2. Parallel Activation in Spoken Word Recognition

How is eye tracking used to measure parallel activation when hearing words?

• people are more likely to look at pictures of the candidates than unrelated words

How did Dahan demonstrate eye tracking activation of sound matches?

“find the bench”

• in the process of hearing bench, they might also activate bed and bell

• multiple words matching the input are activated

• unrelated “apple” is not looked at much

How did Yee and Sedivy demonstrate eye tracking activation of semantic matches?

• a decent amount of time is spend looking at the key (semantically related)

• some continuation of looking even when they’re sure the word they’re looking for is look

In sum, what has been found from eye tracking studies of word activation?

• even as a word unfolds we get activation of similar sounding words and words similar in meaning

However, what could be a limitation of eye tracking studies?

• maybe even seeing the words in the study is partially responsible for activation of the other words

What is the effect of competition in parallel processing?

• different words might have different numbers of other similar words

• if a word sounds similar to many others there will be lots of competition - more difficult to recognise and vice versa

How did Ziegler et al demonstrate parallel activation competition for sounds?

• played recording of words: with lots, little competition or fake words

• had to indicate whether the word they heard was real or not

• slower to respond when words have many sound competitors

How did Zwitserlood demonstrate parallel activation and semantic priming?

• asked to recognise target words on a screen, sometimes real and sometimes not

• if e.g. cold and food are activated in ppts mind they would be more likely to respond faster to say that they are real than unrelated words

• and people are faster to recognise targets that are semantically related to words they hear at the same time

• we see priming for multiple related words

Parallel activation evidence summary

• competition shows parallel activation of similar-sounding words

• priming shows parallel activation of semantically related words

How did Rodd et al demonstrate competition for meanings?

• slower to recognise words with many meanings

• faster with few meaning competitors

Summary: parallel activation

• similar-sounding words

• semantically related words

• multiple meanings of the same word

What are the 3 key principles from speech to meaning?

1. robust identification of words despite variable boundary cues, variability in speakers and variability in noise

2. parallel activation of multiple candidates

3. context influences word recognition

• context constrains parallel activation of words/meanings

How did Swinney demonstrate whether context impact activation of different possible meanings?

• listen to ambiguous words like second, always presented in sentence context where just one meaning was relevant

• shown target words on screen, some real some not and have to decide which are real

• words were either related to the relevant meaning due to the context, or the irrelevant meaning

• do both of the meanings get activated?

• target shown either during or after ambiguous word

• depends when exactly target word was shown to ppt

• when shown after, priming for the irrelevant target goes away

• as people hear ambiguous words multiple meaning generated, regardless of context. but within a few moments it is then constrained down to the one possible option

• context rapidly constrains parallel activation

What are the differences between hearing and reading words?

• hearing: no clear pauses, universal

• reading: clear spaces, invented/taught

What is involved in robust identification of written words despite variability?

• text tends to be standardised

• the same sounds can be written in lots of different ways

• even with jumbles letters words can be processed and understood

How is masked priming used to test letter position variability?

• presented without conscious comprehension

• this masked priming can still affect recognition of the target word

• still get priming with a jumbled word but not when replacing one of the letters

How does parallel activation relate to text?

Do we still see activations of multiple words or just the one we’re seeing in front of us?

Rodd (2004)

had to respond whether words were animals

does reading a word like leopard activate other similarly spelled words

• people are slower when their is competition

• similarly spelled words are activate in our minds when we are reading

• parallel activation when it comes to recognising words in text

• form competition

What is the phoneme restoration effect?

• context of surrounding word/sentence seems to restore missing phonemes in a listeners mind

What did Heilbron et al find regarding the letter restoration effect? influence of context in text?

• trying to recognise degraded letters either embedded in a real word or nonword

• people did recognise letter better in the context of a real word vs nonword

• fMRI evidence showed that the real word restored the letter representation even in early vision

• when people see degraded text first processed in visual region, and then temporal information storing region, then see a top down effect. recognising the word seems to go down and boost the activation in early visual areas

• only happening when recognising the word and not with random word

Conclusions

• parallel activation in both spoken and visual word recognition

• context influences in both spoken and visual word recognition

• flexible processing of visual words in the face of variability