Psychology of Reading

Role of Speech in Reading

* In earlier days, people would not have questioned that talking in inherently linked to reading - silent reading was rare:

* St. Augustine in his "Confessions"

- remarks about monk Ambrose to reading without obvious speech

- "But while reading, his eyes glanced over the pages, and his heart searched out the sense, but his voice and tongue were silent.

Role of Speech in Reading

* Practical importance of this issue -

- How do we teach reading

* phonics method emphasizes grapheme/phoneme conversions

* whole-word method- emphasizes direct connection between the written word as a patttern and its meaning

- How do we teach the deaf reader

- How do we deal with dialect mismatch

Why Propose Speech Process in Reading

* Do you hear your voice or others during reading?

* Recording electrical impulses on skin that lies directly over speech muscles (electromyography or EMG)

- while reading, these muscles are activated!

- Hardyck and Petrinovich '70 study

* Used biofeedback to reduce EMG activity in larynx (Adam's apple)

* suppressed EMG activity led to poorer comprehension of difficult material, but not easy material. Why?

Levels of Speech Representation

* Phonetic Level

- Represented by "Phones" - universal

set of speech sounds. Eg., sounds of the letter 't' in Table, little, cat. Subtle differences that we can not readily recognize in the sounds for 't' - thus separate phones needed.

- Phonological level represented by 'phonemes'.- speech sounds that we can actual recognize. So the 't' sound is represented among all sounds for the 't' sound. Overhead162

Levels of Speech Representation

* Syllable - smallest segments of speech that can be articulated independently. Usually contains a vowel and a consonant.

Evidence for Speech Coding

* Conrad's effect: harder to memorize similar sounding letters, than different sounding letters.

- B,v,t,z,v,z harder to remember than s, t, n, w, q

* Using the speech code is automatic, even in languages without a phonologically based script.

- Tzeng, Hung, and Wang (1977) - Chinese ideographs become confused in memory if they sound similar!

Evidence for Speech Coding

* Lexical decision task:

- speed with which we identify if presentation is a word

- Illegal and nonpronouncable nonwords faster to identify (likj, sagm)

- Slower when nonwords were legal and pronounceable (strig, barp).

- Even slower when nonword homophones presented (e.g., trate, tew).

- Suggests that printed word is translated into some form of speech and then lexicon is consulted.

* So not a direct route from word to meaning without speech overhead167

Evidence for Speech Coding

* Forced choice - Hawkins, Reicher, Rogers, & Peterson, 1976

- Subject presented very briefly a word such as 'coin". Then subject presented word pairs such as join-coin.

- Person indicates which word was presented first - coin or join.

* Person must decide among alternatives (i.e., forced choice)

- What if subject is presented 'cent' real fast, and then is presented cent-sent. Should that be harder than cent-lent. Why?

Evidence for Speech Coding

* Regularity effect: faster to name regular words than irregular words

- Regular words- can be read via the grapheme-to-phoneme correspondence rules, such are 'mode', 'stop'.

- Irregular words- can not apply GPC rules. E.g., deaf

- For irregular words there is a conflict between the GPC pronunciation and the correct pronunciation (e.g., 'deaf' being read like ' leaf')

Speech for Short-Term Memory

- Phonetic - the specific sounds that we can not necessarily perceive, such as the p sound in poke and spoke.

- Phonological - more abstract representation of sounds that we can perceive. The p phoneme is why we perceive similar p sound in poke and spoke.

* After lexical access, keeping speech based representations in working memory is very useful:

- One reason is that sentences often have large distances between related words

- 'Throw the horse over the fence some hay'

Speech for Short-Term Memory

* Kleiman (1975). Asked subjects to 'shadow' digits while also identifying presented word pairs

- Digits presented via earphones, and subject repeated the digits orally.

- Word pairs presented and subject made times yes/no decisions concerning

* Phonemic level: tickle/pickle sound similar, while lemon/demon do not sound similar.

* Graphemic: heard/beard look similar while grace/price do not

* Semantic: mourne/grieve meaning similar but depart/couple do not

- Main