In Language X, an assimilation process copies both [labial] and [round] from one segment to an adjacent one, but never copies [voice] or [nasal] in the same process. In Language Y, a deletion process deletes both [voice] and [spread glottis] together. What does feature geometry predict from these data?
A[labial] and [round] should be terminal nodes in the same branch; [voice] and [spread glottis] should be terminal nodes in a different branch
BAll four features should be grouped under one Laryngeal node, since they all involve laryngeal activity
CFeature geometry cannot explain these patterns because the features involved cross natural class boundaries
DThe two languages have incompatible phonological systems and cannot be compared within a single framework
The core insight of feature geometry: features that spread or delete together share a dominating node. [labial] and [round] spreading together in Language X is evidence they are sisters under a shared node (e.g., a Place or sub-Place node). [voice] and [spread glottis] deleting together in Language Y is evidence they are sisters under the Laryngeal node. The fact that [voice] does not copy with [labial] in Language X confirms they belong to different branches.
Question 2 Multiple Choice
What is the primary evidence that motivates the construction of a feature-geometric tree for a particular language?
AThe universal hierarchy proposed in phonological theory — all languages share the same feature tree structure
BThe phoneme inventory of the language — which sounds exist determines which features must be included and how they are related
CSpreading and deletion patterns — which features consistently behave together as a unit in phonological processes
DAcoustic spectrogram analysis showing which features are produced by overlapping articulatory gestures
Feature geometry is built from phonological behavior, not imposed top-down from a universal template. If processes in a language consistently spread features X and Y together but never X and Z, the tree groups X and Y under a shared node. This makes the tree structure an empirical hypothesis — testable by additional processes discovered in the language — rather than a given.
Question 3 True / False
The fact that nasal assimilation in English (in- → im- before bilabials, iŋ- before velars) can be described as a single rule spreading the Place node — rather than three separate rules — is evidence for organizing place features under a dominating node.
TTrue
FFalse
Answer: True
Without hierarchical organization, three separate rules would be required: one for labial assimilation, one for coronal, one for dorsal. Feature geometry reduces these to one rule (spread the Place node of the following consonant) applied to one structural node — explanatory parsimony that follows directly from the hierarchical grouping of place features. The unification of these patterns is the empirical argument for the Place node.
Question 4 True / False
Feature geometry proposes a single universal tree structure that applies to most human languages, with nearly every language sharing the same hierarchy of nodes and terminal features.
TTrue
FFalse
Answer: False
This is the key misconception. Feature geometry is a framework, not a universal blueprint. The structure must be motivated language by language — the tree is a hypothesis about which features pattern together as units, based on the spreading and deletion processes actually observed in that language's phonology. Some structures (like a Place node) may recur widely, but their justification must come from language-specific phonological evidence.
Question 5 Short Answer
Why is evidence from phonological spreading processes more useful for motivating feature-geometric structure than evidence from the phoneme inventory alone?
Think about your answer, then reveal below.
Model answer: The phoneme inventory tells you which features a language uses, but not how those features are organized relative to each other. Spreading processes reveal behavioral dependencies — when feature A spreads, do features B and C always accompany it? If they do, that is evidence B and C are dominated by the same node as A. The inventory might include [labial], [voice], and [nasal], but only spreading data can reveal that [labial] and [round] pattern together while [voice] does not. Feature-geometric structure is a claim about phonological constituency, and constituency is only visible through behavior.
This is analogous to discovering syntactic constituency through movement and deletion tests rather than from the sequence of words alone. The geometric structure is an abstract representation motivated by phonological behavior — it cannot be read off the surface inventory, only inferred from how features co-vary across phonological processes.