Word Identification

• Phonology (sounds) - Physical characteristics of speech & Orthology (Written representations - letters, characters).
• Morphology - Meaning units
• Lexicon - Words
• Syntax - Grammar, structure
• Semantics - Meaning
• Pragmatics - Communicative context

Without identifying the sounds, you cannot understand language. Syllables need to be extracted and combined into morphemes.

Phonemes are a unit of phonology - they are the smallest units of sound in speech that help to distinguish one word from the other.

The same phoneme can sound different depending on:

• Who says it
• How they say it
• What other phonemes are nearby (coarticulation).
• The written representative (e/E/e)

The French physician Paul Broca (1861) reported a patient who could no longer produce speech - he could only say 'tan'. 

Patient Tan died a few days after Broca saw him and performed an autopsy, discovering a massive lesion in the left frontal lobe.

Broca concluded that this area must be associated with speech production (Broca's Area).

Broca's Aphasia

• Slow, halting speech.
• Poor articulation
• Speech omits propositions, conjunctions and other grammatical filler words

Wernicke's Aphasia

• Reduced comprehension of spoken and written language
• Non-meaningful speech and difficulty finding right word (anomia).
• Patients speak fluently and effortlessly, but pronounce phonemes in the wrong order sometimes.
• Will utter novel words or neologisms (a newly coined word or expression)
• Patients who had a temporary lesion of Wernicke's area say that they were unable to understand others' or their own speech but could not stop talking.

• When connections between levels are weaker, phonological substitutions and nonwords are used more frequently.
• Damaged connections may be the cause for this, leading to more errors, due to an inability to inhibit certain connections.

The interactive activation model created by Rumelhart & McClelland (1982) describes how perception of words results from excitatory and inhibitory interactions of detectors for visual features, letters and words.

• A visual input excites detectors for visual features in the display.These excite detectors for letters consistent with the active features.
• The letter detectors in turn excite detectors for consistent words.
• Active word detectors mutually inhibit each other and send feedback to the letter level, strengthening activation and hence perceptibility of their constituent letters.

The model also produces facillitation for letters in pronouncable pseudowords as well as words. Pseudowords activate detectors for words that are consistent with most of the active letters, and feedback from the activated words strengthens the activations of letters in the pseudoword.

Combining bottom-up and top-down information helps to deal with perceptual noise.

Recognizing letters

• Word and letter detectors at different levels
• Most important step: abstract letter detection

Location of word detectors

• There are high-level visual detectors for objects, faces, houses and words (visual word form area)