Royal praise poetry: c.2.4 (Ur III) and c.2.5 (Isin) versus c.2.6 (Larsa minus Gungunum A/B)

In her dissertation on the literature of the Larsa dynasty, Brisch (2003)¹ sets out to examine the alleged claim that "the language of the Larsa court literature [...] is significantly different from the Sumerian at Nippur, which appears in the hymns of the Ur III and Isin rulers and other texts" (Brisch, p. 127). She goes about this task by giving "examples of selected characteristics of Larsa Sumerian in morphology, syntax, and semantics, and [comparing] these to some features of parts of the Ur III and Isin literary texts — when feasible — or to what would be considered "Standard Sumerian" according to grammatical studies" (p. 131).

The above investigation of the language of the Larsa court literature follows the main part of Brisch's dissertation where she shows that the Larsa literature diverges from the literary tradition found in the Ur III and Isin compositions. This is perhaps most cleary seen in the use of two (new) subscripts: RN lugal-ĝu₁₀ 'O RN my king' and šud₃-de₃ 'prayer'.

The mini-study reported here is an attempt to see if one can get at, or highlight, some of the same "selected characteristics" as Brisch using frequencies and statistical measures. Since this is first and foremost meant as an exercise in how to use frequencies and statistics in studying Sumerian language and literature, only a few selected points from Brisch's dissertation will be commented on. For a detailed analysis of the court literature in the name of the Larsa rulers, Brisch's work should be consulted. Any misinterpretation of the data or Brisch's analysis is of course my responsibility.

Since the divergence mentioned above is first seen in the compositions praising the ruler Sîn-iddinam, earlier Larsa compositions are not part of the comparison below, i.e. Gungunum A (c.2.6.2.1) and Gungunum B (c.2.6.2.a). This unfortunately reduces further the amount of Larsa material, which consists of 12 compositions (c.2.6.6.1 - c.2.6.9.a). The royal praise poetry of the Ur III and Isin dynasties is included in its entirety and comprises 89 composition (c.2.4.1.1 - c.2.4.5.5 and c.2.5.1.2 - c.2.5.8.1). The two text groups will be referred to as LRPP (Larsa royal praise poetry) and UIRPP (UrIII/Isin royal praise poetry) respectively, following the headings used on the ETCSL website.

Contents

1. Items occuring in one but not the other text group
2. Lexeme frequency
3. Word frequency
4. Lexeme bigram table
5. Word bigram table
6. Key words
7. Summing up

A quick look at the lexemes occurring more than once in the Larsa material, but not in the UIRPP, reveals that they are mostly proper nouns. It may suggest that very little new vocabulary is found in the LRPP, as defined here. ki-šu-peš, saĝĝa and ki-ĝar, however, are not attested in the UIRPP, but occur a few times elsewhere in the ETCSL.

The words found in the LRPP, but not in the UIRPP may be more interesting to follow up, to see if they are unique to LRPP. The form unu₇ '(dining) hall', if that is the correct transliteration, seems to be unique to the LRPP.

If we disregard the proper nouns in table 1.2, we get a number of lexemes and words which tie nicely into Brisch's main points. The number of words for composition divisions found in the UIRPP, but not in the LRPP, ki-ru-gu₂, ĝiš-gi₄-ĝal₂-bi-im, sa-gid₂-da-am₃, point to structural differences between the two groups of texts. It would also be interesting to follow up the use of the lexemes di kud, tukul, gul, and munus in the UIRPP to see if this has something to do with a difference in recurrent themes between the text groups. However, once more a note of warning should be sounded. Since we only have 12 compositions from the LRPP to compare with, we can only establish tendencies and not clear-cut differences based on the data at hand.

Finally, let us look at a few bigrams (see also sections 4. and 5.) that only occur in the one, but not the other text group. Right at the top of the first list in table 1.3 is the lexeme that includes the subscript (RN lugal-ĝu₁₀) referred to by Brisch (and others) as a characteristic of LRPP, setting it apart from earlier compositions of royal praise poetry.² The phrase "abula mah" (majestic (city) gate) is also particularly prominent in the LRPP. It only occurs twice elsewhere in the ETCSL. (See Brisch p. 94 for a comment). The lexeme bigram nanna nin-gal is interesting since it occurs four times in the form ^dnanna ^dnin-gal-bi. According to Brisch (pp. 141 - 142), the suffix -bi, which is traditionally seen as an inanimate, posessive determiner, here functions as a conjunction, which is a development seen more clearly in the LRPP than other compositions. We shall return to hul₂ 'to be happy' below.

As regards the right-hand side of table 1.3, i.e. the bigram lexemes occurring in UIRPP, but not in LRPP, a note should be made of the many bigrams containing Enlil and Inana. Notice also the frequent bigrams a2 mah 'great power/strength' and silim dug₄. Why doesn't the praise poetry in honour of the rulers of Larsa mention "great strength"; and why is there no mention of eloquent talk (silim-eš₂ dug₄) or healthy greetings (silim-ma dug₄) in those compositions?

When reading and comparing the data in table 2.1, one may want to focus on the figures in the column headed 10k, which is the estimated number of occurrences of the relevant lexeme per 10,000 words. Based on our data, these figures tell us, for instance, that the lexeme cag4_N_heart is twice as likely to occur in the LRPP than in UIRPP. Without knowing anything about the background of the two text groups, one may also wonder about the relative frequencies of the lexemes en-lil2_N_Enlil and inana_N_Inana (occurs 4 times only in LRPP) in the two lists.

Similar observations to the ones made about lexemes can be made about words. Notice for instance the abscence of the word a2 in the LRPP list. It occurs only twice in the LRPP material. Why is that?

So-called n-grams, that is co-occurrences of two or more items (lexemes, words), may be a better way of teasing out the differences between compositions than the frequencies of single items. Why the happy hearts in Larsa (cag4_N_heart hul2_V_to#be#happy), and why this frequent (self-)refence to black-heads (saj_N_head gig2_V_to#be#black) and Sumer (ki-en-gi_N_Sumer ki-uri_N_Akkad) in the UIRPP.

Apart from noting the dwindling frequencies in table 5.1, no comments will be offered. The fact that we get very low figures when working with small data sets also makes it less relevant to produce additional trigram tables.

The table of key words, table 6.1, in many ways summarises what we have discussed so far. All the lexemes in table 6.1 are statistically significant in the sense that they occur more often (positive numbers) or less often (negative numbers) than expected in relation to the reference corpus (see Comparing ... in the Introduction). So again we can note the high keyness figure for proper nouns, abula, sikil, šag₄, etc., and high negative keyness for e.g. en-lil₂ and a₂.

Note that some of these are weak key words due to a very low frequency threshold of 2 and a relatively high p value (0.1). The values used when comparing the proverbs to the whole corpus were 3 and 0.0001 respectively.

Throughout this mini-study we have noted a distinctive difference in vocabulary between LRPP and UIRPP, e.g. nam-til₃, sikil, šag₄, he₂-ĝal₂, še, and hul₂ versus a₂ and ni₂, etc. Brisch (p. 160) offers this tentative conclusion: "Given the form and content of the Larsa court literature, which is more or less dramatically different from its Ur III and Isin dynasty precursors, one might contemplate a conscious effort of scribes to break with the traditional phrases and expressions in the language of legimitimization as well". This shows, I think, that the kind of lists and tables produced for this exercise can be useful as first step in unearthing differences in vocabulary and style between groups of compositions. However, it must be stressed that this is just an initial step which should always be followd up by a detailed analysis of the relevant items in context. We can finally also note that we have not been able to highlight more subtle differences in the morphology of the two text groups, differences discussed in some detail by Brisch.

¹ Brisch, Nicole Maria. 2003. Tradition and the poetics of innovation: Sumerian court literature of the Larsa dynasty (2003-1763 BCE). Doctoral dissertation, University of Michigan.
² The five most frequent word bigrams of LRPP are

• {d}ri-im-{d}suen lugal-ju10 (8)
• {d}ri-im-{d}suen lugal (6)
• abula mah-a-ke4 (5)
• ur-saj usu (3)
• nam nam-til3-la (3)