Introduction to Semantics & Pragmatics | Linguistics

Speech act theory, developed by philosophers John Austin and John Searle, analyzes how utterances function as actions rather than mere statements. A speech act involves three simultaneous levels: the locutionary act (producing a grammatical utterance), the illocutionary act (the actual action performed, such as commanding or apologizing), and the perlocutionary act (the effect produced on the listener). Searle categorized speech acts into five main types: representatives (stating beliefs, e.g., 'The earth is flat'), directives (aiming to make others act, e.g., 'Please don't smoke'), commissives (committing to future action, e.g., 'I promise'), expressives (expressing psychological states, e.g., 'I'm sorry'), and declarations (changing reality through utterance, e.g., 'I now pronounce you married'). Performative utterances, which perform actions rather than just describe them, can be explicit (containing performative verbs like 'I promise') or implicit (without such verbs like 'I'll be back').

Speech act theory, developed by John Austin and expanded by John Searle, explains that utterances perform actions in the real world beyond merely transmitting information. Austin classified speech acts into three categories: locutionary (the words themselves pronounced), illocutionary (the communicative intention being performed, such as requesting or promising), and perlocutionary (the effect produced on the receiver). Searle later expanded this into five categories: assertive (stating beliefs or facts), directive (influencing behavior), commissive (committing to future actions), expressive (conveying emotions), and declarative (changing the actual state of affairs).

J.L. Austin revolutionized philosophy by treating language as human action rather than mere representation. He distinguished performatives (saying something is doing something: promising, apologizing) from constatives (describing states of affairs). He discovered saying itself is a doing, leading to the general theory of speech acts where all utterances are speech acts. The basic unit of meaning is the speech act—understanding requires knowing the speaker's intentions. Austin distinguished illocutionary acts (statements, questions, commands) from perlocutionary acts (effects on listeners). John Searle continued this work, investigating how language relates to reality through the lens of human action and intentionality.

Speech Act Theory, developed by John Austin and later expanded by John Searle, proposes that when we speak, we don't just describe reality but actually perform actions; for instance, saying 'I promise to help you' doesn't merely state a fact but performs the act of making a promise, demonstrating how language functions pragmatically to accomplish social actions beyond mere information transfer.

Speech act theory, developed by J.L. Austin and John Searle, distinguishes between representational language (statements that can be true or false about the world) and performative language (actions performed through speech, such as promising, ordering, or questioning). A speech act is defined as 'an attempt to do something simply by speaking.' Austin identified performative verbs like 'promise,' 'declare,' and 'swear' that are used to perform specific actions. The theory analyzes speech acts at three levels: locutionary (the linguistic content), illocutionary (the intended action), and perlocutionary (the effects on the listener). Conditions of felicity are necessary conditions for a speech act to succeed, including the speaker's authority, the context, and the relationship between speaker and listener.

This section covers truth-conditional semantics: linguistic meaning is associated between expressions and objects in a model or world. A model is a mathematical representation of a world with certain things happening in it. The denotation of an expression is the object it is associated with in the model. Different types of expressions map to different types of objects: proper names are elements from set theory, verbs are relations (pairs of things), and sentences are truth values from propositional logic. The formal definition states that a sentence s means P if and only if s is true in some situation V if P happens.

Truth-conditional semantics analyzes sentence meaning through syntactic structure rather than word strings. Meaning is composed algorithmically using rules, enabling systematic derivation of literal meaning. Structural ambiguity demonstrates that different syntactic parses yield different meanings, showing semantics operates on syntactic objects. This approach applies universally across languages despite surface differences. Linguistic meaning connects expressions to world objects through denotation: proper names denote individuals, verbs denote relations or sets, and sentences denote truth values. A sentence means something if it is true in some situation under certain conditions, establishing a formal framework for studying language semantics.

In truth-conditional semantics using model theory, linguistic meaning is derived compositionally from syntactic constituents: proper nouns map to individual objects in a model, intransitive verbs map to sets of individuals, transitive verbs map to sets of ordered pairs (subject-object), and sentences map to truth values (true/false); meaning composition follows specific rules where single-branch nodes inherit their daughter's meaning, verb phrases combine verbs with nouns through substitution, and sentences evaluate truth by checking if the subject belongs to the verb's relation set, with negation flipping truth values and conjunction applying standard truth tables.

Truth-conditional semantics is a linguistic framework that analyzes meaning by examining the conditions under which sentences are true or false, rooted in the correspondence theory of truth which holds that a statement's truth depends on its match with reality; the framework addresses the challenge of self-referential definitions in natural languages by translating them into predicate calculus—a formal metalanguage using individual constants (lowercase letters for specific entities) and predicates (uppercase letters for properties, relations, and processes)—through a four-step process involving symbol assignment, world model establishment, formula-model matching, and truth value calculation.

In 'Truth and Meaning' (1967), Davidson proposes a truth-conditional approach to meaning, identifying meaning with truth conditions. His theory rests on three requirements: semantic holism (meaning depends on the entire language), compositionality (finite resources generate infinite sentences), and extensionality (meaning is relations between sentences, not metaphysical entities). Davidson's model uses T-sentences: 'S means that P' equivalent to 'S is true if and only if P.' This approach rejects Quine's 'museum myth' of meaning as discrete objects. Davidson argues sentences are primary linguistic elements, and meaning arises only in assertions, which connect to the world through truth conditions.

Pragmatics studies implicit meaning—understanding beyond literal words. Context determines meaning: linguistic context (surrounding words), physical context (environment), and situational context. Deixis (ديكسس) refers to words that point to people, places, or things in context, such as 'here,' 'there,' 'this,' and 'that.' These words require physical or situational context to understand their meaning.

Deixis (embrayeurs) refers to linguistic elements whose meaning depends entirely on the communicative context, including who is speaking, where, and when; these elements require reference to the speech situation to be understood, as they function as pointers that indicate specific elements in the external world based on the speaker's perspective and the communicative context.

Deixis (from Greek for 'pointing with words') is a speaker-centered linguistic phenomenon that enables speakers to reference spatial, temporal, and discourse elements relative to themselves. It encompasses four main types: person deixis (using pronouns like I, you, he, she to encode speaker, addressee, and third-party references), place deixis (using demonstratives like this, that, here, there and locative adverbs to denote spatial locations), time deixis (using temporal markers like now, yesterday, tomorrow to locate events relative to the moment of utterance), and discourse deixis (using expressions like this, that, next, last to orient within an unfolding conversation or text). The speaker serves as the central reference point, meaning deixic expressions shift their meaning depending on who is speaking and when.

Deixis (or indexicals) are linguistic expressions whose meaning depends on context, specifically the speaker, addressee, time, and place of utterance; they are categorized into three types: person deixis (pronouns like I, you, he, she, we, they), time deixis (today, yesterday, tomorrow, last month), and place deixis (here, there, come in, go out), and require contextual information to be fully understood as they point to specific entities, events, or locations in the communicative situation.

This section establishes the theoretical foundation for analyzing second person pronouns in Spanish. Deixis refers to expressions anchoring reference to speech act participants, while indexicality operates at the pragmatic level. The second person pronoun 'tú' has two distinct interpretations: narrow (deictic) interpretation, which is referentially strong and specifically anchored to the addressee, and broad (non-deictic) interpretation, which is referentially weak and not anchored to any entity. The broad interpretation occurs in non-veridical contexts where the truth of the proposition is not presupposed, including questions, modal adverbs, imperatives, conditional clauses, and desiderative structures.

Semantics refers to the literal meaning of words and sentences according to language rules, while pragmatics refers to the contextual and intentional meaning that depends on the situation, speaker, and listener; effective NLP requires understanding both aspects through techniques like First-Order Logic for representation, syntax-driven semantic analysis, word sense disambiguation using synonyms and bootstrapping methods, and thematic role assignment to enable machines to understand human language accurately.

NLP applications range from small to large: (1) Small: spelling correction, hyphenation, Turkish diacritization; (2) Medium: word sense disambiguation (determining which meaning of a word is intended in context), named entity recognition (identifying persons, places, times, organizations); (3) Large: question answering systems, conversational agents, machine translation. Word sense disambiguation is crucial for machine translation - for example, 'book' in 'I read the book' refers to a physical book, not an ebook. NLP requires knowledge of: phonology (sounds), morphology (word structure), syntax (sentence structure), semantics (word meanings), and pragmatics (manners of communication). This course focuses on morphology, syntax, semantics, and pragmatics.

This section covers Natural Language Processing (NLP) and Machine Translation. NLP enables computers to understand, interpret, and interact with human language through five principles: Syntax (sentence structure), Semantics (word meanings), Pragmatics (context and intent), Morphology (word structure), and Phonology (sound patterns). Applications include translation, chatbots, speech recognition, and search engines. Machine Translation systems automatically convert text between languages using three approaches: Direct Translation, Interlingual Translation (via intermediate language), and Transfer Approach (via abstract representation). The complete workflow includes analysis, representation, and generation stages.

Natural Language Processing (NLP) is the field of study dedicated to the automatic understanding and generation of human natural languages, requiring systems to comprehend multiple linguistic dimensions including phonetics/phonology (sound patterns), morphology (word structure), syntax (word order), semantics (meaning), and pragmatics (contextual meaning), enabling applications like voice assistants, machine translation, and spell checkers.

Logical semantics drives transformative applications across natural language processing and artificial intelligence. In NLP, it powers machine translation by mapping quantifiers and logical structures between languages, question answering by extracting key predicates and entities from knowledge bases, and information extraction by transforming unstructured text into structured data (identifying persons, awards, years, and reasons). In AI and cognitive science, it enables voice assistants to map intentions to actions, chatbots to deduce answers from structured knowledge, and robots to interpret instructions by disambiguating objects and locations. Medical AI systems leverage logical semantics for creating ontologies formalizing relationships (doctor treats patient) and performing automated reasoning about diagnoses based on symptoms. Autonomous vehicles apply these principles to interpret contextual instructions accurately.