Moreover, as already discussed, Prābhākaras claimed that words in a sentence only convey a connected sentence meaning and not a heap of distinct word-meanings. This does not amount to say that the sentence-meaning is partless, but that its parts (the word-meanings) mutually influence each other, so that they would not have exactly the same meaning if they appeared in a different sentence. In a fundamental article on this theory ("The Context principle and some Indian controversies over meaning", Mind 1988), B.K. Matilal and P.K. Sen asked whether this amounts to admit "unsaturated" entities in the world. The following account by Rāmānujācārya might explain why this is not the case:
(that is, the coordinated apprehension of the word meanings occurs through the coordinated apprehension of the sentence meaning, which again depends on suitable rules). Next:
The understanding of the sentence-meaning from the words (pada) occurs indeed in the following way. Firstly the words, while they are heard, cause one to remember (smṛ-) their respective meanings, reciprocally unconnected (ananvita). Then, having remembered them, in the hearer arises a reflection (vimarśa) regarding their own meanings, and beginning in this way: “This is one sentence, this is a split (bhinna) one; this is the meaning the speaker wished to convey (vivakṣita), this is the undesired meaning; this is metaphorical, this is primary; this is worthy of connection; this is principal, this is not principal; this has been prescribed, this has not been prescribed…”. Thereafter, there is a division of fit (yogya) and unfit regarding the coordinated apprehension (anusandhāna) of the exact word-(śabda) meanings through a coordinated apprehension according to the rules (nyāya) appropriate (aupayika) for the ascertainment (nirṇaya) of the sentence-meaning
Then these words, brought (back) to memory, express a related [meaning].
(TR III, emphasis mine)
On the expression of a connected meaning, see here.