The American Heritage Intermediate Corpus

Sample studies of the five-million-word American Heritage Intermediate Corpus, the largest yet for English, show the inadequacies of computerized word counts for lexicographical purposes. Out of about 87,000 word types listed in the American Heritage Word Frequency Book (by JOHN B. C~RorI. et alii, Houghton Mifflin-American Heritage, New York, 1971), only about 40 % seem distinctive lexical items. On the other hand, 22 % of the vocabulary entries required by the 50,000-entry dictionary for which the Corpus was prepared did not occur even once among the five million word tokens. We first examined three randomly selected portions of the listing in the American Heritage Word Frequency Book (AHWFB).side by side with the corresponding entry lists of the American Heritage School Dictionary (AHSD), the Thorndike-Barnhart Intermediate Dictionary (TBID), the Merriam-Webster New Students Dictionary (WNSD), the Holt Intermediate Dictionary (HID), and the Random House School Dictionary (RHSD). Each of the three portions checked in AHWFB was based on a consolidated list of 100 entry words (300 in all), made up of the vocabulary entries in the dictionaries: (1) AFTER. to AHAB, (2) CABBAGE to CALEB, and (3) LAKE to LANOLIN. Our aim was to fred out how useful the Corpus was in assessing the vocabulary of Grades 3-9 and in adding new words to the lexicon of that level as presented in the dictionaries. Here are some of our findings: There are approximately 110 word types in each sample of the main AHWFB listings (once-occurring entries listed at bottom of pages were omitted) corresponding to each 100-entry-word portion from the dictionaries. Of the 110 word types, 36 % are lexically undistinctive items inadmissible as vocabulary entries in dictionaries. Such are: (a) hyphenated loose compounds, e.g. After-Shaving, agreedon, and air-breathing; (b) solid forms of hyphenated compounds; (c) spellings with tmdistinctive initial capital instead of lowercase letter, such as After, Against, Age,