Zero Morpheme

When a paradigm shows systematic contrast in a feature, but one form has no overt marker, analysts posit a zero allomorph to maintain paradigmatic regularity. English noun plurals: cat → cats (overt -s), dog → dogs, sheep → sheep (zero). The plurals contrast with singulars syntactically (verb agreement, determiner choice). Postulating sheep + ∅ rather than treating sheep as exceptional preserves the analysis that English nouns are inflected for number. Without the zero, the system requires special exception lists. Bloomfield's 1933 Language tentatively introduced zero forms; structuralists from Hockett (1954) onward formalized them.

A zero morpheme is structurally present in a paradigm slot but has no phonological exponent. No morpheme means the slot does not exist in the analysis. English uninflected forms — "I run" — carry tense and person/number (non-third-singular present); the morpheme is zero, not absent. Conversely, "the" carries no number marking — there is no number slot for English determiners. The distinction matters for theoretical analysis. Generative morphology (Halle 1990, Anderson 1992) tightens criteria for zero morphemes — present only when paradigmatic contrast demands it.

Many. Russian nouns mark case morphologically; the nominative singular is often zero (stol "table" = stol-∅). Mandarin nouns typically lack overt number marking — singular and plural both zero, with quantifiers or context disambiguating. Hebrew construct state nouns lack overt marker (compared to absolute state, see McCarthy 1986 on root-and-pattern). Mohawk pronoun marking includes zero forms in some persons. Maori, Tagalog, and many other languages use zero for one of two paradigmatic values. The pattern is widespread cross-linguistically; languages tend to mark some categories overtly and leave others zero.

Distributed Morphology (Halle and Marantz 1993) treats morphemes as feature bundles inserted at terminal nodes; phonological exponents are inserted late by Vocabulary Insertion. A morpheme can have a phonologically null exponent (zero allomorph). The framework dissolves the "zero vs. absent" debate — features are present syntactically; their phonological expression is independent. Anderson's A-Morphous Morphology (1992) takes a different view, treating morphology as form-changing rules without morpheme entries. Both frameworks accommodate zero forms within their architecture.

Conversion (or zero-derivation) is the formation of a new word from another part of speech without overt affixation. English "to email" (verb from noun), "the run" (noun from verb), "to chair" (verb from noun, "chair the meeting"). One analysis treats this as derivation by zero affix — N + ∅ → V. Alternative analyses treat conversion as direct category change without affixation. Productivity is high in English compared to other languages. Mark Aronoff's Word Formation in Generative Grammar (1976) and Andrew Spencer's Morphological Theory (1991) discuss the analyses. Conversion poses challenges for any morpheme-based theory.

Some languages have markers that "look like zero" but contrast with other equally-zero forms. Latin's nominative singular -us (homo) contrasts with the genitive singular -is (hominis); both are overt morphemes, but the nominative ending is sometimes zero (consul, consul-is). The point: not every paradigm has zero in the same slot. English plural is zero on sheep; on cat it's -s. Russian nominative is zero on stol "table"; on a different noun it's -a. Cross-linguistic surveys (Greville Corbett, Gender 1991; Number 2000) document zero patterns across languages.

Subtractive morphology removes phonological material — French masculine "petit" [pəti] vs. feminine "petite" [pətit] is sometimes analyzed as subtractive (loss of the final t in masculine). Replacive morphology substitutes — Arabic "kitaab" (book) → "kataba" (he wrote) — same root, different vocalic pattern. Both pose challenges to the morpheme-as-string model and have been adduced as evidence against zero-morpheme analyses (Anderson 1992; Stump 2001). Word-and-paradigm models avoid both zero morphemes and morphemes as strings, treating words as the unit and morphology as paradigmatic alternation.