Garr, W. Randall. Dialect Geography of Syria-Palestine: 1000-586 B.C.E. Winona Lake: Eisenbrauns, 2004.
In this study, Garr attempts to isolate and catalog all dialectally significant linguistic features for the given geographic range and time period in order to classify the first-millenium NWS dialects. Garr analyzes all the available texts including Phoenician, Aramaic, Samalian, Ammonite, Deir Alla, Moabite, Edomite, and Hebrew. Sub-dialects are also differentiated where possible.
Relationships between dialects reflect social, political, and geographic relationships between people groups. The greater the amount of similar features, the greater the mutual intelligibility of the two dialects. However, following the wave model of linguistic change, influence is measured not by all similar features, but only by shared innovations. This is because not all identical linguistic features are evidence of dialect contact. For instance, some features may be retentions from the parent language. Other identical features may have resulted from independent, parallel development, while others are the result of analogical change resulting from the structure of the language. Some may also be independent borrowing from a third dialect. Therefore, only shared innovations are reliable measures of true contact. The greater number of innovations, the greater the likelihood of common development among two or more dialects.
The features analyzed are divided into phonology, morphology, and syntax. Phonological features include mergers and splits of phonemes and conditioned sound changes. For instance, proto-Semitic emphatic *ð (interdental voiced fricative emphatic, which is usually d with underscore and a dot, but apparently that combination is not in Unicode) merges with /ṣ/ in all dialects, except that it is realized as /q/ in Aramaic, Samalian, and Deir Alla. For example, Hebrew ארץ but Aramaic ארק(א).
An example of a morphological feature is the ending of a feminine singular absolute noun. Byblian and Standard Phoenician end with ־ת [ōt]. Samalian and Deir Alla end in ־ה [ā]. Ammonite and Moabite end in ־ת [at]. Aramaic ends in ־ה [ā] (< *at), but also sometimes in ־ת [at] (<*ât). Hebrew ends in both ־ה [ā] and ־ת [t] with the latter being more typical of the northern dialect and the former of the southern.
An example of a syntactic feature is the use of a narrative tense for the historical past. Standard Phoenician uses the infinitive as the narrative tense while Aramaic and Samalian use the perfect for historical past. There are three exceptions in Aramaic from Zkr where the so-called consecutive imperfect is used. It is also used in Deir Alla, Moabite, and of course Hebrew, though it seems to be falling out of use during the course of the sixth century.
After analyzing all the features, Garr suggests that the NWS dialects should be thought of as a continuum. Phoenician and Aramaic represent the two poles, while the other dialects fall in-between. Ammonite, Edomite, Hebrew, and Moabite are closer to Phoenician (in that order), whle Deir Alla stands between Moabite and Aramaic. Old Byblian is a dialect island on the Phoenician side, while Samalian is the same on the Aramaic side. This seems to agree with the biblical evidence that suggests Old Aramaic and and Hebrew were not mutually intelligible (II Kg 18:26 = Is 36:11), while Phoenician, Ammonite, Moabite, and Hebrew were (Jer 27:3).