The Journals of Gerontology Series B: Psychological Sciences and Social Sciences 59:P7-P10 (2004)
© 2004 The Gerontological Society of America
Sentence–Final Word Completion Norms for Young, Middle-Aged, and Older Adults
Cindy J. Lahar1,2,
Patricia A. Tun2 and
Arthur Wingfield2
1 Faculty of Liberal Studies, York County Community College, Wells, Maine.
2 Department of Psychology, and Volen National Center for Complex Systems, Brandeis University, Waltham, Massachusetts.
 |
Abstract
|
|---|
This report describes sentence–final word completion norms for 119 sentence contexts based on the original sentence completion norms of
Bloom and Fischler (1980)
. Four sets of norms are made available for 358 adults, representing young, middle-aged, young-old, and old-old samples. Notable in these norms is a high degree of consistency in responses among all four age samples. Differences in relation to the original Bloom and Fischler norms appear in responses to low contextually constraining sentences. Results show that the recency with which normative data are collected is an important variable to consider when making use of norms.
WORD associations, and word associations constrained by a linguistic context, have seen wide use in experimental studies of semantic knowledge, context effects in word recognition and memory, and studies of word generation and retrieval (Bradshaw, 1984). Many of these studies have made use of the sentence completion norms compiled by Bloom and Fischler (1980). These norms consist of two sets of 120 sentence frames followed by the relative frequency of responses given by 100 undergraduates who were asked to read each sentence and to give the first appropriate single word that came to mind to complete that sentence. This procedure produces so-called "cloze" (Taylor, 1953) norms that reflect the combined effects of the syntactic, semantic, and pragmatic constraints that govern word choice in a sentence context.
The Bloom and Fischler completion norms have been used in studies of sentence constraints on word-selection by children (Stanovich, Nathan, West, & Vala-Rossi, 1985), young adults (Masson, 1986), persons with hearing loss (Fischler, 1983, 1985), and those with Alzheimer's disease (Nebes, Boller, & Holland, 1986; Nebes & Brady, 1991). The norms have received especially wide use in cognitive aging research because of the importance of understanding age-related changes in knowledge and the utilization of linguistic context in older adults (Miller, 2003). For example, Hartman and Hasher (1991) used these sentence stimuli with young and older adults to support the idea of weakened inhibitory processes in later adulthood (see also Stine & Wingfield, 1994).
There are some potential concerns with the use of these norms. One is the question of whether sentence completion responses taken from undergraduate participants are an appropriate baseline against which to test special populations such as healthy older adults or Alzheimer's patients. The second concern comes from the age of the Bloom and Fischler norms. It is possible that changes in verbal style and social environment could influence the probability of particular responses over time. Arcuri, Rabe-Hesketh, Morris, and McGuire (2001), examining a British sample, have also raised the question of cultural influences in sentence completions. Thus, a host of potential influences such as age, time of measurement, and cultural differences may color norms collected from any one sample at any one time.
The possibility that young adults' typical sentence completions might differ from that of older adults has not caused great concern in the literature. In part, this is because general linguistic knowledge is one of the best preserved faculties on into old age (Light, 1990). There is also ample work in the literature showing age consistency in young and older adults' free associations (Burke & Peters, 1986; Lovelace & Cooley, 1982) and category structure (Howard, 1980); and in a sentence completion task, Cohen and Faulkner (1983) found no interaction between age and use of contextual constraint. These studies, however, do not negate the possibility that subtle differences might still be present (Bowles, Williams, & Poon, 1983; Gomez, 2002; Perlmutter, 1979; Riegel & Riegel, 1964).
The primary goal of this present work is to make available for general research use a set of sentence completion norms for several age groups ranging from young adults through older adulthood. We examine whether these different age groups show similar profiles in their response selections, and whether any age differences might be smaller for high-constraint than for low-constraint sentences.
|
Methods
|
|---|
Participants
The 358 native English speakers who participated were separated into four age groups (see Table 1). The samples from Calgary, Alberta, included 86 young, 86 middle, 39 young-old, and 10 old-old adults. From metropolitan Boston, Massachusetts, 71 young-old and 66 old-old adults participated. There were no significant differences between the Alberta and Massachusetts samples in any of the background variables.
The old-old group had significantly more years of education than the other three groups, F(3, 335) = 8.21, MSE = 6.36, p <.001, with Tukey pairwise comparisons (p <.05) showing that the old-old group was the only group that differed from the other groups. No significant group differences were found in measures of vocabulary (mean standard scores were -.19, -.02,.09, and.07 for the young, middle, young-old, and old-old samples, respectively).
Materials and Procedure
Sentence contexts were taken from Bloom and Fischler (1980; Appendix A) who selected the frames to produce a wide range of probabilities of the primary responses across the different sentences. We eliminated one of the original frames because of cultural datedness (In the park the hippie touched the _________).
Participants were given a five-page booklet with instructions identical to those used by Bloom and Fischler (1980). The instructions asked participants to write a single word at the end of each sentence context in order to complete that sentence with a likely ending. Two versions of the booklets were used, presenting the 119 sentence contexts in two randomized orders.
|
Results and Discussion
|
|---|
Sentence Completion Norms for the Four Age Groups
The probability of each response to each sentence frame was calculated separately for each of the four age groups. In our norms, for each sentence frame, we provide all response words given by more than one participant. Each of these words is followed by its response probability for that age group (the number of participants giving that response divided by the total number of participants responding). A final entry of "other" is given for each sentence frame, showing the sum of the probabilities of all responses that were given by only one person.
In their original publication, Bloom and Fischler (1980) presented the sentences in order from highest to lowest constraint. We have retained this same order in our norms, because our data showed a high correspondence with their norms (r =.82 to.88 across the four age groups) for the primary (dominant) response probabilities across the 119 sentence contexts.
In Table 2, we show three examples of the data obtained, one each from a high-, medium-, and low-context sentence frame. For the purposes of this presentation and following analyses, we defined as high constraint Bloom and Fischler's first 25 sentence frames. Their last 25 sentence frames we defined as low constraint, and the remaining frames were defined as medium constraint. Following each response, we show its response probabilities for each of the four age groups sampled. The final column in Table 2 shows the response probabilities for the young adults in Bloom and Fischler's (1980) norms for comparison purposes. The full set of sentence completion norms for all 119 sentence frames is available electronically in html and PDF formats at http://www.yccc.edu/faculty/lahar/norms.htm or by contacting the authors.
Correlational Analyses of Sentence Completions
As a general rule, one would expect to find higher agreement among different participant groups for sentence frames where the context strongly predetermines the response. That is, it is in the lower context sentence frames where the potential differences may primarily be expected to lie. Therefore, Table 3 is based on probabilities of the most common responses for the 25 lowest constraining sentence frames. The upper portion of the table shows correlations between each of our four age groups in the present study. These coefficients represent similarities between the groups in the degree of constraint of the sentences. It is apparent from the top portion of Table 3 that there is a relatively high correspondence in the correlation matrix representing relations among our four age groups, with each coefficient statistically significant at or beyond the p <.02 level. Interestingly, however, one can also see a trend toward a gradation in the strength of the correlations, with groups that are closer in age showing greater similarity.
In the bottom portion of Table 3, we show the correlations between each of the four groups in the present study and the 1980 norms. In contrast to the similarity between the age groups in the present study, the correlations with the 1980 norms are notably low and nonsignificant. This pattern of relationships suggests a time of measurement effect.
As we previously indicated, one would expect fewer differences for high-context sentence frames where probability of responses is more constrained. In all cases for the high-contextual constraint sentences (defined as the first 25 sentences listed in the norms), the specific word that was the dominant response choice in 1980 for college undergraduates was also the same dominant response that each of our four groups generated. The Pearson correlations between the proportion of participants in each group who gave the same specific dominant response in this study and in the 1980 norms ranged from r =.53 to r =.84, with the correlations among our four age groups ranging from r =.47 to r =.82. All of these correlations are significant at or beyond the p <.02 level. It appears, then, that when completing a sentence with a highly constraining context, there is a great deal of similarity in final word choice among adults of different ages and across different times of measurement.
Cultural Influences on Word Choice
The primary focus of this present work was not on cultural differences between norms from multiple locations. Rather, by combining responses from both Americans and Canadians, we hoped to increase the suitability of our norms to a wider circle of research settings. Interestingly, in examining responses from our two locations, we found only one systematic difference in word choice when we separated the Canadian sample from the American. This was in response to the sentence frame The hunter shot and killed a large ________. In Western Canada, the prevalence of elk is high; in fact, only Canadian participants selected the word elk to complete that sentence.
However, Arcuri and her colleagues (Arcuri et al., 2001) have recognized that diverse locations and cultures may be significant factors to consider when using norms in research with different populations. They have reported norms on the Bloom and Fischler sentence frames for an inner-city British sample in order to provide norms appropriate for the United Kingdom. They have also recently collected norms in Portuguese from a Brazilian sample (S. M. Arcuri, personal communication, June 18, 2002). Although the responses from the Canadian and American samples in the present study were very similar, we share the caution that cultural influence could potentially color the norms if sampling occurs with a different population than the population for which the norms are later applied.
Conclusions
Our primary goal has been to make available to investigators a set of sentence–final word completion norms from respondents in four age groups across the adult life span. The high-constraint sentence contexts yielded similar dominant responses across all age groups as well as in relation to the 1980 norms of Bloom and Fischler. Low-constraint sentence contexts also yielded similar probabilities of dominant responses across our four age groups, but not in relation to the norms culled in 1980. For these low-constraint sentence frames, the intervening years between gathering these norms appears to have more of an effect than do cohort or age differences.
Numerous aging researchers have relied on norms collected many years ago from young adults without knowing whether it is appropriate to extrapolate these norms to older adults. As we have shown, in cases of highly constraining sentence contexts, there is strong similarity in the sentence completions of adults of different ages. Even in cases where little context is provided by the sentence frame, the age of the sample completing the sentences does not appear as a strong factor in driving any differences in word selections. The same cannot be said for time of testing, where, at least for low-constraint sentence frames, the time at which the norms were collected appears to have a more important effect on response selections than age differences among a contemporary sample.
|
Acknowledgments
|
|---|
This research was partially supported by University of Calgary Research Grant 69-5683 to the first author and National Institutes of Health Grant AG04517 from the National Institute on Aging to the third author. We also gratefully acknowledge support from the W. M. Keck Foundation. We thank Merri Rosen, Suzanne Toth, Carolyn Crow, Michelle Newman, Maya Rahav, Anthony J. Velez, and Ezekiel Weis for help in data collection and coding. We also express appreciation to I. Fischler, M. Masson, and two anonymous reviewers for critical review and comments on earlier versions of this article. Portions of these data were presented at the International Conference on Communication, Ageing and Health in Queensland, Australia.
Address correspondence to Cindy J. Lahar, York County Community College, 112 College Drive, Wells, ME 04090. E-mail: clahar{at}yccc.edu, tun{at}brandeis.edu, or wingfield{at}brandeis.edu
|
Footnotes
|
|---|
Decision Editor: George Rebok, PhD
Received for publication July 1, 2002.
Accepted for publication August 12, 2003.
|
References
|
|---|
- Arcuri, S. M., Rabe-Hesketh, S., Morris, R. G., McGuire, P. K. 2001. Regional variation of cloze probabilities for sentence contexts. Behavior Research Methods, Instruments & Computers, 33,80-82.[Medline]
- Bloom, P. A., Fischler, I. 1980. Completion norms for 329 sentence contexts. Memory and Cognition, 8,631-642.[Medline]
- Bowles, N. L., Williams, D., Poon, L. W. 1983. On the use of word association norms in aging research. Experimental Aging Research, 9,175-177.[Medline]
- Bradshaw, J. L. 1984. A guide to norms, ratings and lists. Memory and Cognition, 12,202-206.[Medline]
- Burke, D. M., Peters, L. 1986. Word associations in old age: Evidence for consistency in semantic encoding during adulthood. Psychology and Aging, 1,282-292.
- Cohen, G., Faulkner, D. 1983. Word recognition: Age differences in contextual facilitory effects. British Journal of Psychology, 74,239-251.
- Fischler, I. 1983. Contextual constraint and comprehension of written sentences by deaf college students. American Annals of the Deaf, 128,418-424.[Medline]
- Fischler, I. 1985. Word recognition, use of context, and reading skill among deaf college students. Reading Research Quarterly, 20,203-218.
- Gomez, R. 2002. Word frequency effects in priming performance in young and older adults. Journal of Gerontology: Psychological Sciences, 57B,P233-P240.
- Hartman, M., Hasher, L. 1991. Aging and suppression: Memory for previously relevant information. Psychology and Aging, 6,587-594.[Medline]
- Howard, D. V. 1980. Category norms: A comparison of the Battig and Montague (1969) norms with the responses of adults between the ages of 20 and 80. Journal of Gerontology, 35,225-231.[Abstract/Free Full Text]
- Light, L. L. 1990. Interactions between memory and language in old age. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the psychology of aging (3rd ed., pp. 275–290). San Diego, CA: Academic Press.
- Lovelace, E. A., Cooley, S. 1982. Free associations of older adults to single words and conceptually related word triads. Journal of Gerontology, 37,432-437.[Abstract/Free Full Text]
- Masson, M. E. 1986. Comprehension of rapidly presented sentences: The mind is quicker than the eye. Journal of Memory and Language, 25,588-604.
- Miller, L. S. 2003. The effects of age and domain knowledge on text processing. Journal of Gerontology: Psychological Sciences, 58B,P217-P223.
- Nebes, R. D., Boller, F., Holland, A. 1986. Use of semantic context by patients with Alzheimer's disease. Psychology and Aging, 1,261-269.[Medline]
- Nebes, R. D., Brady, C. B. 1991. The effect of contextual constraint on semantic judgments by Alzheimer patients. Cortex, 27,237-246.[Medline]
- Perlmutter, M. 1979. Age differences in the consistency of adults' associative responses. Experimental Aging Research, 5,549-553.[Medline]
- Riegel, K. F., Riegel, R. M. 1964. Changes in associative behavior during later years of life: A cross-sectional analysis. Vita Humana, 7,1-32.
- Stanovich, K. E., Nathan, R. G., West, R. F., Vala-Rossi, M. 1985. Children's word recognition in context: Spreading activation, expectancy, and modularity. Child Development, 56,1418-1428.
- Stine, E. A. L., Wingfield, A. 1994. Older adults can inhibit high-probability competitors in speech recognition. Aging and Cognition, 1,152-157.
- Taylor, W. L. 1953. "Cloze" procedure: A new tool for measuring readability. Journalism Quarterly, 30,415.
TOP
Abstract
Methods
