Classic Study- Baddeley 1966


  • Previous studies into STM found that when words are acoustically similar, recall is negatively affected, more so than when words were semantically similar 
  • Questions raised whether STM and LTM are as separate as first thought because interference has been shown to cause forgetting in both
  • Because interference affects recall in both STM and LTM, to look at STM learning and recall must occur within around 30 seconds, for LTM it needs to be beyond 30 seconds 

Baddeley had to conduct a 3 part study to ensure the credibility of the results because of confounding variables had occured in the first 2 

1 of 9

Aim and Procedure- Experiment 3


  • To see if LTM was like STM --> if in LTM acoustic similarity of words would lead to more memory impairment than with semantic similarity of words 


  • laboratory experiment


  • independent measures - 4 different groups, 1 condition (1 list of words)

IV --> the type of word list given

DV --> the recall of order of word list (remember correct order of words)


  • male and females selected from 'The Applied Psychology Research Unit'
2 of 9

4 Conditions- Experiment 3

A --> Acoustic Similarity (experimental)

  • participants learned words that sounded similar to each other 

B --> Acoustic Dissimilarity (control)

  • participants learned words that didn't sound similar

C --> Semantic Similarity (experimental)

  • participants learned words that mean the same kind of thing

D --> Semantic Dissimilarity (control)

  • participants learned words that didn't mean the same thing but are used as much as those in condition C 
3 of 9

How the 4 conditions were conducted

  • Each list of 10 words presented via projector- 1 word every 3 seconds in the correct order
  • After, had to complete 6 tasks involving memory for digits
  • Asked to recall the word list in 1 minute in the correct order 
  • Repeated over 4 learning trials (repetition- ensure words had gone into LTM)
  • Word list in a random order visible on a card because the test was based on order rather than knowing the words 
  • After the 4 learning trials, 15 minute interference task- copy 8 digit sequences at their own pace - this was to prevent rehearsal 
  • Participants were then given a surprise retest on the word sequence 
4 of 9


Acoustic Similarity:

  • Hard to learn list of acoustically similar words in STM
  • Trials 3+4 there was a smaller effect - acoustically similar- no significant difference 

Semantic Similarity:

  • Trial 1&2 no effect in STM- control and similar list had similar results 
  • Trial 4 and retest, control did significantly better than the similar list (% recall lower for semantically similar words)
  • Foe the similar list, there was no difference in recall between trial 4 and the retest 
5 of 9


  • Acoustically similar sounding words were more difficult to encode --> participants found it more difficult to recall list A in in initial phase of learning so STM is largely affected 
  • Encoding in the LTM is largely semantic --> C&D during retest, semantically similar list impaired 
  • So, STM and LTM are affected differently by different types of coding 

-Learning of word sequences impaired by semantic similarity

-LTM based on semantic encoding, STM based on acoustic encoding

-Transferral to LTM involves an intermediate stage where material is in STM. This is shown by greater difficulty in learning list in experiment 3 when STM is minimised 

6 of 9

Evaluation Points


  • The study is low in generalisability because participants were males and females selected from The Applied Psychology Research Unit. This means that the results are limited due to all of the participants being from a psychology research unit, so results may not apply to those from a non-psychology background
  • The study is low in generalisability. Baddeley's sample size was small- less than 20 participants in the key groups and they were all volunteers. This can produce a biased set of results as participants would be more focused and potentially having something in common- more likely to volunteer. So, the results may not be representative of the target population.


  • The study can be tested for reliability. Baddeley conducted the study in a highly controlled lab environment in which he used a standardised procedure e.g. all participants learnt the same no of words (10)and the same prevention of rehearsal tasks to complete. There was control over extraneous variables such as noise. This means the study can be replicated by other researchers to see if the same results can be obtained. 
7 of 9

More Evaluation Points


  • The findings can be applied to real life such as in education. The finding that we encode semantically in our LTM has enabled teachers/professionals to promote the best learning strategies to improve memory. This means students can use techniques such as mindmaps to create links and ensure that information being learnt has a meaning so it can be stored


  • The study is low in external validity because the experiment took place in a lab, which is an artificial environment where participants would not have been behaving naturally. This means participants could have been showing demand characteristics and so the study does not measure real life behaviour
  • The study is high in interal validity due to the highly controlled nature of the experiment e.g. timing how long a stimulus is visible for. This means the extraneous variables were controlled and the researcher can be more sure that it is only the IV having an effect on thr DV. So, cause and effect can be established.
8 of 9


  • Some ethical guidelines were broken. The surprise recall task at the end of the study may have breached informed consent as participants were unaware that this would be required. This means that the participants were deceived, however this was necessary to gain accurare results
9 of 9







This is good. I've been trying to find this experiment in this much detail.

Similar Psychology resources:

See all Psychology resources »See all Cognitive Psychology resources »