This thesis defends the following argument:

1. Signs of macro-structural planning indicate a translator has monitored the performance of the product as Greek discourse.

2. Unique target language items that generate inferences in the product are one sign of macro-structural planning.

3. The product contains unique target language items that generate inferences.

4. So, the translators monitored the performance of their product as Greek discourse.

5. So, the translators had a pragmatic presupposition that their intended audience would detect inferences generated in the product.

If sound, this argument demonstrates that utterances in the product with unique target language items were communicatively independent from their source. The best explanation for the amount and distribution of such utterances is that the product itself was intended to be read as a monolingual corpus. This thesis then proceeds to demonstrate that the translators deliberately licensed discourse-pragmatic meaning borrowed from the source in cases of literalism:

6. If (5), then neither the translators nor their intended audience were oblivious to the same inferences when they were borrowed from the source.

7. So, macro-structural planning also includes licensing inferences borrowed from the source.

8. So, the translators took a maximal stance toward crosslinguistic symmetry in the product.

If successful this argument demonstrates that the Septuagint was intended to be read as a monolingual corpus even when its meaning had been borrowed from the source. This argument provides an explanation for a high degree of literalism in the product: written convergence was an epiphenomenon of crosslinguistic symmetry. Alternation between different modes of translation arose as the translators balanced the consequences of asymmetric contact with a desire for interpretive resemblance with the source, leading to a product that appears to be both ‘literal’ and ‘non-literal’ at the same time.