned as indexed to specific morphological contexts.
From your study of morpheme types, you know that morphemes are the minimal units of meaning or grammatical function — roots like *dog*, prefixes like *un-*, suffixes like *-ed*. From phonological systems, you know that sounds are organized into abstract categories (phonemes) that have predictable realizations depending on their sound environment. Allomorphy is where morphology and phonology collide: a single underlying morpheme can surface with different phonological shapes depending on context, and understanding those shapes requires both kinds of knowledge at once.
The English plural morpheme is the canonical example. We write it consistently as *-s*, but listen carefully: *cats* ends in /s/, *dogs* ends in /z/, and *buses* ends in /əz/. These are three distinct phonological forms — allomorphs — of the same morpheme. The conditioning is phonological: /s/ follows voiceless consonants (/t/ in *cat*), /z/ follows voiced sounds (/g/ in *dog* and all vowels), and /əz/ follows sibilant sounds (/s/, /z/, /ʃ/, /ʒ/, /tʃ/, /dʒ/) because two sibilants cannot be adjacent without a buffer vowel. This is phonologically conditioned allomorphy: the same distributional logic applies to the third-person singular -*s* (*walks/runs/pushes*) and the possessive -*'s*. Once you see the rule, three "different" endings collapse into one underlying morpheme with predictable surface shapes.
Not all allomorphy is this tidy. Suppletive allomorphy is the linguist's term for cases where the allomorphs share no phonological resemblance at all — they are simply learned as exceptions. *Go/went* is the English past tense with maximum suppletivity: nothing in the form of *went* is derivable from *go* by phonological rule. Similarly, *good/better/best* and *bad/worse/worst* show suppletive comparison. These forms must be stored in the lexicon as irregular paradigm entries, not derived by rule. Suppletive allomorphy is common cross-linguistically in high-frequency items, which fits with what we know about memory: frequent forms get memorized directly rather than computed.
Between rule-governed allomorphy and full suppletion lies a middle ground of morphologically or lexically conditioned variation. The English past tense -*ed* surfaces as /t/, /d/, or /əd/ (the same voicing logic as the plural), but some verbs take "strong" or irregular pasts (*sing/sang*, *drive/drove*) indexed to specific lexical classes inherited from Germanic. The theoretical debate is whether to capture these with abstract morphophonological rules (treating vowel alternations as triggered by an underlying abstract morpheme) or to list them as stored alternations. Modern approaches generally use Optimality Theory or similar constraint-based frameworks to handle the interaction between phonological well-formedness and morphological faithfulness, weighting constraints differently across languages to generate the observed patterns.