In this work, Cohen seeks to analyze the influence of syntax on the morphological development of the verbal system in Semitic. Specifically, he argues that it is the potential of the nominal sentence to produce new verbal forms that drives the a cycle of change in the verbal system.
A nominal sentence should not be seen as an ellipsis of the verb, but as a predicate expression in its own right. For instance, in Biblical Hebrew you can not alternate a verbal phrase for a nominal without altering it fundamentally, introducing additional semantic notions as well as a determination of tense.-aspect. For example, adonī ḥākām can be translated as “my lord is wise”, a present tense. To insert hāyāh is to add a perfective nuance, yihyēh imperfective, yehī jussive, etc. Only the nominal phrase expresses the simple present. Thus, Cohen argues that the present is not a notion attached to the verbal system in Semitic, but to the nominal phrase. Tense is expressed by syntax in such a case, not morphology.
Specifically, the predicate nominal undergoes a process of verbalization. Cohen argues that in Semitic a verb can be differentiated from a noun by three features: A) it is marked as a predicative form; B) it constitutes a complete statement, being comprised of a verbal base and a morphologically marked personal subject; C) it enters into the system of oppositions expressing tense-aspect.
This can be illustrated by comparing the Akkadian stative with the West Semitic suffix conjugation. The Akkadian stative/permansive is quite clearly a conjugated predicate adjective with features A and B above, however it has not completed the process of verbalization as it is not a part of the tense-aspect system. In West Semitic, however, the form has completed the process, being brought into the opposition qatala : yaqtulu.
Arabic has obviously levelled the paradigm by using the -k- of the 1cs form, while Ethiopian has levelled the paradigm by using the -t- of the second person forms. Thus Semitic in an ancient phase must have had a permansive-stative form which developed from the nominal phrase, and which in the West Semitic languages was integrated into the verbal system to express the perfect. Paradigms of a similar structure in the West continued to form sporadically, such as Classical Mandean, by agglutination of reduced forms of the subject pronoun to an adjective in the predicate state.
Cohen calls this type of construction direct, in contrast to the indirect construction such as is found in Neo-Aramaic. Here a passive participle is combined with a complement introduced by the dative particle l-. Thus šmīe l-X is a nominal sentence “(it is) heard by X ” which has developed into a true verbal form expressing the perfect “X heard”. This is likely also the origin of the Egyptian sdm.n.f form with -n- being some sort of dative particle.
The history of Aramaic, longest and best attested of the Semitic languages, illustrates how these verbalized nominals cause the rearrangement of the verbal system as a whole. The base of the system seems to be the introduction of special forms to express concomitance. For example, “he writes (in general, non concomitant)” versus “he is writing (write now, concomitant)”.
Often the main opposition of accompli : inaccompli seems to be subdivided into concomitant : non concomitant. The more complicated systems develop this subdivision for both the accompli and inaccompli, the less only for the inaccompli. The active participle tends to be developed into a concomitant form in the inaccompli, the passive participle for the accompli (as a perfect).
The use of the participle then begins to encroach on the function of the older forms (for example the base prefix and suffix conjugation). For example, as concomitant, the active participle naturally begins to be used for the real present “he is writing”, limiting the prefix conjugation to the general present “he writes”. The participle also naturally expresses imminent future as in English “he is going to write”. The older forms therefore begin to be limited to modal and subordinate functions, until eventually being pushed out of the system altogether as even newer forms develop.