Language - Week 8

What is language?

  • A system of communication using sounds or symbols that enables us to express our feelings, thoughts, ideas and experiences.

  • Complex cognitive function that allows humans to communicate to record information, to express emotions

  • System of symbols and rules

  • Language influences thinking, perception and memory

Design features of language (Hockett, 1963)

  • 13 features - Non-human primates have first 9 but only humans have:

  • Displacement - Refer to things removed in time & space.

  • Productivity - Creates never-before-heard utterances that others understand w/ novel meanings.

  • Cultural transmission - Not inborn (feral children)

  • Duality of pattering - Recombine units into near infinite number of sentences.

Psycholinguistics

  • The psychological study of language.

  • Four main areas:

    • Comprehension - How do people understand spoken and written language?

    • Representation - How is language represented in the mind and in the brain?

    • Speech production - How do people produce language?

    • Acquisition - How do people learn language?

Components of language

  • Phonology (sounds)

  • Morphology (words)

  • Syntax (grammar)

  • Semantics (meaning)

  • Pragmatics (context)

Components of language

  • “The cat is hunting birds”

    • Phonology: the sound of words, while you read you also know how to pronounce words - can even pronounce them in your head (c-a-t)

    • Orthography: spelling of words such as cat and bird (kat? burd?)

    • Morphology: small units of language that have meaning or grammatical function (hunt, -ing, bird, -s)

    • Semantics: meaning, cat is going to chase a bird, cat is not playing with the bird, cat is going to eat very soon…(depends on the cat)

Reading versus Speech Perception

  • Although they both revolve around the words, reading and speech perception differ in a number of ways

    • Reading (easier)

      • Word seen as whole

      • Less ambiguous

      • Less demanding

        • Words still available

    • Speech Perception

      • Word spread out in time, transitory

        • Difficult to tell when one word ends and another begins

      • Ambiguous signal

        • Sub-optimal conditions

        • E.g, Background noise - we have to disintangle between speech and background noise

      • More demanding

        • Words no longer available (transitory)

    • Speech Perception (easier)

      • Prosodic cues (pitch, intonation, stress and timing) hint at sentence structure.

      • Gestures (hand and faces) often accompany speech.

Speech segmentation

  • Our ability to perceive individual words even though there are often no pauses between words in the sound signal.

  • Supported by:

    • Knowledge of the language - known words in a foreign language “pop-out”.

    • Meaning of the sentence - “be a big girl and eat your vegetables”, The thing Big Earl loved most in the world”.

    • Patterns of sounds - some sounds usually go together in words “prettybaby” “pre-tty” (correct) “tty-ba” (incorrect).

    • Speech comprehension errors can occur when the bottom-up signal is degraded (e.g. background noise) “see it, say it, sort it”

Speech: Perceiving individual words in sentences

  • Speech perception is difficult due to ambiguous signal (accents, pronunciations “d’ya wanna” “do you want to”.)

  • Pollack & Pickett (1964):

    • Recorded conversations of participants in waiting room.

    • Participants then presented with single words from conversation.

    • Could identify only half the words, even though they were listening to their own voices1

      → Ability to perceive words is aided by the context of the words and sentences that make up the conversation.

Speech: The phonemic restoration effect (Warren, 1970)

  • Participants asked to indicate where in the sentence the cough occurred.

  • Could not identify the correct position of the cough.

  • None of them noticed that /s/ in “legislatures” was missing.

    → Meaning & context influences our perception speech sounds.

    → “Filling in” the missing phoneme based on word & sentence context = top-down processing.

Reading: The word superiority effect (Reicher, 1969)

  • Target letter more easily identified if present in word than non-word.

  • Letters are easier to recognise when they are contained in a word than when they appear alone or are contained in a non-word.

  • Processing of letters in words is faster and more accurate.

  • Letters in words are not processed one by one, each letter is affected by the context within which it appears.

  • Just as context affects how we hear phonemes in spoken language, context affects how we see letters in written language

Word Recognition: Interactive Activation Model (McClelland & Rumelhart, 1981)

  • Visual word processing taking into account bottom-up and top-down proccessing.

  • Recognition units at three levels

    • Word

    • Letter

    • Feature

  • Extract vertical line I at Feature Level

    • Activation / excitation to all letters with I present (e.g., H, N, D, M, L, F, R, P, J, B, E)

    • Inhibition to thpse without(e.g., A, Q, W etc…)

  • Letters identified at Letter Level (H)

    • Activation/excitation to all ‘4 letter word’ units with letters in that position (e.g., HOOD, HEAD, HELF, etc…)

    • Inhibition for 4 letter word units without (e.g., MOOD)

  • Words recongised at word level

    • Activated words increase level of excitation letter-level units for letters in that word.

  • Letter Level

    • Activated letters increase level of excitation of feature-level units for features in that letter

  • Accounts for the Word Superiority Effecr through top-down processes.

  • Activation at the word level increases activation at the letter level (A)

    • Decreases activation at letter level (N)

  • Top-down processing occurs about 200ms after word onset.

  • Model does not take into account

    • meaning in role of processing words

    • phonological processing of words

    • ease with which can read Cambridge email

    • words> 4 letters

Word Recognition: Reading Psuedowords (Coltheart el al., 2001)

  • Dual-route Cascaded Model

    • Three routes for printed words and speech

  1. Non-lexical Analysis

    • Relies on conversion between individual letters and their sounds.

    • Does not always work for all letters (in English)

  2. Direct Lexical …. (finish later)

IA model vs. DRC model

Understanding words

  • Lexicon = a person’s knowledge of what words mean, how they sound and how they are used in relation to other words.

The word frequency effect (Gardner et al., 1987)

  • We responf to high frequency words more quickly than low frequency words.

  • E.g., home (547 times per million words)….

robot