Ambiguities and Their Types
On this page
Ambiguities and Their Types
Recall first
Name the ambiguity in each: I saw her duck; old men and women; The chicken is ready to eat; He deposited money at the bank. Can one sentence contain more than one type?
First principles
Ambiguity occurs when one form supports multiple analyses or interpretations. Vagueness is different: a form has a broad or underspecified range rather than discrete competing analyses. NLP systems must preserve candidates or use evidence to rank them.
- Lexical ambiguity: one word has multiple senses or categories:
bank(finance/river),duck(bird/verb). - Morphological ambiguity: a surface form has multiple word analyses/features.
unlockablecan mean “not able to be locked” or “able to be unlocked”; an inflected form may have multiple lemma/POS analyses. - Syntactic/structural ambiguity: a sentence has multiple parse trees.
old men and womenmay coordinate[old men] and [women]or[old] [men and women]. - Semantic ambiguity: a syntactically plausible structure has different compositional meanings, including scope: “Every student read a book” allows different quantifier scopes.
- Referential/anaphoric ambiguity:
Ravi told Amit that he wonleavesheunresolved. - Pragmatic ambiguity: interpretation depends on speaker intention and situation: “Can you open the window?” is often a request, not a capability question.
- Phonological/orthographic ambiguity: homophones or missing spacing can create alternatives; these matter especially in speech and languages with different writing conventions.
Worked example. I saw her duck has at least a lexical/POS ambiguity (duck noun vs verb), and if duck is a verb, her can be an object or possessive determiner in a different reading. A tagger proposes categories, a parser proposes structures, and semantic/pragmatic context ranks the readings. Do not call every later-stage failure “semantic ambiguity”; identify where the alternatives arise.
English relies heavily on word order and function words. Indian regional languages vary in script, tokenization, morphology, case marking, compounding, free-word-order tendencies, code-mixing, and resource availability. These are linguistic and data conditions, not a license to claim that every language has the same ambiguity pattern.
Exercise — reveal after answering
Identify the main ambiguity in The students discussed the problem with the teacher.
Answer: Structural PP attachment: with the teacher may be the discussion partner or a property of the problem (the latter is less plausible but syntactically possible). Context resolves it.
Exam lens
- Define ambiguity before listing types.
- Give a sentence and explain competing analyses, not just labels.
- Distinguish lexical ambiguity (word-level alternative), syntactic ambiguity (tree alternative), and pragmatic ambiguity (context/intention).
- Mention regional-language variation carefully: script, segmentation, morphology, and resources differ.
Rapid revision checklist
- Define ambiguity vs vagueness.
- Recall lexical, morphological, syntactic, semantic, referential, pragmatic types.
- Trace one example through tagger → parser → context.
- Avoid treating English assumptions as universal.
Key takeaways
- Ambiguity is layered; one sentence can trigger several layers.
- Candidate generation and contextual ranking are complementary.
- Language-specific writing and morphology alter where ambiguity appears.