Worthington's book is an excellent contribution to both textual criticism and Assyriology. The book goes far beyond what it promises and tackles questions of orthography, phonology, morphology and interpretation as well. It is very well written and structured and should thus be easy to understand even for people without a background in Akkadian or Semitic.
As many classicists will be unfamiliar with Akkadian, I shall give an outline of its most important orthographical features, some of which will be referred to below. Akkadian is written in cuneiform taken over from Sumerian. The script has logographic as well as syllabic elements, with many signs having multiple functions and many syllables potentially being written by a variety of signs. Thus, a single sign may function as a Sumerian logogram (for instance Sumerian LUGAL standing for šarru 'king'); as a syllabogram with its original Sumerian sound value; as a syllabogram with an innovated Akkadian sound value; as a determinative, that is, as an indication of what semantic field the following or occasionally preceding noun belongs to rather than as an element to be pronounced (this is important for ambiguous sumerograms which could be read as two different Akkadian words); and, finally, as a phonetic complement, that is, a sign used after a sumerogram either to disambiguate it or to show the Akkadian case ending. Conversely, a syllable may be written with different signs, for instance the syllable u may be written as u or ú.
But let us now return to Worthington's book. The first chapter serves as a general introduction to textual transmission. As Worthington points out, even if a text has a colophon, it is not always clear whether the text was copied by sight, through dictation or from memory, as colophons are often worded ambiguously. The kinds of errors committed are not necessarily helpful, either. Phonetic errors need not point to dictation because they can easily arise if a copyist reads out the text to himself; and errors of misreading can occur in dictation as well because the person dictating is typically reading a text rather than making one up. It is also worth pointing out that the degree of fidelity a copyist aims for depends on a number of factors. Sacred texts are generally copied more exactly than utilitarian documents, for which the precise wording does not matter much; and different scriptoria may also adhere to different standards.
Chapter 2 deals with methodological problems. Worthington introduces concepts which should be familiar to any classicist, such as the difference between an autograph and a reconstructed archetype, which may be quite far removed from the autograph in time and phrasing. The reader is also introduced to stemmatic analysis conducted on the basis of genealogically diagnostic anomalies. Three points in this chapter are worth dwelling on. First, while Akkadian never developed a fully standardized orthography, there are some conventions and some tendencies; for instance, nidintum 'gift' may be spelled ni-din-tum, but not *nid-in-tum, because a single consonant between two vowels is obligatorily interpreted as a syllable onset. Second, not every misspelling is a true error. Some misspellings may reflect current pronunciation rather than conservative convention, but in practice it is often impossible to distinguish between misspellings and language change. And third, in texts with a long transmission history, a later, corrupted version may have become more influential than an earlier, more correct one, and editors need to make it clear whether they are aiming for the original text or for the one that would have been more familiar to the actual speakers of Akkadian.
Chapter 3 discusses the mechanisms of textual change. Sign similarity is a common problem. On p. 92, Worthington mentions the case of a man who asked for pots, but received straw instead, because the sumerograms for straw (INmeš, a plural) and pots (KANmeš) are similar, a problem that would not have arisen if the writer had used syllabograms, as the words are quite different in Akkadian (tibnu versus diqāru). Other errors arose when inherently ambiguous signs were misinterpreted, or when sumerograms were wrongly used for words which had homophones that could be written with sumerograms. Most other error types have close parallels in Greek or Latin, for instance when words or word groups are left out due to homoearchon or homoeoteleuton, when signs are forgotten randomly or by haplography, when dittography or sign inversion occurs or when words receive wrong endings by attraction to neighbouring words. Interpolations, glosses and misguided corrections of supposed errors are also found. A semantic basis can be found in polar errors ('hot' for 'cold') or synonym substitution (malku for šarru, both meaning 'king'1). Assyrianisms, cases where Babylonian texts are coloured by the Assyrian vernacular in the course of transmission, may sometimes be deliberate.
Chapter 4 treats patterns of orthography, phonology and morphology. Within a word, a writer may go for phonological spellings or morphological ones. Thus, in iparrasū 'they (masculine) decide', from a root PRS, the first and last vowels are affixes. A writer opting for phonological spellings would write i-par-ra-su, but one opting for morphological ones would write i-par-ras-u, showing the affixal nature of the last vowel. While phonological spellings predominate within words, word boundaries normally constitute syllable boundaries in writing, regardless of pronunciation; so šumu ukin 'he established a name' would be written šu-mu-ú-ki-in. However, in speech words are not separated from each other, and this may be reflected in occasional sandhi-spellings such as šu-mu-ki-in. Similar sandhi-spellings occur even in combination with sumerograms.
Worthington also examines the 'honorific' nominative of names or titles, used instead of syntactically appropriate accusative or genitive. This phenomenon is already attested in Old Babylonian, if only occasionally. To what extent it foreshadows the collapse of the case system is unclear, as personal names and honorifics often behave idiosyncratically across languages; thus, the unrelated Urartian normally has subject-object-verb order, but deities in non-subject function will mostly occupy the first position in the clause.
On p. 235-8, Worthington discusses the interesting case of plene spellings among contracted verbs. If we have a contracted verb such as ilqû 'they took', the contracted final vowel is commonly written plene (il-qu-ú). When followed by connective ma, an enclitic, some tablets consistently use non-plene spellings (il-qu-ma). Can this tell us anything about pronunciation? Is the vowel shortened, as in the well-known Latin phenomenon called Kürzung durch Tonanschluß (compare sī quidem versus sĭquidem 'if indeed')? The answer is negative. Not all tablets follow this practice, and it is consistently ignored if there is a final long vowel that did not arise through contraction. We seem to be dealing with an attempt at spelling standardization.
Chapter 5 deals with the question how easily Akkadian was sight-read in antiquity. Worthington's analysis is thorough and perceptive and does not rely on modern scholars' preconceptions. Various factors would have helped the ancient reader. For instance, many signs had only one function, as sumerograms or determiners, which decreases the ambiguity inherent in the script. Sometimes, plene spellings were used in order to help readers select the correct sound value of the preceding sign (in which case one ought not to be quick to draw conclusions about vowel length). Some tablets go further and use only ša1 for the subordinator and only šá for the syllable within a word. Exceptions are rare and can often be explained linguistically; for instance, the phrase 'without number' is ša lā nībi (literally 'which not number', a relative clause without verb), and here we do find the spelling šá, but this may point to univerbation, with šalā becoming 'without'.
Other factors must have made reading more difficult. The ubiquitous lack of word separation was probably not as much of a problem as it may seem to a reader trained in European languages, but the ambiguity of so many signs must have been troublesome even to a trained professional. However, most professionals received much less training than modern schoolchildren, and this training would not have extended much beyond the text types which they were going to specialize in. Worthington concludes that even the best readers would not have been able to read Akkadian as fluently as we read newspapers.
Chapter 6 is a short discussion of issues of edition and interpretation. Worthington points out that what one might consider a codex optimus is often incomplete, forcing the editor to rely on lesser tablets. Similarly, the principle of eliminatio codicum descriptorum, a standard practice in Latin and Greek, is in many cases inappropriate because Akkadian tablets are often copied from more than one source at a time, making a purely stemmatic analysis difficult or impossible. Finally, when we are editing a Latin or Greek author, it is of course our aim to come as close to the original wording as possible. Akkadian texts, on the other hand, are often authorless, and there is something to be said for relying on the majority of tablets, even if they are inferior, in order to recreate the text that had the largest cultural influence in the Akkadian-speaking world.
Chapter 7 is a very brief summary of the results of the book. A detailed table of contents, several lists of abbreviations, a thorough bibliography and a clear index will help any reader using this beautifully produced book, which I can recommend unreservedly.2
1. It is hard to tell whether these are complete synonyms. In Hebrew, מלך does refer to a king, but a סר is a king's son, a prince. In Akkadian, Gyges is referred to as guggu šar māt luddi 'Gyges, šar of the land of Lydia', which shows that this role is higher than that of a prince.
2. I am grateful to Grant Frame for going through my review and alerting me to some problems.