The investigation tested the effectiveness of machine instruction in the teaching of spelling to a sample of fifty-four secondand third-grade students. All subjects were located in the same school, and groups were formed on the basis of spelling achievement matching. The analy sis of variance statistical technique was employed in the treatment of the data. Although no differences were found in main effects (i.e., machine versus teacher), a differ ential (Interaction) effect was found at the .002 level of confidence. Further analysis revealed that: (1) on the third-grade level, the machine had proved more effective (.01 level of confidence) than the traditional method, and (2) on the second-grade level the traditional method had produced significantly more achievement (above the .01 level) than the machine method. The strength of the reversal is strongly suggestive of an age level below which this type of machine instruction loses its effectiveness. EDUCATORS, learning theorists, and laymen have greeted the appearance of the teaching ma chine in the nation's schools with varying degrees of enthusiasm. In some quarters the machines receive almost unqualified support; in others, they are as cordially condemned. One of the more commonly heard objections to the use of machine instruction is that the ma chine forces learning to proceed only in small steps and requires highly uniform performance. As spelling is a school content subject that re quires procession in small steps through memori zation and repetition, it appears that this subject may be uniquely suited for machine instruction. In addition to the fact that teaching machines incorporate attributes believed to be necessary in the teaching of spelling, research findings re lated to spelling in recent years indicate a definite need for improved methodology in this sub ject area. Experimental research involving machine instruction with spelling content would seem to represent a logical and promising line of investigation. The presently reported project was begun with the investigator's development of a detailed se quence of spelling lessons for machine use in grades two through six. The experiment was con ducted as a test of the effectiveness of segments of the machine program appropriate for use by second and third graders. The Spelling Program The spelling program developed for use in the present study was designed to teach the student to spell words on paper after hearing them pro nounced. The 1,998 words included in the pro gram were taken from those listed for grades two through six in Improving Your Spelling Pro gram.1 The words were re-organized into groups that were similar in spelling or pronunciation and were ordered within groups according to complexity. Only noun plurals were omitted in the construction of the program. Part I of the program (1,002 words) includes one-syllable words only; and Part II (996 words) includes two, three, and four syllable words. Each part is divided into four units. Unit I in each part con tains words that are spelled as they sound. Unit II of both parts contains words with double conso nants. Unit III of both parts contains words with double vowels, and Unit IV contains words that are not spelled as they sound. In all of the units the words are in groups of five. Isofar as possible, words of similar sound and spelling are grouped together. In preparing the program, the investigator ob served the fundamentals of programmed instruc tion but made no attempt to adhere to any par ticular programming doctrine. The resulting pro gram is classified as linear in that only one pre determined series of responses is required of the student. Reinforcement is provided with each written response, and the possibilities of error have been minimized insofar as is consistent with the fading of prompts. The program was prepared for use with a mag This content downloaded from 157.55.39.46 on Sat, 11 Jun 2016 07:07:32 UTC All use subject to http://about.jstor.org/terms