Language and reading

Language

Stroop (1935)

Experimental Methods

Lexical Decision:

Where Ps discriminate (using keypresses) between words and nonwords (or pseudowords). RT and accuracy rates are measured.

Naming:

A word is presented, and PS say the word aloud as quickly and as accurately as possible. RT and accuracy rates are measured.

Key finding: the RVFA

When a word is presented in the periphery, recognition is better in the right visual field (RVF) than in the left visual field (LVF).

Furthermore, words varying in length are equally well recognised in the RVF. In the LVF, longer words take longer to recognise.

Right visual field advantage:

Word processing in the periphery is characterised by a dissociation: RVF (LH) words are processed efficiently; LVF (RH) words are not

Experimental methods:

Eye movement recording: Where Ps read text (silently/aloud) while eye position is tracked. Yields a range of information…

Neurological/Neuropsychological/Neuro-scientific Techniques: Any one of the above tasks implemented while neural activation is recorded (fMRI, MEG, EEG). We can then examine the cortical bases of word recognition.

Patients with brain trauma can provide insights into cortical localisation of language...

Language and the left hemisphere

This was originally discovered by Paul Broca, a French physician.

A patient called Leborgne suffered progressive loss of speech, with all other cognitive functions left intact.

“Tan”

A later autopsy revealed frontal lesion in the LH. Later cases confirmed the neural locus of articulated language.

Visual Word Form Area:

Cohen et al. (2002)

Key findings: Cerebral Specialisation

The human visual system is contralateral in nature.

The right visual field (RVF) initially projects to the left cerebral hemisphere (LH); the LVF projects to the RH.

Hubel and Wiesel (1959)

Recorded cell activity in the primary visual cortices of anaesthetised cats. Cells selectively respond to lines of different orientations.

Interactive activation Model (McClelland & Rumelhart, 1981)

Orthographic Neighbour: A word that differs from a target word by a single letter in the same position

The masked prime procedure (e.g., Forster, 1987)

<30 ms = orthographic information (Grainger & Ferrand, 1996)

~60 ms = phonological information (Ferrand & Grainger, 1993)

Interim Summary

Humans are highly adapted for language and reading, as evidenced by the Stroop task, word superiority and cortical specialisation.

The experimental paradigms strongly associated with this field of enquiry are lexical decision tasks, and the masked priming procedure.

One of the key theories of word recognition is McClelland and Rumelhart’s (1981) interactive activation model.

Now we move on from recognising individual words to reading strings of words…

Reading

Optimal viewing position (AKA perceptual span)

When reading text, fixations tend to land towards the initial 3-4 letters of a word and up to 15 letters from the end (Raynor, 1975).

One theory held this was to maximise the number of letters in the RVF (and projected to the LH).

Key findings: cerebral specialisation

The human visual system is contralateral in nature.

The right visual field (RVF) initially projects to the left cerebral hemisphere (LH); the LVF projects to the RH.

Paterson (2014) compared English (left->right) and Urdu (right->left) readers, manipulating the amount of text visible within a sentence.

If OVP is to maximise projection to the LH, both reader types should benefit from visible window extension to the right.

If other factors give rise to OVP then English readers should benefit from right extension; Urdu readers a left extension.

Eye Movements and Reading

Fixation: Where the eyes remain stationary on an object, location, or word. To be considered a fixation, the eyes have to be static for ~250 ms.

Saccade: An eye movement made between fixations. A saccade typically takes around 20-30 ms.

How do we read?

The “obvious” assumption is reading is a serial process:

• Fixate a word (N), recognise N, saccade to N+1, recognise N+1…

However, the numbers don’t quite add up (200ms to recognise a word + 85-200ms to prepare a saccade = 400 ms per word?!)

Eye movement patterns reveal:

• reading is a rapid process, spending considerably les than 400ms on each word

• scan pattern unpredictable

• text alignment has huge effect on scan pattern

The E-Z Reader model (Reichle et al., 1998) accounts for observed scan patterns during reading.

• The model assumes an interplay between attention, word processing, and saccade programming (in pretty much that order).

Evidence

Reingold et al (2012): monitored the eyes of participants as they read passages such as:

During a saccade towards TABLE, the word was replaced with an unexpected word (BANJO).

• If the model is correct, TABLE will have received some attention, BANJO would not.

Although Ps did not consciously detect the word change, their eyes fixated on BANJO longer than TABLE.

Critisisms

The Cambridge email (Rawlinson, 1976)

Acorcrdnig to reaserch at Cabmrdige Unvieristy it deosn’t mtater in waht odrer the lettres ina wrod are. The olny imroptant thnig si that the fisrt and lsat lettres be in the righit plcae. The rest canbe a totle mses and you can stili read it witohut prolbem. This is becusse the huamn mind deos not raed evry lettre by itslef butthe wrod asa wohle.

The theory focuses on the interplay between attentional orienting and saccades, but says little about the lexical processing taking place once a word is attended or fixated.

Summary

A range of techniques (behavioural, neurological, neuropsychological) can reveal the loci and timing of word recognition processes.

The interactive activation model describes how bottom-up and top-down processes combine to enable individual word recognition.

The E-Z reader model describes how we process multiple words (i.e., read sentences). Focuses on attention and saccades, but doesn’t explain higher lexical processing (semantics, syntax).