A validation study was undertaken to determine whether or not the Branch Count, as restricted by Peter Menzel, is a valid measure of syntactic length and complexity in written language. In answer to the demand for quantitative measures of oral and written language acquisition, Menzel worked out a set of guidelines for drawing tree diagrams over the surface structure of syntactic units and then counting branches to determine both length and complexity of the units. / The Branch Count was validated by correlating its results with those of two established means of studying written language--Kellogg Hunt's words per T-unit count and John Bormuth's cloze measures--when these instruments were used to analyze thirty-two written passages. Moreover, the Branch Count results were also compared with a lesser known scale used by Walter Loban to study elaboration in communication units. / The analyses and the statistical treatment carried out to compare the Branch Count with the other measures of syntactic length and complexity indicate the following: (1) There is a high degree of correlation (.991) between the mean branch count per T-unit and the mean word count per T-unit in the passages used. Regression equations based on the results of analyzing these samples make it possible to convert word per T-unit counts to branch counts per T-unit and vice versa. (2) A high degree of correlation (.970) also exists between the mean branch count per T-unit and the mean number of elaboration index points per T-unit for the passages. Again regression equations allow for conversion of one count to the other. (3) The branch count per T-unit, analyzed as a function of both word count per T-unit and elaboration index points per T-unit, shows a positive correlation of .992 with these other measures in multiple correlation. When the mean branch count per T-unit is treated as a function of the sum of the other two counts, the coefficient of simple correlation is .989. (4) The Branch Count comparison with Bormuth's various cloze measures shows that: (a) The branch count per T-unit has a negative correlation of .852 with cloze means established by administering cloze tests on the thirty-two passages. A negative correlation is to be expected since the more syntactically complex the structure, the lower the expected cloze mean whereas the branch count is expected to rise with increased complexity. Regression equations again allow for converting one measure into an estimate of the other. (b) The correlation coefficient for the relationship between the branch counts per T-unit and the formula estimated cloze means is -.831. Regressions are given for this relationship, too. (c) The branch count per sentence treated as a function of both letters per sentence and / letters per word has a high multiple correlation with these measures. The coefficient of multiple correlation is a positive .988. Regression equations make it possible to convert a branch count per sentence into an estimate of letters per sentence and letters per word so that a researcher can use these estimates with Bormuth's table based on cloze means to estimate reading levels for passages. / These results lead to the conclusions that Menzel's Branch count is a valid instrument for measuring syntactic development and that it does measure both length and complexity of syntactic units. However, the Branch Count is judged to more effective for measuring syntactic maturity in composing written language rather than in reading written language. / Source: Dissertation Abstracts International, Volume: 41-01, Section: A, page: 0230. / Thesis (Ph.D.)--The Florida State University, 1979.
| Identifer | oai:union.ndltd.org:fsu.edu/oai:fsu.digital.flvc.org:fsu_74017 |
| Contributors | HANNON, SANDRA FURR., The Florida State University |
| Source Sets | Florida State University |
| Detected Language | English |
| Type | Text |
| Format | 279 p. |
| Rights | On campus use only. |
| Relation | Dissertation Abstracts International |
Page generated in 0.0022 seconds