24
R.G. Bias | [email protected] | Name that tune. Song title? Performer(s)? 1

Name that tune

  • Upload
    caitir

  • View
    37

  • Download
    0

Embed Size (px)

DESCRIPTION

Name that tune. .  Song title?  Performer(s)?. Psycholinguistics. “The Knower” 2/1/10. Objectives. After this class you will be able to (it is my hope!): Describe some reasons why speech perception is hard. Describe how we still do it, seemingly easily. - PowerPoint PPT Presentation

Citation preview

Page 1: Name that tune

R.G. Bias | [email protected] |

Name that tune.

 Song title?  Performer(s)?

1

Page 2: Name that tune

R.G. Bias | [email protected] |

Psycholinguistics

“The Knower”2/1/10

2

Page 3: Name that tune

R.G. Bias | [email protected] |

ObjectivesAfter this class you will be able to (it is

my hope!): Describe some reasons why speech

perception is hard. Describe how we still do it, seemingly

easily. Understand the Stroop Effect. Understand much about how we read. Know what “phonological recoding” is. Demonstrate an even DEEPER

appreciation for how we can use behavioral data to make inferences about brain function.

Page 4: Name that tune

R.G. Bias | [email protected] |

Psycholinguistics The psychology of language. What goes on from the time I get an

idea until you have the same idea,–Whether I speak my idea (speech

production, auditory science, speech perception)

– Or write my idea (motor movements, visual system, reading)

Page 5: Name that tune

R.G. Bias | [email protected] |

Last week . . . We talked of selective attention. Get this – the Stroop Effect. I need a volunteer. http://iws2.ccccd.edu/lipscomb/

16_week_course/stroop_lab.htm

Page 6: Name that tune

R.G. Bias | [email protected] |

Speech perception . . . . . . is EASY!!

I need three volunteers to each read just a few words out loud.

Page 7: Name that tune

R.G. Bias | [email protected] |

Did anyone NOT hear . . . DOG FISHING ROD LADDER

So, speech perception must be easy, eh?

Page 8: Name that tune

R.G. Bias | [email protected] |

Yeah, consider this: To understand ONE word, you need to :– Transform sound waves into neural signals

(remember – tympanic membrane, middle ear bones [auditory ossicles], cochlea, basilar membrane, hair cells).

– Apply “pattern recognition routines” to figure out what letter a particular pattern of sounds represents.

– Hold early letters in working memory to put together one word’s worth of letters.

– Look up the meaning of that word in your mental lexicon.

Page 9: Name that tune

R.G. Bias | [email protected] |

Furthermore . . . . . . adults speaking English speak about

15 speech sounds (letters, but when spoken “phonemes”) per second, thus about 900 per minute.

You gotta pick out those sounds (from the speaker) from other ambient noise.

Plus, there are NOT blank spots between words, in the acoustical stimulus. (Notice how hard it is to parse individual words in a language you don’t speak?!)

PLUS . . .

Page 10: Name that tune

R.G. Bias | [email protected] |

Variability in Phoneme Pronunciation!!

Differences in pitch and tone of different speakers. (Men vs. women. Adults vs. kids. Tall people vs. short people. Native English speakers vs. not.)

Even an individual doesn’t always pronounce a phoneme the same way.

Plus, coarticulation effects – your articulators (lips, tongue, jaw) are different positions when you START to utter a particular phoneme, depending upon the context, and you’re preparing to pronounce the next phoneme. And soooo . . . – The phoneme sounds different from word to word!– So all those /d/ phonemes you perceived as the same are

actually QUITE different, acoustically (physically) – they’re firin’ different patterns of hair cells!

Speech perception is a miracle! How do we do it?

Page 11: Name that tune

R.G. Bias | [email protected] |

How do we do it? Word boundaries – we use our knowledge of our

language to insert those word boundaries that aren’t really there.

Visual cues Context. We are “active listeners.” More o’ this

“top-down” perception – we use our knowledge of language, and the context of the utterance, to perceive “ambiguous stimuli.”– For example – Warren and Warren (1970) study on

“phonemic restoration” – a cough replaced the asterisk below, but people had no problem getting the “right” word.• “It was found that the *eel was on the axle.”• “It was found that the *eel was on the shoe.”• “It was found that the *eel was on the orange.”

– So phonemic restoration is a sort of illusion.

Page 12: Name that tune

R.G. Bias | [email protected] |

So, in speech perception . . .

. . . we receive a less-than-crisp acoustic signal, this non-unique speech stimulus, and IMPOSE meaning on it by . . .– Building in those word boundaries–Utilizing visual cues, when available,

and– Attending to context

Page 13: Name that tune

R.G. Bias | [email protected] |

The Psychology of Reading Except for fairly rare cases of

“phonetic symbolism” (onomatopoeia) words have no inherent meaning.– (And rarer cases of “orthographic

symbolism”!!) So, READING is the interpreting of

words, the acts that go on to impose meaning, from within, on external visual stimuli.

Page 14: Name that tune

R.G. Bias | [email protected] |

Some facts about reading Eyes of the mature reader move rhythmically across the

page (from left to right). Eye movement consists of fixations, saccades, regressions,

and return sweeps. No information is taken in during saccades (10-25 msec),

regressions (same duration), or return sweeps (40 msec). During fixation (250 msec) a visual pattern is reflected

onto the retina. Span of perception = amount of print seen during a single

fixation. Span of perception = 12 letter spaces for good readers, 6

for poor readers.

Page 15: Name that tune

R.G. Bias | [email protected] |

More facts Span of recognition – 1.21 words for senior high, 1.33

words for college readers. So, 7 to 8 fixations per line of print. As content gets tougher, duration of fixations, not number,

changes (increases). Regressive movements aren’t systematic. Used when

attention is faltering. College readers have 1 regressive movement per 3 or 4

lines of print. Immature readers have 3 or 4 regressions per line.

Page 16: Name that tune

R.G. Bias | [email protected] |

Iconic Memory Remember in Week 1 I mentioned a two-stage memory

process – STM and LTM. A third stage, Iconic Memory: The unidentified, “pre-

categorical” pattern of lines, curves and angles; formed in about 100 msec. Lasts just about 500 msec.– Echoic Memory lasts about 2 sec.

Icon can hold up to 20 letter spaces. Pattern recognition routines are applied to the lines,

curves. It takes about 10 – 20 msec to read each letter out of the

iconic memory. Neural signal takes about 30 msec to go from the retina to

the visual cortex.

Page 17: Name that tune

R.G. Bias | [email protected] |

Iconic Memory (cont’d.) At some point, thanks to pattern recognition routines,

letters are read out. Letters are transformed into abstract phonemic

representations. The abstract phonemes are used to search the mental

lexicon. About 300 msec after the eye has fallen upon the page, the

first word is “understood,” i.e., placed in Primary Memory (STM, Working Memory).

Syntactic and semantic rules are applied to gain the meaning of the sentence.

Page 18: Name that tune

R.G. Bias | [email protected] |

How do you know, Randolph?

Psycholinguists employ a variety of methods to acquire this data about human behavior.

One question: Why do we think readers routinely transform the visual representation into a phonological representation?– Cognitive economy – all (healthy) new readers

come to the task as skilled hearers.– “I thought you said something about data?”

Page 19: Name that tune

R.G. Bias | [email protected] |

Rubenstein et al. (1971) Used a lexical decision task (word/nonword?). Two types of nonwords – homophonous (with real

words), like burd and nonhomophonous like rolt. Equally “wordlike.”

Longer latencies for burd. Similarly, longer for real homophones like meat. Pointed to “false matches” in the mental lexicon.

Page 20: Name that tune

R.G. Bias | [email protected] |

More Data McCusker et al. (1977) proofreading experiment– Homophonous typos (e.g., furst) went

undetected more often than nonhomophonous typos (e.g., farst).

Gough and Cosky (1977) used the Stroop task.– Nonwords homophonous with color words (e.g,. bloo)

led to more interference than control words (e.g., blot) or nonwords nonhomophonous with color words (e.g., blop).

I found readers took longer to process words with irregular “spelling-to-sound rules” (e.g., pint) than words with regular rules (e.g., hint) (Bias, 1978).

Page 21: Name that tune

R.G. Bias | [email protected] |

The Point The reasons for this somewhat

esoteric discourse on the psychology of reading are:– To communicate the complexity that is

human information processing– The illustrate the ways scientists go

about answering questions about info processing

– To sensitize you to the sorts of things known about human behavior

Page 22: Name that tune

R.G. Bias | [email protected] |

SO WHAT? Given that we’re so all-fired

complex, what does this have to say about how we design computer interfaces?– Depth cues.– Color perception.– Effects of context on perception.–What’s easy to read? – Recognition vs. recall.

Page 23: Name that tune

R.G. Bias | [email protected] |

Resources Matlin, M. W. (2009). Cognition.

Hoboken, NJ: John Wiley and Sons, Inc.

Bias, R. G. (1978). Phonological recoding of words in isolation and prose. Unpublished doctoral dissertation. The University of Texas at Austin, Austin, TX. (Wink.)

http://iws2.ccccd.edu/lipscomb/16_week_course/stroop_lab.htm

Page 24: Name that tune

R.G. Bias | [email protected] |

Today’s song was “Ten Thousand Words” by The Avett Brothers.

Why do you suppose we chose to play it before THIS class?

24