Given vs New Information

How information status works

A speaker is constantly modelling the hearer. Some things — the time of day, last week's weather, the speaker's own name — the hearer already has in mind. Other things — what happened on the speaker's flight this morning, the price of a new gadget — they don't. Every sentence allocates its words across these two zones: given for what the hearer can already access, new for what the speaker is adding.

The split predates information theory. Halliday (1967) first formalized it for English; Wallace Chafe (1976) embedded it in cognitive psychology with the notion of activation; Ellen Prince (1981) widened the taxonomy to include intermediate cases; Knud Lambrecht (1994) integrated it with topic and focus. The terminology varies — given/new, theme/rheme, hearer-old/hearer-new, accessible/inaccessible — but the underlying intuition is shared.

Three properties cluster on the given side: brevity, anaphoric form, and prosodic deaccenting. Given entities are referred to with pronouns or zero anaphora rather than full noun phrases; they sit in syntactic positions that signal continuity (subject, sentence-initial); and they carry no pitch accent. New entities take the opposite profile: full noun phrases, late position, accented.

The crucial constraint is the given-new contract (Haviland and Clark 1974): speakers package their utterances so the given portions match something the hearer already has. When the contract holds, comprehension is fast. When it breaks — when given is presented as new, or new as given — the hearer must repair it through accommodation, often slowing reading times measurably.

Why this matters

• Discourse coherence. The given-new alternation is the engine of paragraph flow.
• Definiteness across languages. Articles, demonstratives and bare nouns realize information status.
• Reading and comprehension. Information packaging predicts reading time and recall accuracy.
• Prosody and TTS. Synthesizers that ignore givenness produce flat or wrongly-accented speech.
• Translation. Article-less to article languages requires repackaging definiteness; the reverse strips redundancy.
• Headlines and journalism. Lede paragraphs introduce new entities; later paragraphs maintain them as given.
• Second language acquisition. Article use is one of the slowest features for L2 English to master.

Given vs new

	Given	New
Cognitive status	Active or accessible	Inactive, freshly introduced
Form (English)	Pronoun, definite NP, zero anaphora	Indefinite NP or full proper name
Position (English default)	Clause-initial, subject	Clause-final, postverbal
Prosody	Deaccented	Carries the nuclear pitch accent
Article	the, demonstratives	a, an, some, existential there
Length	Short, often pronominalized	Often longer, descriptively rich
Recall in memory tests	High — hearer integrates with prior representation	Lower until repeated
Russian / Mandarin marker	Sentence-initial bare noun	Sentence-final or indefinite NP

Cross-linguistic markers

English uses articles. The signals identifiability — the hearer can pick out the referent uniquely; a/an signals an indefinite, often discourse-new entity. The contrast does extra duty for given/new: I saw a dog. The dog was barking. The first mention is new (indefinite); the second is given (definite).

Russian has no articles. Information status surfaces in word order: a sentence-initial bare noun defaults to given/definite, a sentence-final bare noun to new/indefinite.

Книга на столе.   "The book is on the table."  (book = given)
На столе книга.   "There is a book on the table." (book = new)

The same words in opposite orders shift the entire information-structural reading.

Mandarin Chinese uses related strategies. The numeral-classifier yī ge "one CL" introduces indefinite/new referents; bare nouns and demonstratives mark given/definite. Word order also matters: preverbal subjects are typically definite, post-verbal objects indefinite.

一个 男孩 来 了。     A boy came. (NEW)
那 个 男孩 来 了。    That boy came. (GIVEN)

Japanese overlays the topic particle wa (typically given) on the bare noun system. The particle ga marks subjects of new information (the answer to "who came?"). Demonstratives kono / sono / ano mark definiteness.

Czech and other Slavic languages push the new information to the end of the clause obligatorily — the functional sentence perspective tradition, developed by the Prague School, made this systematic.

Tagalog and other Philippine languages use ang-marking to flag the most prominent referent of the clause, often the topic; non-ang noun phrases tend to be new or peripheral.

Worked examples

Default given-new flow. Yesterday I bought a book. The book was about pragmatics. First clause: I = given (speaker), a book = new. Second clause: the book = given (just introduced), about pragmatics = new.

Existential introduction. There's a strange noise coming from the kitchen. The existential there-clause exists primarily to introduce a new entity at the end of the sentence; a strange noise is hearer-new.

Bridging. I went to a wedding. The bride was tall. The bride has not been mentioned, but the wedding licenses her as inferable. The definite article is felicitous despite the noun being technically new.

Deaccenting given material. I saw a DOG. Then the dog ran AWAY. The second occurrence of dog is given and deaccented; the predicate away takes the focal accent.

Russian word order. Маша поцеловала Петю "Masha kissed Petya" (Masha given, Petya new) vs Петю поцеловала Маша "It was Masha who kissed Petya" (Petya given, Masha new). Free word order does focus-marking work.

Pronoun substitution. Marie Curie won two Nobels. She was the first woman to win one. Once the entity is given, English replaces the full name with a pronoun. Repeating Marie Curie would feel stilted.

Accommodation. Walking into a room: Sorry I'm late, the bus broke down. The hearer has no prior bus in mind, but accommodates a definite reference because of plausibility — the speaker presupposes a bus and the hearer absorbs it. Stalnaker (1974) called this presupposition accommodation.

Related variants

• Hearer-old vs discourse-old. Prince (1992): an entity may be in the hearer's general knowledge (hearer-old) without having been mentioned in the current discourse (discourse-old). Famous people are hearer-old by default.
• Activation states. Chafe (1987) distinguished active (currently in focal consciousness), semi-active (peripheral), and inactive. Definiteness is sensitive to activation, not just identifiability.
• Accessibility hierarchy. Ariel (1990) ranked referring expressions by how accessible the referent must be: zero anaphora and clitic pronouns at the high-accessibility end, full names and demonstratives at the low end.
• Bridging references. Clark's "associated entities" — readers compute the link between mentioned and unmentioned but related entities.
• Specific vs nonspecific indefinite. I'm looking for a doctor (any doctor) vs I'm looking for a doctor (a specific one). Nonspecific indefinites resist later anaphoric reference.
• Generic noun phrases. The lion is a noble beast uses the not for givenness but for genericity — referring to the kind, not an instance.

Common pitfalls

• Confusing given with definite. Definiteness is about uniqueness/identifiability; givenness is about prior activation. The Pope is identifiable to almost any hearer but is not necessarily given in the current discourse.
• Treating given/new as binary. Most accounts use a hierarchy: active > accessible > inactive. Mid-discourse referents are neither fully given nor fully new.
• Ignoring deaccenting. Skilled English speakers deaccent given material. TTS systems and L2 learners often miss this, producing stilted prosody.
• Over-introducing. Repeating full names where pronouns suffice violates the given-new flow and reads as condescending.
• Under-introducing. Using a pronoun for a referent the hearer cannot recover violates the contract — they will pause to search for an antecedent that does not exist.
• Forgetting accommodation. Hearers will quietly absorb new presuppositions packaged as given when they are plausible — a tool used by skilled writers, weaponized by manipulative ones.