Questions: Distributed Morphology

Question 1

In a Distributed Morphology analysis, why does English use 'went' rather than '*goed' as the past tense of 'go'?

• A. Because 'go' and 'went' are stored as a linked pair in the lexicon, with a stipulated exception to regular past tense formation
• B. Because the Vocabulary Item [PAST, √GO] → 'went' outcompetes [PAST] → '-ed' due to the specificity principle: a more specific VI takes priority
• C. Because DM treats suppletion as phonological reduction that obscures the regular past tense morpheme
• D. Because syntax cannot attach the '-ed' morpheme to monosyllabic verb roots

Answer: (B)

DM explains suppletion through Vocabulary Insertion: competing Vocabulary Items apply to feature bundles, and the most specific match wins. The VI [PAST, √GO] → 'went' matches both features, beating the default [PAST] → '-ed', which matches only one. This is not a lexical quirk or stored exception — it is a principled competition predicted by the same mechanism that handles all morphology. Option A is the Lexicalist account that DM explicitly rejects.

Question 2

What does 'Late Insertion' mean in Distributed Morphology, and what problem does it solve?

• A. Morphological affixes are inserted late in the surface phonological string, after all syntactic movement
• B. Phonological form is assigned to abstract morphosyntactic features only after syntactic computation is complete, not pre-stored in a traditional lexicon
• C. Vocabulary Items are inserted after semantic interpretation, ensuring meaning is not affected by phonological form
• D. Inflectional morphology is acquired later in language development than derivational morphology

Answer: (B)

Late Insertion is the architectural claim that syntax operates on abstract, phonologically unspecified features — phonological material is only assigned at the morphophonological interface, after syntactic operations are done. This solves the problem of explaining how the same abstract feature (e.g., [PAST]) can receive different phonological exponents in different contexts (regular '-ed' vs. suppletive 'went') without pre-listing these as separate lexical entries.

Question 3

In Distributed Morphology, words are assembled in the mental lexicon before being inserted into syntactic structures.

• T. True
• F. False

Answer: False

This is the Lexicalist view, which DM explicitly rejects. DM claims there is no pre-syntactic lexicon in the traditional sense. Instead, syntax builds structures from abstract morphosyntactic features, and phonological form is inserted only afterward (Late Insertion). The three-list architecture (syntactic features, Vocabulary, Encyclopedia) replaces the lexicon, with no pre-built words entering syntax.

Question 4

Distributed Morphology predicts that syncretism — when one phonological form expresses multiple distinct feature bundles — results from under-specification in Vocabulary Items.

• T. True
• F. False

Answer: True

Syncretism is a natural prediction of DM's architecture. A Vocabulary Item that is under-specified (matching multiple feature contexts) will insert in all those contexts, producing the same phonological form for what are structurally distinct feature bundles. For example, a VI matching [PAST] without specifying person or number would appear wherever those features are absent. This is a unified account of a phenomenon that Lexicalism treats as coincidental formal identity.

Question 5

How does DM's three-list architecture differ from the Lexicalist model, and what specific architectural feature does 'Late Insertion' contribute?

Model answer: Lexicalism stores fully formed words (with their phonological form and meaning) in a pre-syntactic lexicon that feeds syntax. DM replaces this with three separate lists: (1) abstract morphosyntactic features manipulated by syntax, (2) the Vocabulary (mappings from feature bundles to phonological forms), and (3) the Encyclopedia (idiosyncratic, non-compositional meanings). Late Insertion means phonological form is assigned only after syntactic computation — syntax never sees phonological content, only abstract features. This allows the same derivation to produce different phonological outputs depending on context (enabling suppletion, syncretism) without pre-specifying exceptions in the lexicon.

The key shift is from a word-based to a feature-based architecture. By separating phonological realization from syntactic computation, DM can account for morphological idiosyncrasy through principled competition mechanisms rather than stipulated exceptions, and makes the unified prediction that morphological and syntactic complexity reflect the same underlying hierarchical feature structures.