Electronic data processing has made the task of producing reverse dictionaries easier. A reverse dictionary had been prepared as early as 1873 for the language of the Rigveda, and others, such as for Latin and Greek, all produced by means of hand-written slips, soon followed.(1) For Assyriologists, the reverse dictionary (Rucklaufiges Worterbuch, henceforth RW) of Karl Hecker represents a novum, although a forerunner appears at the end of Simo Parpola's Neo-Assyrian Toponyms (AOAT 6 ), as a "reverse index of the key words" (pp. 387-402), as was pointed out to me by my colleague Remko Jas.(2)
Reverse dictionaries go back, in fact, to a much more distant past. The dictionary that can claim the honor of having been the first reverse dist and thus provides no clue to show that quadriliterals roots all have a liquid (I or r) as the second of the four radicals. The problem of coding verbs with identical first and third radical (e.g., sagasu, karaku) would also have been better solved, I believe, by using the letter P in both first and third positions rather than the letter S twice, as, e.g., in iSaRRiS (isaggis), or SuRRuS- (sullus-), though the notation works well for the mediae geminatae and the tertiae infirmae, i.e., the use of the radicals R-S-S in the first case and of P-R- in the second.
The list of morphemes resulting from this approach is so far unique in a traditional grammatical description of Akkadian; its use has been sanctioned, it seems, by the reverse index, although, of course, several other types of lists of interdigitated morphemes can be envisaged.
A pendant to the list of "independent" morphemes is the main part of RW, section IV, titled "Lexeme." (A lexeme, in the author's usage, seems to mean a word stem without suffixes.) Whereas in the morpheme list, section 11, part 1, the consonants of the root are represented by one or more from among the capital letters K, P, R, or S, in the appropriate position interdigitated with the vowels and the pre- and infixes, in section IV these "place holders" are replaced by actually occurring radicals. Thus, for saPRuSat of section IIa we find sapsuqat in section IV; for uPtataRRiS- we find untatazziq-, for iPtanaRRis- we find istanaddih-. Some of the forms end in a string of letters that do not belong to the root, as the above cited saPRuSat; other forms are scattered throughout the list according to the stem-final consonant. By the same token, the (actually attested) items in section IV provide an easily accessible list of possible consonant clusters, and even of the possible epenthetic vowels to resolve these clusters in word final.
The second part of section 11, "Gebundene Morpheme, Suffixe, Enklitika," lists all the possible suffixes and suffix combinations, as well as most enclitics. The separation of the two lists, whose elements can combine with one another, makes for a more economical presentation. The reverse list of the bound morphemes (gebundene Morpheme) marks, by a hyphen, not only the boundary between the omitted "lexeme" and the bound morpheme - which is redundant - but boundaries between the several morphemes of a string that occurs in stem-final. The bound-morpheme list comprises stem-final suffixes or suffix sequences, including the enclitic particle -ni (also known as the "Assyrian subjunctive"), the possessive suffixes, adverbial endings (e.g., -amma, -atum), and derivational endings, both in absolute final position and when they are followed by a further bound morpheme. The inflectional endings are stripped off, but hinted at by a final hyphen. Elements that do not stand in word final are not included, and this particular information is not available in RW or other reverse dictionaries. infixes are not listed, nor are prefixes. Many concatenated endings are rare and thus it is difficult, if not impossible, to identify the word or words so formed, as in the case of -(a)tatu-sa-ni, -utti-sunu-ni or utta-sna. It is not stated which bound morpheme or morphemes can occur with which "lexeme" even though most "lexemes" end in a hyphen to indicate that they are followed by (bound) morphemes. The homophonous suffixes under "gebundene Morpheme" in this section (Ilb) are distinguished neither as to meaning nor co-occurrence nor for dialect distribution. For example, [-kunu.sub.1] that follows a noun to indicate the possessive suffix of the second person plural is not distinguished from [-kunu.sub.2] that follows a verb to mark the second person plural object. The fact that [-kunu.sub.2] is restricted to the Assyrian dialect is not mentioned either (in the Babylonian dialect this ending appears as -kunuti and thus is alphabetized in a different place). Eventually, the meanings-or at least the functions-of the bounist and thus provides no clue to show that quadriliterals roots all have a liquid (I or r) as the second of the four radicals. The problem of coding verbs with identical first and third radical (e.g., sagasu, karaku) would also have been better solved, I believe, by using the letter P in both first and third positions rather than the letter S twice, as, e.g., in iSaRRiS (isaggis), or SuRRuS- (sullus-), though the notation works well for the...