Single-Letter Retrieval Cues for Anagram Solution

Single-Letter Retrieval Cues for Anagram Solution – Statistical Data Included

Kenneth L. Witte

ABSTRACT. Anagram solution, as related to single-letter retrieval cues and first letter of the solution word (consonant or vowel), was examined. In Experiment 1, college-aged solvers were presented both types of 5-letter words and either the first letter of the solution word as a cue, or no cue. In Experiment 2, the effects of four types of retrieval cues (first, middle, or last letter or no cue) upon solving consonant-beginning words was examined. Finally, Experiment 3 examined the solution of both types of solution words as related to the preceding four types of retrieval cues. The results of all 3 experiments showed that a single letter can be an effective cue. For consonant-beginning words, the middle and last letters were as effective as the first letter. For vowel-beginning words, the first letter was more effective than either the middle or last letter. It was concluded that solvers select one letter of the anagram, typically a consonant, to serve as the first letter of the solution word, and then rear range the remaining letters.

Key words: anagram, retrieval cues, single-letter

RESEARCH in the area of anagram problem solving has a long history (see Johnson, 1966). Because anagrams provide a well-defined task in which performance can be carefully measured (Richardson & Johnson, 1980), they have been used to study linguistic variables, a variety of individual difference variables, and general cognitive processes (Weinstock, 1979). Early research was centered primarily on the anagram task itself, and studies focused on anagram solution as it related to characteristics of the anagrams and of the solution words, as well as individual difference variables (see Johnson, 1966, for a review). More recently, researchers have used anagrams to examine a variety of phenomena, including many involving cognitive processes. For example, anagrams have been used to study implicit memory in younger adults (Rajaram & Roediger, 1993) and in older adults (Java, 1992), as well as issues involving the discreteness or continuity of cognitive processes (Smith & Kounios, 1996). In addition, individual differe nce variables in cognitive functioning have been examined. This research has involved schizophrenic individuals (Miller, Chapman, Chapman, & Collins, 1995), home-dwelling elderly individuals (Witte & Freund, 1995), and institutionalized elderly individuals (Phillips, 1993).

Given the widespread use of anagrams in current cognitive research and the fact that the anagram task is seen by some (viz., McAndrews & Moscovitch, 1990) as particularly appropriate for addressing questions about repetition of problem-solving operations, a more thorough understanding of the processes involved in anagram solution would seem to be warranted.

The processes of anagram solution typically involve a rearrangement of presented letters. Solvers must compare the rearrangement with words searched for and retrieved from their lexical memories (Mendelsohn, 1976). Investigators have been concerned with three questions: (a) What are the features of the words stored in memory? (b) In what manner are the words retrieved? (c) How does the comparison between the words and the anagrams occur? (see LeMay, 1972). Typically, priming procedures have been used to address these questions (see, e.g., Antonietti & Girotti, 1991). The present study was concerned with the first process of anagram solution, namely, the rearrangement of the letters of the anagrams.

Some theorists (e.g., Solso, Topper, & Macey, 1973) have hypothesized that individuals often select a single letter (or two) as the beginning of the solution word and then rearrange the remaining letters, searching for a combination that matches a word in their lexicon. Solvers’ self-reports confirm this hypothesis (Kaplan & Carvellas, 1968; Mendelsohn, 1976). Given this approach to anagram solution, providing the first letter of the solution word should facilitate anagram solution. Three previous studies have addressed this issue.

Using 5-letter solution words, Dewing and Hetherington (1974) found that performance was facilitated if the first and last letter of the solution words were simultaneously presented. Murray and Mastronardi (1975), using solution words ranging from 3 to 9 letters, presented just the first letter of the solution word. They did not present any inferential statistics to compare their various cue groups, because their goal was to determine whether a power function could describe their data. Nonetheless, their Figure 1 reveals that the first letter apparently was an effective cue for the longer anagrams (7 or more letters), but not for the shorter anagrams. Finally, Dominowski (1968) found that a single letter, regardless of its position in the solution word, did not facilitate solutions for 5-letter anagrams.

It is surprising, for at least two reasons, that the effects of presenting the initial letter of the solution word are not more robust. First, as indicated earlier, solvers report that they often select a single letter as the beginning of the solution word. Second, Brown and Knight (1990) found that the first letter was a particularly effective aid for retrieving target words based on their definitions. The purpose of the present research was to further examine the role of single-letter cues for anagram solution. The three studies reported herein used 5-letter anagrams, because this length is most commonly used in anagram research. The findings cited earlier notwithstanding, we hypothesized, for reasons to be expounded shortly, that providing a single letter as a cue would facilitate anagram solution.

EXPERIMENT 1

If, as according to Solso et al. (1973), solvers select a single letter as the beginning of the solution word, then they are more likely to choose a consonant, because more words start with consonants than with vowels (Mendelsohn, 1976). Accordingly, anagrams whose solution words start with consonants should be easier to solve than those starting with vowels. Research supports this prediction (Gilhooly & Johnson, 1978; Hunter, 1959; Mendelsohn, 1976). Given these findings, it might be expected that a first-letter cue would be more helpful for solution words starting with vowels, because this information would keep solvers from needlessly spending time looking for words starting with consonants. It is worth noting those studies (Dominowski, 1968; Murray & Mastronardi, 1975) in which there was no facilitatory effect of providing the first letter of the solution words employed solution words that started with consonants. Perhaps facilitatory effects would have been demonstrated if words beginning with vowels, r ather than consonants, had been used. In the present study, participants solved anagrams whose solutions started either with a vowel or with a consonant. Half of the participants were given the first letter of the solution word as a cue and the other half were not. It was predicted that the anagram performance of the cue group would exceed that of the no cue group, particularly for those solution words starting with a vowel.

Method

Design and Participants

The design was a 2 x 2 factorial, with absence or presence of the first letter of the solution word as a between-subjects variable, and beginning letter (consonant or vowel) of the solution word as a within-subject variable. The participants were 48 general psychology students whose participation partially fulfilled a course requirement. Participants were block randomly assigned to the two cue conditions.

Materials and Procedure

Twenty anagrams were used. Half of the solution words started with vowels (two of each vowel), and the other half started with consonants (two each of five randomly selected consonants). The vowel words were album, apron, enact, entry, inlet, impel, orbit, orang, ultra, and usher, and the consonant words were demon, dogma, haven, horde, neigh, nudge, token, tulip, waver, and woven. All solution words were composed of three different consonants and two different vowels. Each solution word had a frequency of occurrence ranging from 1 to 17 per million (Thorndike & Lorge, 1944). The average frequency was 7.5 per million for the vowel words and 7.9 per million for the consonant words. All anagrams involved three letter moves to reach the solution word; a different order was used with each anagram. The order of the two types of solution words was initially block randomly ordered and then the order of the 20 anagrams was randomly determined, with the restriction that a word beginning with a given letter could not occur a second time until all other beginning letters had occurred once. The resulting order of the 20 anagrams was the same for all participants.

A 22-page test booklet was prepared for each participant. The first page was an informed consent form and the second page contained the instructions. The anagrams were presented, one to a page, on the remaining 20 pages. Each anagram was typed in capital letters in exactly the same location on each page. Under each anagram were two lines, one for recording the solution word and one for recording the solution time. Half of the test booklets had the first letter of the solution word on the solution line, whereas the other half did not.

Participants were tested in small groups. After signing the informed consent form, participants read the instructions. They were told that they had 60 s to solve each anagram; solution times were observed using a stop clock on the front wall of the testing room. Timing was controlled by the experimenter, who told the participants when to start, stop, and turn the page to the next anagram. Participants were told that the solution words did not include foreign words, plurals, or proper names, and that they could use paper and pencils as aids. For participants in the cue condition, the last line of the instructions informed them that the first letter of each solution word would be provided.

Results

A 2 x 2 factorial analysis of variance (ANOVA) was conducted for each of two dependent variables (number of anagrams solved and median solution time). Participants who did not solve any of a particular type of anagram were assigned a score of 61 s for that condition. The .05 level of significance was used for all tests in this and subsequent experiments. More words beginning with consonants were solved than those beginning with vowels F(1, 46) = 35.97 (see Table 1). Further, participants presented with the first letter of the solution word solved more anagrams than those not presented with a cue, F(l, 46) = 70.69. Finally, the interaction was significant F(1, 46) = 20.40. Although the cue condition effect was reliable, this effect was more pronounced for words beginning with vowels than for words beginning with consonants. It should be noted, however, that follow-up analyses indicated that the cue condition effect was reliable for both types of solution words. The analysis of the median time to solution prod uced results that generally paralleled the number solved results. Words starting with consonants were solved faster than words starting with vowels, F(1, 46) = 6.40 (see Table 1). Solution times in the cue condition were faster than those in the no cue condition, F(l, 46) = 11.02, and this effect was greater for vowel-beginning words than for consonant-beginning words, F( 1,46) = 10.14. Follow-up analyses of the interaction indicated a reliable cue condition effect for vowel-beginning words but not for consonant-beginning words. The intention was to use anagrams that formed single-solution words. However, orang could also be solved as either argon (word frequency = 5 per 4 million), groan (34 per million), or organ (48 per million), and neigh could be solved as hinge (10 per million). In the previously reported analyses, any of the preceding words were counted as correct. With the exception of organ, these multiple solutions could be attained only by participants in the no cue condition and thus, could potent ially minimize the effect of providing the first letter as a cue. Although the cue condition effect was reliable for both number correct and time to solution, we decided to reanalyze the data, excluding the two multiple-solution anagrams. The basic pattern of results in these analyses mirrored that found earlier; for both dependent measures, all three effects were significant at the same levels as those previously reported.

Discussion

The present results unequivocally demonstrate that graphemic cues, in the form of the first letter of the solution word, can be effective cues for anagram solution. As such, the findings are consistent (a) with the theoretical postulation (Solso et al., 1973) that solvers select a single letter of the anagram as the beginning letter of the solution word and (b) with previous findings suggesting that this letter is more likely to be a consonant than a vowel (Gilhooly & Johnson, 1978; Mendelsohn, 1976). Knowing that the first letter is a vowel keeps solvers from pursuing a solution word beginning with a consonant. On the other hand, knowing that the first letter is a consonant, although still helpful, merely informs the solver which of the three consonants to put at the beginning of the solution word.

The present results are inconsistent those of with previous studies (Dominowski, 1968; Murray & Mastronardi, 1975) that failed to find cuing effects for 5-letter solution words. However, closer inspection of these two studies suggests several reasons for the discrepant findings. First, both studies involved words that began with consonants. As indicated by the present findings, cuing effects are generally smaller for such words than for words beginning with vowels. Second, Murray and Mastronardi (1975) used solution time as their dependent measure. The significant interaction found for this measure in the present study was due to the fact that cuing effects were reliable only for solution words starting with vowels. Perhaps if these authors had also analyzed number of correct solutions, then they would have demonstrated reliable cuing effects. Thirdly, Dominowski (1968) did find more correct solutions for his cue group (M = 11.00; maximum possible = 36) than for his control group (M = 8.75). This difference is of a magnitude comparable to that found in the present study. Further, Dominowski found that the cue group improved from the first block of 12 anagrams to the last two blocks, whereas the control group did not. His Figure 2 reveals that the two groups showed comparable performance initially, although the number of correct solutions was greater for the cue group than for the control group across the last two blocks of 12 anagrams. It is possible that a follow-up analysis might have revealed a reliable cuing effect for the latter two blocks. Dominowski interpreted the improvement in performance across blocks for the cue group as probably reflecting solvers’ learning to use the letter-sequence information provided. In the present study comparable practice effects did not occur. An analysis of the number of correct solutions for the first 10 versus the second 10 anagrams revealed that although the same pattern of results occurred for both halves of the task, there was a slight decrease in performance for the c ue group (Ms = 7.62 and 7.12, respectively) across the two blocks.

EXPERIMENT 2

Ronning (1965) proposed that anagram solvers approximate an algorithmic process of permuting the letters of an anagram and that they can “rule out” certain permutations because the initial letter, bigram, or trigram is unlikely to begin an English word. Thus, anagrams having a high rule-out factor (more permutations eliminated) should be easier than anagrams having a low rule-out factor. Ronning’s results were consistent with this prediction, and Gribben (1970) subsequently replicated and extended these findings.

Dominowski (1968) indicated that if the position of one of the letters in a 5-letter word is indicated, only 24 permutations remain, regardless of the letter’s position in the word. Thus, the effectiveness of a single letter as a cue should be independent of its position in the solution word. This analysis is inconsistent with accounts (e.g., Solso et al., 1973) that suggest that the first letter should be the most effective cue inasmuch as solvers first attempt to determine the initial letter of the solution word. Dominowski’s results were consistent with his position, as the effect for letter position was not significant. However, Dominowski failed to find a significant cuing effect; the cue group did no better overall than the control group did. Furthermore, Dominowski manipulated letter position using a within-subject design. It is possible that this approach induced a different problem-solving strategy than a between-subjects design would have. For further assessment of the effectiveness of a single let ter as a cue for anagram solution, in the present study, four groups of solvers were presented the anagrams used by Ronning (1965). Three of the groups received a single letter of the solution word (first, middle, or last) as a cue, whereas the fourth group received no cue. On the basis of the results of the first study, we expected that the anagram performance of solvers presented the first letter would exceed that of the control group. Of particular interest in this study was the question of whether the middle and last letters would be as effective as cues as the first letter.

Method

Design and Participants

The design was a 2 x 2 x 4 factorial design, with rule-out factor (high, low) and Thorndike-Lorge word frequency (high, low) as within-subject variables, and cuing condition (no cue, first, middle, or last letter of solution word) as a between-subjects variable. The participants were 80 general psychology students whose participation partially fulfilled a course requirement. Participants were block randomly assigned to the four cue conditions.

Materials and Procedure

Ronning’s (1965) 20 solution words were used. Half of the solution words were high frequency (AA) words (light, month, stand, child, plant, water, laugh, white, price, their), as determined by the Thorndike and Lorge (1944) counts. The other half were low frequency (7 to 8 per million) words (wring, flint, scant, cramp, blunt, chaos, pouch, deity, scour, broil). The first 6 words from each frequency category entailed high rule-out totals (90-94 of the 120 permutations could be eliminated by logical analysis), whereas the remaining half involved low rule-out totals (70-76). The order of the letters in each anagram was the same as that in Ronning’s study. The order of the four types of anagrams was block randomly determined and was the same for all participants.

A 22-page booklet was prepared for each participant. The general format of the booklets was comparable to that used in Experiment 1. In one fourth of the booklets, the initial letter of each solution word appeared in the first space of each solution line. In another quarter, the middle letter was presented in the middle space, whereas in another quarter the last letter was presented in the final space. The remaining quarter contained blank solution lines.

Participants were tested in small groups. The basic procedure was identical to that of Experiment 1. For participants in the three cue conditions, the last line of the instructions informed them that the first, middle, or last letter, respectively, of each solution word would be presented.

Results

A 2 x 2 x 4 factorial ANOVA was conducted on the number of anagrams solved and the median solution time. Participants who did not solve any of a particular type of anagram were assigned a score of 61 s for that condition. Analysis of the number solved produced a reliable main effect for cuing conditions, F(3, 76) = 6.88. Participants in the three cue conditions solved more anagrams than those in the control condition (see Table 2). Newman-Keuls tests indicated that the three cue groups did not differ from one another, and all three differed from the control.

Three other effects were significant. More high frequency (M 8.01) than low frequency (M = 6.69) words were solved, F(l, 76) = 36.18, and more high rule-out (M = 8.19) than low (Ms = 6.51) rule-out words were solved, F(1, 76) = 62.85. In addition, the interaction between these two variables was reliable, F(1, 76) = 3.88. The rule-out effect was of greater magnitude for the high (Ms = 4.51 vs. 3.50) than the low frequency words (Ms = 3.68 vs. 3.01).

The analysis of the time scores produced three reliable effects. High frequency words (M = 13.48) were solved faster than low frequency words (M = 19.48), F(1, 76) = 45.06. High rule-out words (M= 13.20) were solved faster than low rule-out words (M = 19.76), F(l, 76) = 31.33. The Frequency x Rule-out interaction was also significant, F(l, 76) = 10.95. The rule-out effect was again of greater magnitude for high frequency (Ms = 8.89 vs. 18.07) than for low frequency words (Ms = 17.51 vs. 21.45).

Discussion

The present results, like those of Experiment 1, clearly demonstrate that a single letter can be an effective cue for anagram solution, in terms of number of correct solutions. Given the fact that all the solution words started with consonants, the present failure to find cuing effects for the solution time measure is comparable to the findings of Experiment 1, in which cuing effects were demonstrated for vowel-beginning words but not for consonant-beginning words. Of special import is the fact that cuing effects were independent of the position of the cue letter in the solution word. This finding is consistent with what Dominowski (1968) found, with the major exception that cuing effects were significant in the present study. The first letter of the solution word was not a more effective cue than either the middle or last letter, which is consistent with Dominowski’s argument that the effectiveness of a single letter as a cue should be independent of its position in the solution word, because the number of r emaining permutations of the letters is identical in each case. This result is seemingly inconsistent with the hypothesis that the first letter should be the most effective retrieval cue. However, it is possible that, in some cases, the first letter might be the most effective cue. As noted earlier, solvers tend to seek a solution word that starts with a consonant. This tendency would have been reinforced in the present study because of all the solution words started with consonants. It may be the case that, with the present sample of solution words, revealing the middle or last letter provides the solver with a good guess as to the identity of the initial letter. For example, presenting the anagram ITLGH along with the information that G (T) is the middle (last) letter may make it apparent to the solver that the trigram of GHT constitutes the end of the word, and that, therefore, L must be the initial letter. If this analysis has merit, then, in essence, the three cuing conditions are functionally equivalent . Perhaps with other solution words, (e.g., words starting with vowels, or words having more than 5 letters), the initial letter may be a better cue than either the middle or last letter.

Two additional comments are appropriate regarding the present results. Anagrams with high rule-out values were solved more readily than ones with low rule-out values, replicating earlier findings (Gribben, 1970; Ronning, 1965). Also, anagrams forming high frequency words were solved more readily than anagrams forming low frequency words. This finding was first demonstrated by Mayzner and Tresselt (1958) and has been repeatedly demonstrated (see, e.g., Johnson, 1966).

EXPERIMENT 3

In Experiment 2, we found that the initial letter of the solution word was no better as a cue than either the middle or last letter. However, we hypothesized that with some words, such as those beginning with a vowel, providing the initial letter would lead to more solutions than would providing either the middle or last letter. In the third experiment, we tested this hypothesis.

Method

Design and Participants

The design was a 2 X 4 factorial, with first letter (consonant or vowel) of the solution word as a within-subject variable and cuing conditions (no cue, or first, middle, or last letter of the solution word) as a between-subjects variable. The participants were 80 general psychology students whose participation partially fulfilled a course requirement. Participants were block randomly assigned to the four cue conditions.

Materials and Procedure

The 20 solution words used in Experiment 1 were used, with the exception that the multiple-solution words (orang, neigh) were replaced by ogler and nobly, respectively. All anagrams involved three letter moves to reach the solution word; a different order was used with each anagram. The order of the anagrams was block randomly determined, as in Experiment 1.

A 22-page booklet was prepared for each participant; the general format of the booklets was like that used in the previous experiments. And, as in Experiment 2, one fourth of the booklets had the initial letter of the solution word in the first space of the solution line, another quarter had the middle letter in the middle space, another quarter presented the last letter in the final space, and the remaining quarter presented just a blank solution line.

Participants were tested in small groups, and the same general procedure used in the earlier studies again was employed.

Results

A 2 X 4 factorial ANOVA was conducted on the number of anagrams solved and the median solution time. For the number solved correctly, all three effects were significant. The total number of anagrams solved differed across the four cue conditions, F(3, 76) = 12.86, more consonant-beginning words were solved than vowel-beginning words, F(l, 76) = 178.79, and the cue effect differed depending on the type of word, F(3, 76) = 5.73 (see Table 3). Follow-up tests of the interaction indicated that the cue condition effect was reliable for each type of word (p [less than] .01 in each case). Newman-Keuls tests indicated that, for vowel-beginning words, the three cue groups differed from the no cue group, and the first letter group differed from the other two cue groups, which, in turn, did not differ from one another. For consonant-beginning words, the three cue groups differed from the no cue group, but did not differ from one another.

For solution time, the four cue groups differed, F(3, 76) = 5.46, and vowel-beginning words took longer to solve than consonant-beginning words, F(1, 76) = 51.59. The interaction was also reliable, F(3, 76) = 4.18. Follow-up tests of the interaction revealed a reliable group effect for vowel-beginning words but not for consonant-beginning words, Fs(3, 76) = 6.14 and 1.88, respectively. Follow-up Newman-Keuls tests of the former effect indicated that the first letter group differed from the other three groups, which, in turn, did not differ from one another.

GENERAL DISCUSSION

The present results, in part, replicate the results of the two previous studies. As was the case in Experiment 1, the group presented with the first letter of the solution word solved more consonant- and vowel-beginning words, and solved the latter more quickly, than the group that received no cue. And, as was the case in Experiment 2 with consonant-beginning words, the three cue groups, although not differing from one another, solved more anagrams than the no cue group.

In addition to replicating the earlier results, the present findings indicate that, under some conditions, the first letter of the solution word is a better cue than either the middle or last letter. Solvers provided with the first letter solved more vowel-beginning words and solved them faster than solvers in the other groups. As such, these findings are consistent with both solvers’ reports (Kaplan & Carvellas, 1968; Mendelsohn, 1976) regarding the solution process, namely, that they seek the beginning letter(s) of the solution word as a cue to retrieve the solution word from lexical memory and with theoretical accounts of the process (e.g., Solso et al., 1973). As indicated earlier, solvers apparently are more likely to select a consonant as their retrieval cue. For consonant-beginning words, providing the middle or last letter facilitates solution to the same extent as providing the first letter because knowing either the middle or last letter reduces the number of possible permutations remaining, and/or provides useful information as to which consonant is the first letter of the solution word. However, the middle and last letters are not nearly as effective as the first letter for solving vowel-beginning words. Although the number of permutations is reduced to the same degree by knowing any letter (first, middle, or last), solvers who are given the middle or last letter apparently still seek solutions beginning with a consonant, at least initially. Eventually, solvers in these two conditions realize that the solution word does not start with a consonant and start to look for a solution word beginning with a vowel. Seemingly, enough time is left for solution that solvers in these two groups do better than solvers in the no cue condition, in terms of number solved, but there is insufficient time to do as well as solvers provided with the first letter. Providing the first letter as a cue not only reduces the number of permutations remaining, but it also keep solvers from spending time needlessly seeking solutio ns beginning with consonants.

In sum, the results of the three experiments are consistent with the premise (Mendelsohn, 1976) that anagram solution is, in part, a retrieval process, and that retrieval cues, in addition to the anagram itself, can aid this process. Previous research has shown that providing solvers with cues, in addition to the anagram itself, can aid the process of anagram solution. Both categorical cues (Dewing & Hetherington, 1974; Murray & Mastronardi, 1975; Schubert, Spoehr, & Haertal, 1979) and associative cues (Safren, 1962) have been shown to facilitate anagram solution. However, previous studies, using single-letter cues with shorter anagrams (Dominowski, 1968; Murray & Mastronardi, 1975), did not find such cues to facilitate the solution process. The present three experiments provide compelling evidence that a single letter can facilitate anagram solution. Furthermore, apparently due to solvers’ tendencies to seek solution words beginning with consonants rather than with vowels, the initial letter of the solution word is more effective than either the middle or last letter for solution words starting with vowels.

The research reported in this article was supported by funds from the Marie Wilson How-ells Fund.

REFERENCES

Antonietti, A., & Girotti, G. (1991). Anagram solving is facilitated by intra- but not crossmodal priming. Perceptual and Motor Skills, 72, 403-408.

Brown, A. S., & Knight, K. K. (1990). Letter cues as retrieval aids in semantic memory. American Journal of Psychology, 103. 101-113.

Dewing, K., & Hetherington, P. (1974). Anagram solving as a function of word imagery. Journal of Experimental Psychology, 102, 764-767.

Dominowski, R. L. (1968). Anagram solving as a function of letter-sequence information. Journal of Experimental Psychology, 76, 78-83.

Gilhooly, K. J., & Johnson, C. E. (1978). Effects of solution word attributes on anagram difficulty: A regression analysis. Quarterly Journal of Experimental Psychology, 30, 57-70.

Gribben, J. A. (1970). Solution-word letter sequences in anagram solving. Journal of Experimental Psychology, 85, 192-197.

Hunter, I. M. L. (1959). The solving of five-letter anagrams. British Journal of Psychology, 50, 193-206

Java, R. I. (1992). Priming and aging: Evidence of preserved memory function in an anagram solution task. The American Journal of Psychology, 105, 541-548.

Johnson, D. M. (1966). Solution of anagrams. Psychological Bulletin, 66, 371-384.

Kaplan, I. T., & Carvellas, T. (1968). Effect of word length on anagram solution time. Journal of Verbal Learning and Verbal Behavior, 7, 201-206.

LeMay, E. H. (1972). Anagram solutions as a function of task variables and solution word models. Journal of Experimental Psychology, 92, 65-68.

Mayzner, M. S., & Tresselt, M. E. (1958). Anagram solution times: A function of letter order and word frequency. Journal of Experimental Psychology, 56, 376-379.

McAndrews, M. P., & Moscovitch, M. (1990). Transfer effects in implicit tests of memory. Journal of Experimental Psychology: Learning, Memory, and cognition, 16, 772-778.

Mendelsohn, G. A. (1976). An hypothesis approach to the solution of anagrams. Memory & Cognition, 4, 637-642.

Miller, M. B., Chapman, J. P., Chapman, L. J., & Collins, J. (1995). Task difficulty and cognitive deficits in schizophrenia. Journal of Abnormal Psychology, 104, 251-258.

Murray, D. J., & Mastronardi, L. L. (1975). Anagram solution times, word length, and type of accessory clue. Bulletin of the Psychonomic Society, 5, 119-121.

Phillips, K. T. (1993). The effectiveness of cueing on anagram solving by cognitively impaired nursing home elderly. Loss, Grief and Care, 6, 107-116.

Rajaram, S., Roediger, H. L., III. (1993). Direct comparison of four implicit memory tests. Journal of Experimental Psychology: Learning, Memory and Cognition, 19, 765-776.

Richardson, J. T. E., & Johnson, P. B. (1980). Models of anagram solution. Bulletin of the Psychonomic Society, 16, 247-250.

Ronning, R. R. (1965). Anagram solution times: A function of the “ruleout” factor. Journal of Experimental Psychology, 69, 35-39.

Safren, M. A. (1962). Associations, sets, and the solution of word problems. Journal of Experimental Psychology, 64, 40-45.

Schubert, R. E., Spoehr, K. T., & Haertel, R. J. (1979). Solving anagrams: Category priming and the differential availability of category solutions. Quarterly Journal of Experimental Psychology, 31, 599–607.

Smith, R. W., & Kounios, J. (1996). Sudden insight: All-or-none processing revealed by speed-accuracy decomposition. Journal of Experimental Psychology: Learning, memory and Cognition, 22, 1443-1462.

Solso, R. L., Topper, G. E., & Macey, W. H. (1973). Anagram solution as a function of bigram versatility. Journal of Experimental Psychology, 100, 259-262.

Thorndike, E. L., & Lorge, I. (1944). A teacher’s wordbook of 30,000 words. New York: Teachers College Press, Columbia University.

Weinstock, R. B. (1979). Anagram solving as influenced by solution word frequency, anagram transition probability, and subject’s vocabulary level. Bulletin of the Psychonomic Society, 14, 375-378.

Witte, K. L., & Freund, J. S. (1995). Anagram solution as related to adult age, anagram difficulty, and experience in solving crossword puzzles. Aging and Cognition, 2, 146-155.

TABLE 1

Means and Standard Deviations for Anagram Solution Performance as

Related to First Letter of Solution Word and Cuing Condition

First letter

Vowel Consonant

Condition M SD M SD

Number solved

No cue 2.67 2.08 5.62 1.79

First letter 7.17 1.34 7.58 1.25

Median solution time (s)

No cue 27.38 16.23 17.54 7.07

First letter 13.79 6.74 14.92 8.33

TABLE 2

Means and Standard Deviations for Anagram Solution Performance

as Related to Cuing Condition

Response measure

Number solved Solution time (s)

Condition M SD M SD

No cue 12.35 3.10 15.48 4.88

First letter 15.75 2.69 16.63 8.14

Middle letter 14.35 3.65 18.76 8.17

Last letter 16.35 2.56 15.04 5.22

TABLE 3

Means and Standard Deviations for Anagram Solution Performance as

Related to First Letter of Solution Word and Cuing Condition

First letter

Vowel Consonant

Condition M SD M SD

Number Solved

No cue 1.65 1.46 5.40 1.88

First letter 5.65 1.60 7.40 1.70

Middle letter 4.15 2.80 6.85 1.95

Last letter 3.75 2.07 8.00 1.56

Median solution time (s)

No cue 33.05 20.40 21.65 9.82

First letter 18.35 8.66 15.60 8.06

Middle letter 31.45 11.71 19.55 8.58

Last letter 35.75 12.34 17.75 6.85

COPYRIGHT 2001 Heldref Publications

COPYRIGHT 2001 Gale Group