The Use of On-Line Databases in Neology

The Use of On-Line Databases in Neology David Jost, Beth Rowen, and Susan Schwartz A he major American commercial dictionaries continue to collect citational evidence the old-fashioned way, although electronic methods beckon and have been used in supplementary ways. In this brief note we will present evidence showing that electronic methods offer valuable assistance in deciding what new words should be entered in a major revision. Our evidence shows, however, that human collection and decision remain important. Our study tested the choice of new entries in three dictionaries by comparing these choices against the instances of use of these same words in DIALOG, an on-line service containing vast amounts of newspaper and periodical text. Each new entry so tested was unique to one of the three dictionaries, The American Heritage College Dictionary , Third Edition, 1993 (AH3C), Merriam Webster's Collegiate® Dictionary, Tenth Edition, 1993 (WlO), and Random House Webster's College Dictionary, 1991 (RHWC), all of which were published in the period 1991-1993. We searched for instances of each word during two periods of time in one of the DIALOG databases called PAPERS, containing full-text versions of 54 United States newspapers. The first period was the time from the last edition of the dictionary until two years before publication of the most recent edition, and the second was from the end of the first period until July 1994. The first period corresponds to the time span each dictionary had to collect evidence for the most recent edition. A look at the second period gives us some idea of how frequently each word has been used following the data collection that presumably led to entry of each of these words. In addition we give the total number of instances of a word in the PAPERS database. There may be some duplication, with the same article appearing in more than one database. In such cases, the duplicate uses of a word are counted as separate instances. Table 1 NEW WORDS IN THREE DICTIONARIES: FREQUENCIES OF OCCURRENCE IN SOURCES AH3C 19831992 1991 1994 Total WlO 19841992 1991 1994 Total RHWC 19761989 19901994 Total chemical dependency crewed disemploy Generation X lake effect magical realism meroplankton meta-analysis mountain bike ombudsperson potato skin 56032436 879433 8 12 2273594 280 339 1 57 2987 46 192 258 330 1 114 4372 61 71 8155 1372 21 3981 569 682 2 173 7476 108 265 benign neglect1437 computer-9 phobe ganache406 happy talk1060 karaoke meals-onwheels radwaste redial sudser technobabble wire fraud 734 5004 32 480 66 59 6416 866 287 623 3928 4095 16 462 54 99 2732 2418 16 704 1764 4834 9617 55 985 131 160 9450 dancercise 46 dis1872 genomic DNA0 global3920 warming house music350 metalhead29 narcoterrorism26 Populuxe42 quality time1680 techno-thriller7 triple710 witching hour 36 84 50907060 3 4 1854422885 14511827 223256 6190 4284 52187065 6477 6831400 18___________David Jost, Beth Rowen, and Susan Schwartz_____________ The table on the previous page gives our results. In this note we can merely point to a few possible implications of diis evidence. The first is that, on the basis of their frequency in the database, a number of these words could well have been included in all three dictionaries, e.g., global warming, meals-on-wheeh, wire fraud, chemical dependency, mountain bike, quality time, and dis. On the other hand it would appear that some choices seem marginal considering the DIALOG evidence. But odier considerations besides sheer amount of evidence enter into decisions about inclusion. Meroplankton , for example, is an important type of plankton. Ombudsperson is an example of a gender-neutral term that some users might want to use even though its frequency is low. Lake effect is frequent orally, especially on television weather reports in the Great Lakes region, although it is rare in printed sources. In a complete citation gathering program, DIALOG and similar databases will grow in importance. A traditional reading program can help lexicographers turn up new words and senses while DIALOG and its cousins can help lexicographers determine frequency of this new material. Such database evidence, however, cannot replace reasoned decisions, nor can it replace the careful reading of texts by a dictionary staff to...