Decomposing vowel melodies in root-and-pattern morphology
Abstract
Root-and-pattern (R&P) morphology has long been analysed as evidence for non-concatenative word formation, where consonantal roots interdigitate with vocalic templates. This paper develops a syntactic alternative inspired by Kalin (Princeton University, 2026), proposing that vowel melodies are not unified morphological objects but decomposable exponents of distinct syntactic heads. Under this hypothesis, apparent discontinuity in vowel patterns reduces to linear spell-out of hierarchical structure rather than templatic insertion. Evidence from Modern Hebrew and Neo-Aramaic supports a systematic mapping between individual vowels and functional heads such as Voice, Tense, and v. Comparative data from Mubi (Chadic) and Yucatec (Mayan) delimit the empirical scope, showing partial convergence and non-decomposable systems. The analysis eliminates melodic morphemes, reduces representational redundancy, and reframes R&P morphology as a surface reflex of syntactic structure rather than a distinct morphological module.
1. Introduction
A central problem in morphological theory is how grammatical structure is realised when linear concatenation fails. Root-and-pattern systems in Semitic languages provide a canonical challenge. In Modern Hebrew:
Here, consonantal roots and vocalic patterns are non-linear. While root discontinuity has received extensive treatment, the status of vowel melodies remains theoretically unsettled.
Traditional analyses treat vowel patterns as unified templates interleaved with consonantal roots. This approach captures surface regularities but introduces a representational tension: a single morpheme must be discontinuous, invariant across environments, and syntactically sensitive.
This paper addresses this tension by focusing on a stronger claim: vowel melodies do not exist as morphological units.
2. Background: templatic morphology
Autosegmental approaches (McCarthy 1981; Bat-El 1994; Arad 2005; Borer 2013) analyse R&P morphology as the interaction of consonantal roots and vocalic templates across separate tiers. A template (e.g., CaCaC, CiCeC) is associated with a root via prosodic mapping.
Although empirically successful, this model faces three limitations. First, vowel patterns vary systematically with morphosyntactic context, suggesting they are not purely phonological. Second, the notion of a single melodic template requires abstract representational machinery not independently motivated. Third, discontinuity is stipulated rather than derived.
These issues motivate syntactic approaches in which morphological structure is fully derived from syntactic composition and Vocabulary Insertion.
3. Hypothesis: vowels as syntactic exponents
Following Kalin (2026), this paper adopts a restrictive hypothesis:
Each vowel in a root-and-pattern system corresponds to a distinct syntactic head.
Vowel sequences such as i–e are therefore not unified morphemes but the linear realisation of independent functional projections.
Illustratively:
- i → Tense
- e → Voice
- a → v-type specification (context-dependent)
The consonantal root provides lexical content, while vowels realise distributed syntactic features. Under this view, discontinuous melodies are epiphenomena of linearisation.
4. Hebrew evidence
Modern Hebrew verbal morphology provides a primary testing ground. Across binyanim, systematic variation in vowel quality correlates with morphosyntactic distinctions:
A decompositional analysis yields the following generalisations:
V1 correlates with Tense-related features
V2 correlates with Voice
v-type distinctions condition stem structure
Crucially, vowels behave as independent variables across paradigms rather than as a unified template. This supports a model in which each vowel is inserted locally at a distinct syntactic position and linearised post-syntactically.
5. Eliminating melodic discontinuity
The key theoretical consequence is the elimination of melodic discontinuity.
Under classical accounts, vowel patterns must span non-adjacent root positions, requiring templatic insertion. Under the present analysis, this discontinuity disappears.
Three consequences follow:
The derivation reduces to ordered spell-out:
√ROOT → v → Voice → T → PF
Thus, what appears as a vowel melody is the surface reflection of syntactic ordering.
6. Cross-linguistic comparison
6.1 Neo-Aramaic (Amadiya)
Neo-Aramaic shows strong convergence with Hebrew. Distinct vowel positions align systematically with Voice and Aspectual distinctions. Importantly, V1 and V2 exhibit independent distributional conditioning, supporting decomposition.
This yields a near-isomorphic mapping between syntactic heads and vowel exponence, reinforcing the hypothesis.
6.2 Mubi (Chadic)
Mubi presents partial support. Vowel alternations correlate with aspectual and derivational distinctions; however, both vowels tend to pattern jointly rather than independently. This reduces the empirical motivation for decomposing vowels into separate heads, although decomposition remains formally possible.
Thus, Mubi suggests a gradient rather than categorical implementation of vowel decomposition.
6.3 Yucatec (Mayan)
Yucatec diverges significantly. Vowel alternations are primarily prosodic and tonal, with no evidence for multiple independent vowel exponents. Alternations are more economically analysed via autosegmental tone and length rather than syntactic decomposition.
This indicates that vowel-melody decomposition is not universal but conditioned by language-specific morphophonological architecture.
7. Theoretical implications
The decomposition of vowel melodies has three implications for morphological theory.
First, it eliminates templatic primitives from grammatical architecture. Apparent templates reduce to distributed exponence.
Second, it strengthens a syntax-first view of morphology in which morphological structure is entirely derived from syntactic configuration.
Third, it reframes discontinuity as a property of linearisation rather than representation.
Formally:
Morphology = Syntax + Vocabulary Insertion + Phonological Linearisation
No additional templatic component is required.
8. Urdu and Saraiki: a contrastive perspective
Although Urdu and Saraiki do not exhibit root-and-pattern morphology, they provide a useful contrast class.
Urdu shows:
Saraiki shows parallel structures with auxiliary stacking and causative layering.
These systems exhibit:
- linear suffixation
- transparent syntactic stacking
- external realisation of functional structure
Unlike Semitic systems, functional heads are not compressed into vowel space but expressed linearly. This contrast supports a broader typology:
Both reflect a single underlying computational system with different phonological realisation strategies.
9. Conclusion
This paper has argued that vowel melodies in root-and-pattern morphology are not unified morphological objects but distributed syntactic exponents. Evidence from Hebrew and Neo-Aramaic supports systematic decomposition, while Mubi and Yucatec delimit its empirical applicability.
The central result is a simplification of morphological architecture: discontinuous vowel patterns are not primitives of grammar but emergent effects of syntactic spell-out.
Root-and-pattern morphology, rather than constituting an exceptional morphological system, thus provides a transparent window into how hierarchical syntactic structure is mapped onto phonological form.
