scispace - formally typeset

Journal ArticleDOI

Word length effects are not due to proactive interference.

01 Mar 2002-Memory (Taylor & Francis (Psychology Press))-Vol. 10, Iss: 2, pp 139-149

TL;DR: No empirical support was found for PI as an explanation of the word length effect and strong word length effects were present but there was little evidence for PI influencing either overall levels of recall or the wordlength effect.

AbstractIn immediate serial recall short words are better recalled than long words. The word length effect has become pivotal in the development of short-term memory models. The current research tests one explanation of the word length effect; that it is related to proactive interference (PI). We report two experiments in which the relationship is directly tested. In the first experiment we show that word length effects can be observed over the first few trials in an experiment and that the effect shows itself primarily in the number of omissions made. In the second experiment we simultaneously test for PI and word length effects. Strong word length effects were present but there was little evidence for PI influencing either overall levels of recall or the word length effect. In short, no empirical support was found for PI as an explanation of the word length effect.

Topics: Interference theory (51%)

Summary (3 min read)

Materials

  • Each trial in the experiment consisted of five words.
  • These words were all disyllabic words that had been matched for word frequency.
  • The four trials of short Word Length and PI 8 words were created by randomly assigning the five words to the five serial positions on each trial.
  • To create the second set trials another two five-word pools were created.
  • The order of the 50 trials was then randomised for each subject.

Procedure

  • The participants studied the six sets of trials across two days.
  • The subjects were all tested in a single group.
  • Given that it took less than two minutes to present the four trials and that there was at least a two-hour break between sets of trials, the expectation was that there should not be any PI between the different sets of materials.
  • Subjects were instructed to write from left to right and to leave blank spaces for any words they could not remember.

Results

  • Recall was scored as correct only if the word was recalled in its correct serial position.
  • The authors also compared performance on the very first trial in each group.
  • There was the odd extra list intrusion that usually had similar phonological characteristics to the forgotten target, the odd repetition of a word in the list, and the odd intrusion from a prior list or a perseveration of an earlier response, but these errors were very infrequent.
  • The mean number of errors collapsed across serial position is depicted in Figure 1 , and were analysed in the same way as the targets.

Discussion

  • The authors replicate the Nairne et al. (1997) findings that word length effects do not emerge over four trials when the Cowan et al. words are used as experimental stimuli.
  • Furthermore, word length effects can be observed on the first trial in a block and across the subsequent three trials.
  • That is, there is very little direct evidence for PI effects in the task, yet there are strong word length effects with the three and seven phoneme words.
  • Secondly, although there are differences in the number of intrusions, intrusions represent a minor source of error in the task.
  • As was the case in Experiment 1, more omission errors were made with long words that with short words.

Experiment 2

  • In Experiment 2 the authors explore the same issues but with a task that directly measures PI effects.
  • The two-block trials are the critical trials for manipulating PI.
  • On these trials the second block always contains four words that have to be recalled in serial order.
  • In the no-interference condition the first block contains four consonants.
  • The choice of consonants and words as appropriate materials was based on the fact that these materials produce reliable release from PI effects in the Brown-Peterson task (Wickens, 1972) .

Word Length and PI 15

  • To create the items in the closed pool list, the authors first randomly selected 16 items from each of the 120 item word pools and these 16 items then became the new short and long word pools used for creating the interference and no-interference trials.
  • Creating the lists involved similar procedures to those used in the open pool condition, save that on each trial the relevant 16-item word pool was accessed and items were selected without replacement on each trial.
  • This ensured that the short and long words were repeated frequently during the course of the experiment.
  • Again 10 one-block trials were added and the order of the trials was randomised.

Procedure.

  • Participants were told that they would study a series of one or two-block trials and they only had to remember the most recent block that they had seen.
  • Thus, when they discovered that a trial was a two-block trial they were to forget the first block and concentrate on remembering the second block because it would be on this block that they would be tested.
  • The list items were then presented at a rate of 1 item per second.
  • Subjects had a prepared answer sheet on which to record their answers.

Word Length and PI 16

  • The results of the present experiment, where performance has been collapsed over serial position, are shown in Figure 2 .
  • A 2*2*2 repeated measures ANOVA with interference, word pool and word length as within-subject was conducted on correct recall.
  • To solve this problem the authors collapsed block-1 and prior list intrusions into one group.
  • For combined block-1 and prior list intrusions, there were more intrusions in the interference condition than the no-interference condition, F (1,19) = 20.

General Discussion

  • The current research was designed to test one explanation for the word length effect.
  • Nairne et al.'s (1997) findings that word length effects were not present across the first four trials in an experiment was taken as evidence for the proposition that word length was related to the build-up of PI.
  • In Experiment 2 there was no difference in target recall between the interference and no interference trials.
  • There is some indirect evidence for PI effects in the first experiment.
  • The analysis confirmed that short words were better remembered than long words and that performance on trial one was reliably better than performance on trial two.

Word Length and PI 20

  • One way that the PI explanation could be maintained is if it is assumed that in Experiment 1 the PI effects are experiment wide and word length effects emerge across the experiment.
  • On balance, the research indicates that robust word length effects can be observed where there is little evidence of PI.
  • Long words are more likely to be omitted from the recall protocol than short words.
  • Given that there are probably many different possible factors that could produce an omission error, the finding of differential omission errors is probably not all that helpful.
  • They found that under some circumstances the two items could effectively block each other such that neither item could be recalled.

Did you find this useful? Give us your feedback

...read more

Content maybe subject to copyright    Report

Word Length and PI 1
Complete Citation: Tehan, Gerald and Turcotte, Josée (2002). Word length effects
are not due to proactive interference. Memory, 10 (2), 139-150. ISSN 0965-8211.
Accessed from USQ ePrints
http://eprints.usq.edu.au
Word Length Effects Are Not Due to Proactive Interference
Gerald Tehan
University of Southern Queensland
Toowoomba, Queensland, Australia
Josée Turcotte
Laval University
Québec, Québec, Canada
Mailing Address:
Gerry Tehan
Department of Psychology
University of Southern Queensland
Toowoomba, 4350
AUSTRALIA
Phone number: (07) 4631 2376
e-mail: tehan@usq.edu.au

Word Length and PI 2
Abstract
In immediate serial recall short words are better recalled than long words. The
word length effect has become pivotal in the development of short-term memory
models. The current research tests one explanation of the word length effect; that it is
related to proactive interference (PI). We report two experiments in which the
relationship is directly tested. In the first experiment we show that word length effects
can be observed over the first few trials in an experiment and that the effect shows
itself primarily in the number of omissions made. In the second experiment we
simultaneously test for PI and word length effects. Strong word length effects were
present but there was little evidence for PI influencing either overall levels of recall or
the word length effect. In short, no empirical support was found for PI as an
explanation of the word length effect.

Word Length and PI 3
Word Length Effects Are Not Due to Proactive Interference
Across the years there has been continual debate concerning whether
forgetting is caused by decay or by interference. One interesting facet of this debate is
that interference effects are usually seen as a prime source of forgetting in long-term
memory settings, but decay is seen as the principle means in immediate memory
settings (Baddeley, 1986; Burgess & Hitch, 1996; Henson, 1998; Page & Norris,
1998). Consequently, over the last thirty years or so, the vast majority of models of
immediate memory have assumed that the short-term memory trace that supports
immediate recall decays very rapidly, unless it is renewed by verbal rehearsal. Not
surprisingly these models are now often referred to by the generic name of "trace
decay plus rehearsal" models (Brown & Hulme, 1995).
For those who argue in support of decay, the word length effect, the fact that
span for short words is larger than span for long words, is one of the key short-term
phenomena. In the original research that established this effect, Baddeley, Thomson
and Buchanan (1975) first established a span advantage for one-syllable words over
five syllable words. However, in subsequent experiments they established that the
prime determinant of the word length effect was the spoken duration of the words.
Span for the short duration words was significantly larger than span for the long
duration words. On the basis of such findings the decay plus rehearsal models assume
that short-term memory traces rapidly decay in real time. Given that per given period
of time more short words can be rehearsed than long words, more short words can be
kept in an active state than long words.
In recent years, however, the trace decay plus rehearsal assumptions have
come under increasing amounts of pressure. The initial research that implicated
spoken duration has proved difficult to replicate (Lovatt, Avons & Masterson, 2000).

Word Length and PI 4
Word length effects have been found when pronunciation rates have been controlled
for (Caplan, Rochon & Waters, 1992), or rehearsal has been prevented (LaPointe &
Engle, 1990) and there are instances of where there are no-span differences where
there are clear differences in pronunciation rates (Service, 1998). Cowan et al (1992,
see also Dosher & Ma, 1998) have suggested that word length effects occur during the
recall process itself rather than during rehearsal prior to recall. In short, simple decay
notions appear to be inadequate as either a necessary or sufficient explanation for the
word length effect.
Currently, there are three alternative explanations for the word length effect,
two of which are based upon interference assumptions. Neath and Nairne (1995)
suggested that word length effects might result from intra unit interference. Melton
(1963) had demonstrated that items within a study list produced mutual interference
on each other. Neath and Nairne extended Melton's idea to features within words.
They suggested that words had to be compiled from sets of smaller features and that
the more features that had to be compiled, the greater the likelihood that an error
would be made. Since long words were assumed to require the compilation of more
features than short words, these words would be more error prone. Neath and Nairne
incorporated these ideas into the feature model (Nairne 1990) and were able to
provide good fits of existing data. The feature model with its assumptions about intra
unit interference has subsequently made novel predictions concerning the conditions
under which the word length effect would and would not be found and the data have
so far been consistent with the predictions of the model (Neath, Surprenant &
LeCompte, 1998)
The second interference-based explanation is that proactive interference (PI)
plays a role in producing the word length effect (Nairne, Neath & Serra, 1997).

Word Length and PI 5
Nairne et al. argued that if decay was the causal factor underpinning the word length
effect the effect should be as strong on the first trial of an experiment as on the last.
Consequently, they examined performance on a trial-by-trial basis. In their first
experiment they presented subjects with four 5-word lists that consisted of two-
syllable words that differed in spoken duration (Cowan et al., 1992). 220 students
were given four trials of short words and another group of 220 students was given
four trials of long words. No word length effects were present on any of the trials.
This result could be simply interpreted as another failure to replicate previous
findings, or simply be a cohort difference. However, these explanations were ruled
out in the second experiment where subjects studied 24 trials of either short or long
words. Again word length effects were absent on the first four trials, but they did
emerge on the next block of four trials and were also present on the remaining blocks
in the experiment. This finding that word length only emerged after four trials was
clearly problematic for trace decay plus rehearsal explanations. Moreover, the result
was reminiscent of Keppel and Underwood's (1962) finding that forgetting in the
Brown-Peterson task gradually emerged over three or four trials. Given this
correspondence and the fact that the Keppel and Underwood data are generally
attributed to PI, they suggested that word length effects could be related to PI. Thus,
both PI and word length effects build up over trials.
From a theoretical perspective, while the similarities between the Nairne et al
and the Keppel and Underwood are enticing, it is hard to see how word length effects
could be incorporated within standard trace discrimination explanations of PI. The
standard explanation of PI in the Brown-Peterson task is that on the first trial there are
only a small number of items available for recall. There are no problems in
discriminating these items from other items in the experiment because as yet there are

Citations
More filters

Journal ArticleDOI
TL;DR: It is suggested that the word length effect may be better explained by the differences in linguistic and lexical properties of short and long words rather than by length per se.
Abstract: The word length effect, the finding that lists of short words are better recalled than lists of long words, has been termed one of the benchmark findings that any theory of immediate memory must account for. Indeed, the effect led directly to the development of working memory and the phonological loop, and it is viewed as the best remaining evidence for time-based decay. However, previous studies investigating this effect have confounded length with orthographic neighborhood size. In the present study, Experiments 1A and 1B revealed typical effects of length when short and long words were equated on all relevant dimensions previously identified in the literature except for neighborhood size. In Experiment 2, consonant-vowel-consonant (CVC) words with a large orthographic neighborhood were better recalled than were CVC words with a small orthographic neighborhood. In Experiments 3 and 4, using two different sets of stimuli, we showed that when short (1-syllable) and long (3-syllable) items were equated for neighborhood size, the word length effect disappeared. Experiment 5 replicated this with spoken recall. We suggest that the word length effect may be better explained by the differences in linguistic and lexical properties of short and long words rather than by length per se. These results add to the growing literature showing problems for theories of memory that include decay offset by rehearsal as a central feature.

77 citations


Journal ArticleDOI
TL;DR: Except for the within-list semantic similarity condition, there was a buildup in PI levels in the form of protrusion errors across trials, but the magnitude of the WLE did not increase with the PI buildup, suggesting that it was not affected by PI across trials.
Abstract: We examined the influence of semantic similarity and proactive interference (PI) on the word length effect (WLE) in immediate serial recall. Word length was manipulated by comparing memory for monosyllabic versus multisyllabic words. PI effects were evaluated by manipulating semantic similarity in the to-be-remembered lists and examining its impact on the WLE’s magnitude across eight-trial blocks. Words were sampled from a single semantic category across the entire block, from a single category within the list, or from different categories. Robust WLEs were observed in single-category blocks and when words were from different categories. However, when all the within-list words were from the same semantic category, the WLE was sharply attenuated. Except for the within-list semantic similarity condition, there was a buildup in PI levels in the form of protrusion errors across trials. However, the magnitude of the WLE did not increase with the PI buildup, suggesting that it was not affected by PI across trials.

6 citations


Journal ArticleDOI
TL;DR: The reading rate is not affected by changing the wavelength of the light, however, the mean differences in wpm were affected byChanging the wavelengths, and introducing positional noise affects word recognition differently with different wavelengths.
Abstract: BACKGROUND: Reading can be described as a complex cognitive process of decrypting signs to create meaning. Eventually, it is a way of language achievement, communication, and sharing information and ideas. Changing lighting and color are known to improve visual comfort and the perceptual difficulties that affect reading for those with poor vision. AIM: This study aims to investigate the effect of changing the wavelengths and different levels of positional noise on reading performance for participants with best-corrected distant visual acuity (BCVA) of 6/6 or better. METHODOLOGY: Twenty English speakers with BCVA 6/6 or better were asked to read words presented in a printed format. The stimuli were black print words in a horizontal arrangement on matte white card. They were degraded using positional noise produced by random vertical displacements of the letter position below or above the horizontal line on three levels. RESULTS: Introducing positional noise affected word recognition differently with different wavelengths. The role of short wavelength in enhancing orthographic reading and word recognition is clear – they reduce the effects of positional noise. The error rate and duration time have different effects with different wavelengths, even when positional noise is introduced. CONCLUSION: The reading rate is not affected by changing the wavelength of the light. However, the mean differences in wpm were affected by changing the wavelengths. Also, introducing positional noise affects word recognition differently with different wavelengths.

1 citations


Dissertation
01 Jan 2004
Abstract: When participants are presented with lists of items for immediate serial recall, tradition would suggest that a race begins - between the need to constantly refresh or recycle the memory trace of that list, and a tendency for the memory trace to decay. Standard models in the literature assumed a complex interaction of mental subsystems whereby a controlling attentional process strove to keep the memory of such a list alive for a sufficiently long period of time so it could be remembered and output in order, using a type of recirculating loop rehearsal and storage mechanism to offset the decay process. Evidence supporting such models stemmed from the observation that more short words could be remembered in order than long words (Baddeley, Thomson, & Buchanan, 1975). This word length effect, described in the second chapter, was a crucial piece of evidence for rehearsal and decay models, in the example given, the recirculating loop was seen as being time-based and extremely limited in capacity, such that memory was deemed equivalent to the amount of information which could be cycled through the rehearsal loop in about two seconds. A number of recent challenges to this model of remembering have cast doubt on the nature of the process as described in such models as that of Baddeley (1990; 1996). Chapter 1 began by providing an overview of the development of such models from their earliest form, and also introduced some alternative ideas about the structure and function of human memory. A processing view was described, in which the probability of recalling a list of items depended not upon a race between decay and rehearsal, but on differential processing of items based on their nature. As remembering a list in its original order involves not only remembering the items themselves, but also information about how they relate together in the list, an alternate theory was advanced that in some cases the processing of information about the items, and information about their serial order could dissociate, producing a processing tradeoff. As individual items were better remembered, information about their presentation order diminished. This observation (Nairne, Riegler, & Serra, 1991) was introduced as the item-order hypothesis. The item-order hypothesis suggested that under certain conditions increased item processing could lead to deficits in order processing, and that this produced a dissociation in performance between item and order memory tasks. The generation effect (Slamecka & Graf, 1978) was one such example, as was the perceptual interference effect (Mulligan, 2000), and these were discussed in Chapter 3. The word length effect was seen as another instance where this tradeoff might be observed. A design incorporating elements of item and order tasks based upon Nairne et al. (1991) was detailed in the fourth chapter, leading on to empirical testing of the word length effect (Chapter 5), the generation effect (Chapter 6) and the perceptual interference effect (Chapter 7). This series of experiments compared word length and generation effects under serial recall and single item recognition tasks, using a range of test conditions designed to allow replication and extension of existing data from these separate streams of research. Results did not appear as predicted for some aspects of generation and all aspects of perceptual interference, and further experiments in Chapter 8 attempted to address the current findings. For the experiments involving word length, short words were better recalled than long words on the serial recall task, but long words were better recognised in the recognition task. Following additional manipulations in Chapter 8, the generation effect began to produce a similar pattern, but the results for perceptual interference were inconclusive. Word length data, however, were consistent with the item-order approach and supported a novel explanation for the word length effect. Implications and conclusions were discussed in Chapter 9.

Cites background from "Word length effects are not due to ..."

  • ...Tehan and Turcotte (2002) have recently attempted to replicate the Nairne et al. (1997) study by using the Cowan et al. (1992) words, but also including a stronger manipulation of word length by varying the number of syllables in the short and long words....

    [...]


References
More filters

Journal ArticleDOI
TL;DR: A computerised database of psycholinguistic information is described, where semantic, syntactic, phonological and orthographic information about some or all of the 98,538 words in the database is accessible, by using a specially-written and very simple programming language.
Abstract: This paper describes a computerised database of psycholinguistic information. Semantic, syntactic, phonological and orthographic information about some or all of the 98,538 words in the database is accessible, by using a specially-written and very simple programming language. Word-association data are also included in the database. Some examples are given of the use of the database for selection of stimuli to be used in psycholinguistic experimentation or linguistic research.

2,175 citations


Journal ArticleDOI
Abstract: A number of experiments explored the hypothesis that immediate memory span is not constant, but varies with the length of the words to be recalled. Results showed: (1) Memory span is inversely related to word length across a wide range of materials; (2) When number of syllables and number of phonemes are held constant, words of short temporal duration are better recalled than words of long duration; (3) Span could be predicted on the basis of the number of words which the subject can read in approximately 2 sec; (4) When articulation is suppressed by requiring the subject to articulate an irrelevant sound, the word length effect disappears with visual presentation, but remains when presentation is auditory. The results are interpreted in terms of a phonemically-based store of limited temporal capacity, which may function as an output buffer for speech production, and as a supplement to a more central working memory system.

1,825 citations


"Word length effects are not due to ..." refers background in this paper

  • ...In the original research that established this effect, Baddeley, Thomson, and Buchanan (1975) first established a span advantage for one-syllable words over fivesyllable words....

    [...]


14 Sep 2001

921 citations


"Word length effects are not due to ..." refers methods in this paper

  • ...In each case, 20 short and 20 long words were selected from the MRC database (Coltheart, 1981) to have the same characteristics as those used in the previous set....

    [...]

  • ...These words were selected from the MRC Psycholinguistic Database....

    [...]

  • ...A total of 120 short and 120 long words were selected from the MRC Psycholinguistic Database (Coltheart, 1981)....

    [...]


Journal ArticleDOI
TL;DR: A new model of immediate serial recall is presented: the primacy model, which produces accurate simulations of the effects of word length, list length, and phonological similarity.
Abstract: A new model of immediate serial recall is presented: the primacy model. The primacy model stores order information by means of the assumption that the strength of activation of successive list items decreases across list position to form a primacy gradient. Ordered recall is supported by a repeated cycle of operations involving a noisy choice of the most active item followed by suppression of the chosen item. Word-length and list-length effects are attributed to a decay process that occurs both during input, when effective rehearsal is prevented, and during output. The phonological similarity effect is attributed to a second stage of processing at which phonological confusions occur. The primacy model produces accurate simulations of the effects of word length, list length, and phonological similarity.

715 citations


"Word length effects are not due to ..." refers background in this paper

  • ...One interesting facet of this debate is that interference effects are usually seen as a prime source of forgetting in long-term memory settings, but decay is seen as the principal means in immediate memory settings (Baddeley, 1986; Burgess & Hitch, 1996; Henson, 1998; Page & Norris, 1998)....

    [...]


Journal ArticleDOI
Abstract: Summary Three experiments were performed to determine the relationship between certain variables influencing proactive inhibition in long-term retention of lists of verbal items and the influence of these variables on short-term retention of single items. More particularly, retention of single items over 18 sec. should, if the laws of long-term retention are applied, decrease with number of previous items to which S has been exposed. In addition, amount of forgetting should be a direct joint function of number of previous items and length of the retention interval. In Exp. 1 each S was presented consonant syllables singly, with retention being measured after 3, 9, and 18 sec. Forgetting of the first item presented (T-1) was less than for the second (T-2) or third (T-3) item, but forgetting of the latter (T-2 vs. T-3) did not differ. On all three tests forgetting was directly related to length of retention interval, but no interaction was evident between number of previous items and length of retention interval. In Exp. 2 a higher degree of initial learning of the items was achieved. Forgetting increased directly as a function of number of previous items presented. The predicted interaction was indeterminate since retention was essentially 100% on T-1 for all retention intervals. Experiment 3 tested retention of six successive items over 3- and 18-sec. intervals. Retention after 3 sec. showed an initial drop and then a rise over the six tests, the rise suggesting a practice effect. Forgetting over 18 sec. increased directly from T-1 to T-6 and there was no indication that a constant amount of proactive interference had been reached. The interaction between length of retention interval and number of potential proactively interfering items was very evident. The results were interpreted to mean that proactive inhibition in short-term memory of single items follows the same laws as proactive inhibition in long-term memory of lists of items.

705 citations


"Word length effects are not due to ..." refers background in this paper

  • ...This is the pattern that Keppel and Underwood (1962) found and is typical of performance on the build-up trials in the release from PI paradigm....

    [...]


Frequently Asked Questions (1)
Q1. What are the contributions in "Word length effects are not due to proactive interference" ?

The authors report two experiments in which the relationship is directly tested. In the first experiment the authors show that word length effects can be observed over the first few trials in an experiment and that the effect shows itself primarily in the number of omissions made. Melton ( 1963 ) had demonstrated that items within a study list produced mutual interference on each other. Neath and Nairne incorporated these ideas into the feature model ( Nairne 1990 ) and were able to provide good fits of existing data. However, these explanations were ruled out in the second experiment where subjects studied 24 trials of either short or long words. In the following experiments the authors attempt to replicate and extend the Nairne et al. research. In the first experiment the authors attempt a close replication of Nairne et al. save that they use a standard presentation rate of one word per second and word length is manipulated within subject rather than between subject. Here the authors explicitly examine PI effects and their relationship to word length. Each subject studied three blocks of short words and three blocks of long-words, with each block containing four trials. As was the case in the Nairne et al. experiment, all subjects studied the same set of trials, in the same order. The participants studied the six sets of trials across two days. Consequently, a between subjects design could be used if the different materials could be studied with a long period between the different sessions. The five words were then projected onto a screen at the rate of one word per second. The authors also compared performance on the very first trial in each group. The results of the first experiment show that word length effects are observable across the first four trials in an experiment provided that a strong manipulation of word length is used. They are instructed to study each block of words as it is presented on the computer screen but they are asked to remember only the most recent block of four words ; either the first block on a one-block trial or the Word Length and PI 13 second block on a two-block trial. Secondly, and more importantly, recall of words from the first block ( can, window, tile or switch being recalled ) or words from previous trials provides direct evidence for the influence of prior memories. None had participated in Experiment 1. Materials Word Length and PI 14 Subjects again studied two sets of 50 trials in which 10 were one-block trials and 40 were two-block trials. Each subject studied a set of lists that had been constructed from the open word pools and one that had been constructed from the closed word pools. The trials in the open pool condition were created in the following way. Using three and seven phoneme words the authors replicate Experiment 1 by showing that short words are better recalled in position than long words. The results of this experiment are consistent with other research that indicates that immediate recall of sub-span lists is immune to the effects of PI ( Dempster & Cooney, 1986 ; Sanders & Willemsen, 1975 ; Tehan & Humphreys, 1995 ; Wickens, Moody & Dow, 1981 ). If the first experiment is conceptualised as a single experimental session consisting of 24 trials in which 4-trial blocks of short and long words alternated, then word length effects in the latter four blocks could be explained in terms of the build up of PI across those trials as was the case in the Nairne et al study. The authors conducted a post-hoc test of this position by presenting 36 students with a single trial of either five three-phoneme words or five seven-phoneme words ( the first trials of the closed pool condition in Experiment 1 ). Cowan et al ( 1992, see also Dosher & Ma, 1998 ) have suggested that word length effects occur during the recall process itself rather than during rehearsal prior to recall. Neath and Nairne ( 1995 ) suggested that word length effects might result from intra unit interference. They suggested that words had to be compiled from sets of smaller features and that the more features that had to be compiled, the greater the likelihood that an error would be made. Given this correspondence and the fact that the Keppel and Underwood data are generally attributed to PI, they suggested that word length effects could be related to PI. There are also a number of methodological issues that require further examination before the role of PI in the word length effect can be unequivocally accepted. Furthermore, word length effects can be observed on the first trial in a block and across the subsequent three trials. Furthermore, there is no Word Length and PI 18 difference between short and long words when it comes to intrusions. This pattern suggests that the emergence of a word length effect after four trials in the Nairne et al. Furthermore, one would have to assume that this process was more likely to occur with long words than with short words.