Linguistics Flashcards
Morphemes vs. Syllables
- Syllables can be seen as the phonological building blocks of words
Speaker → “speak” + “er” (Morphemes) vs “speak” + “ker”(syllables
Free morphemes can be used as words, bound morphemes cannot
Words
- Every word is composed of one or more morphemes
- In some languages, characters represent complete words(or syllables)
Morphological normalization
- Identification of a single canonical representative for morphologically related wordforms
- Used in NLP to identify different forms of the same word
Normalization methods
- Stemming(Derivation): The text analysis that identifies the stem of a token(”derive” - > “deriv”, “am” →”am”)
- Lemmatization(Inflection): The text analysis that identifies the lemma of a token(”derive”→”derive”, “am”→”be”)
Syntax
- The syntax of a language is defined by a grammar
- the structural relationships between words, usually within a sentences
Two types of word classes
- Open(lexical words): Theoretically, infinitely many members per class(Nouns, Adjectives, Adverbs, Main Verbs(see, played) and digital numbers(3233)
- Closed(functional words): Number of members is fixed in principle(Modal verbs (can, had), determiners(the, some), conjunctions(add, or), pronouns(its, their), prepositions, particles(off, up) and interjections(Ow, Eh)
Structural Relationships
- Part of speech. The class of a word is decided by its syntactic context(is on the boundary between morphology and syntax)
Abstract classes: Noun, verb, adjective, adverb, preposition, pronoun, conjunction, interjection, determiner
POS tagging: the text analysis that assigns a part-of-speech tag to each token
Clauses
- Clauses: Grammatical units that express complete propositions
Clauses are the smallest grammatical unit that can express a complete preposition
Two basic types of clauses:
- Main Clauses: Independent, can stand alone as a sentence: I remained dry
- Subordiante clause: is reliant on a main clause and thus depends on it: (Usually stars with a subordinating conjunction) Although it rained, “because I was inside the building”
Grammars
a description of the valid structures of a language, is defined by a set of rules
Syntactic parsing
(aka full parsing): the text analysis that determines the grammatical structure of a sentence with respect to a given grammar(Types: constituency parsing and dependency parsing
Phrase vs Dependency grammar
Phrase grammar: Models the constituents of a sentence and how they are composed of other constituents and words(Constituency tree, inner nodes are non-terminal, leaf are terminal)
Dependency grammar: models the dependencies between the words in a sentence(Dependency tree, all nodes are terminal, the root is nearly always the main verb(of the first main clause)
Semantics
syntactic ambiguity: arises when a piece of text has more than one valid syntactic structure
### the meaning of single words and compositions of words
Meaning:
- Propositional content in terms of validity or truth conditions
- Often requires common-sense reasoning based on world knowledge(Max can open Tim’s safe, he knows the combination)
- Includes expressed emotional content(That poor cat, Fortunately, Max can ope
Linguistic Form vs context of use
- Meaning implied by linguistic form
- Context-independent meaning, such as “its raining”
- What a speaker publicly commits to, such as “It is wet outside”
- A speaker’s private state, such as “You should take an umbrella”
- Meaning dependent on context of use
- Scope of quantifiers, such as “Every student reads some book”
- Word sense ambiguities, such as “I’m making it”
- Semantic relations between noun in compounds, such as “play book”
- Meaning dependent on non-lunguistic perception
- Time, such as “now”, tomorrow,..
- Location, such as “here”, “there”, “That’s a beautiful city”
- Speaker and hearer such as “I”, “You”,..
Lexical Semantics
- Word senses: Distinctions in meaning between different uses of the same form
- Polysemy: related word senses with the same lexical entry(newspaper(physical object vs abstract content))
- Homonymy: Unrelated word senses that have the same lexical entry(Bank ( river bank vs money bank))
- Semantic roles: Number of arguments of a predicate, specific relationship the arguments bear to the predicate
- The roles the arguments of a predicate have in the state or activity captured by the predicate
- Different predicates have different semantic roles: She saw Max, She kissed Max, She ressembled Max
Compositional semantics
The meaning of word compositions in phrases, clauses and sentences
- Relations:
- Semantic: Relations between entities from the world
- Temporal: Relations between describing courses of events
- Linguistic Operators:
- Quantifiers: Indicating quiantities of objects, such as “Every man read”
- Hedges: Limiting the impact of propositions, such as “Probably every”
- Negation: inverting adjectives/predicates, such as “Probably not every”
- Semantic relations:
- Word compositions that capture relational predicates with arguments
- Typically: who did what to whom, where, when, how and why?
- Common relations types:
- Binary relations: Relations with two arguments
- Events: Relations with multiple arguments, possibly nested relations
- Connotation: What word choice conveys beyond truth-conditional semantics, such as “Insightful results”vs “interesting results”
Between Lexical and compositional Semantics
- Multi-word expression: A lexical larger than a word that can bear both compositional and idiomatic meanings(driving instructor, vice versa)
- On the boundary between lexical an compositional semantics(Long time no see, Kick the bucket)
- Word n-grams: An alternative to identifying multi-word expression to simply use word bigrams, trigrams, or similar instead
Entity: An entity represents an object from the real world
- Named entities: objects that can be denoted with a proper name: Names like In Hannover or at Leibniz University Hannover
- Numeric entities: Values, quantities, proportions, ranges or similar. “in this year”, “2023-03-04”, $10 000, etc.
Pragmatics
Meaning in context of discourses
Discourse
Describes linguistic units that are larger than a sentences
Discourse → Monologues
Dialogue → Conversational discourse with two or more parties
Discourses-level semantics:
- Coreference: Different expressions may be used to refer to one thing
- Coherence: Understandable discourse has continuity in meaning
Discourse-Level Units
- Paragraph: Grammatically, a paragraph is a sequence of one or more sentences, whose boundaries are denoted by line breaks.
- Discourse structure: the structure that represents the organization of an entire text
- Discourse segment:
- A linguistic unit serving as a single building block of a discourse
- May consist of multiple smaller adjacent segments
- Elementary discourse unit. Minimum segment, usually a clause(Tempting as it my be)
- Coherence relations:
- Describes how two segments relate to each other
- May be semantic or pragmatic
- Discourse parsing:
- The text analysis that infers the discourse structure of a text
- Implicit segments and relations are what makes parsing hard
- Discourse segment:
- Coreference: Two or more expression in a text that refer to the same thing(Max waljed in, he sat down)
- Coreference resolution: the text analysis that maps all references to unambiguous identifiers and coreference resolution may require deep text understanding
- Local vs global coherence:
- Local: Coherence in adjacent discourse segments
- Global: Coherence of the entire discourse of given tex
Pragmatics:
- Pragmatics deals with how language is used to accomplish goals
- Relates to the author’s intention and to the context of use(I never said she stole my money)
- covers speech acts, presupposition and implicature, and much more
Speech acts
A linguistic utterance with a performative function
the terms is mostly used to refer to illocutionary speech acts
Types:
- Locutionary act: the act of saying something meaningful(Smoking is bad for your health)
- Illocutionary act: A direct or indirect act performed by performing a locutionary act(Assertion that smoking is bad for your health(direct), Warning not to smoke(indirect))
- Perlocutionary act: an act which changes the cognitive state of the interlocutor(Causing you to adopt the intention to stop smoking)
Implicature
Presupposition: Implicit assumption about the world related to an utterance whose truth is taken for granted(Max cousin took an aspirin → Max has a cousin, someone’s called max)
Implicature: What is suggested by a linguistic utterance, even though neither expressed nor entailed
