Level?
Etymological parallels to the *ḏV/*ˀVl(l)- system in Akkadian are not quite transparent. As argued by J. Huehnergard (2006:116‒118), *ḏV is cognate with the Akkadian relative pronoun ša, whose prototype *ṯu can be reliably established on the basis of the Sargonic orthography with ŠV signs. An obvious problem faced by this comparison (otherwise quite appealing if only for distributional reasons) is the apparently unmotivated shift from *ṯ to *ḏ (or vice versa).
These difficulties notwithstanding, it still seems easier to assume (with Hasselbach 2007) that the WS situation reflects the original PS picture, whereas Akkadian, here again, has innovated.