A Study in Syriac Corpus Linguistics: Syntax, Change and Definiteness in the Syriac Noun Phrase

Abstract

Historical evidence for Aramaic's three-thousand-year development is incredibly complex and this is no better reflected by its definite article suffix, which first comes into view during Old Aramaic in 8th-9th century BC before its subsequent weakening in Imperial Aramaic by the 5th-6th century BC and eventual extinction in Late Aramaic by around the 5th-6th century AD. The `birth' of the article, so to speak, has received ample attention in the Semitic literature, but its widespread `death' remains a mystery.

Recent developments in computational linguistics and the Digital Humanities make it easier to process large amounts of Aramaic manuscripts for research on language change. Using corpus-based methods, with a view to understanding the distribution between Emphatic (definite noun) and Absolute (bare noun) forms, this study builds a specialised and balanced corpus of the Late Aramaic dialect, Classical Syriac, comprising approximately 400,000 words and structured across ten authors spanning thirteen centuries. An average of 63.13% of total words across all manuscripts were tagged for State Morphology, including Emphatic and Absolute forms. Key quantitative and qualitative findings emerge from corpus searches on the distribution of Emphatic-Absolute noun forms in Cardinal (1-10), Nominal Negation ("dla") and Quantifier ("kl") phrases.

The average proportion of Emphatic-Absolute noun dependents in Cardinal phrases represented a 62-38 distribution in favour of the Emphatic across the corpus period. I propose the weakening and generalisation of the definite article morpheme had already neutralised Cardinal phrases in a pre-Syriac context. Interestingly, nearly 70% of all attestations involving an Absolute noun dependent comprised plural noun fossils of time (e.g., "years", "days", "hours") with remaining productive nouns representing statistically marginal cases. Based on these facts, I suggest that Absolute noun relics play an outsized role in Syriac grammars, which overgeneralise their productive function. Nominal Negation phrases display a similar distribution between Emphatic and Absolute noun dependents with a 72-28 ratio, respectively. Approximately 70% of all attestations involving an Absolute noun dependent comprised a lexicalised collocate phrase (e.g., "without flaw" > "flawless"). Interestingly, a renewed Nominal Negator form "dlaw" is only attested in later manuscripts. I propose this renewed form was innovated via extension from relativised verbal negators sometime after the 4th century. Finally, I demonstrate evidence supporting the fact that the semantic distinction between Emphatic-Absolute nouns in singular Quantifier "kl" phrases collapsed around the 4th-5th century, evidenced by the prevalence of proleptic quantifiers, which took over the function of the definite article in Emphatic nouns. I propose a source of this change via contact-induced extension from other Aramaic dialects.

This study underscores the value of corpus linguistics for studies in historical syntax in an Aramaic context. It fills a lacuna in the typological literature on the nature of definite article loss in human languages, while also contributing fresh evidence in support of syntactic change mechanisms (e.g., reanalysis, extension) in an under-represented Semitic language. The study also contributes new methods in corpus research in Syriac by offering a Manuscript Selection Framework as a starting point for standardising corpus design practices, which uses a range of philological, textual and linguistic criteria. The part-of-speech (POS) tagged corpus also holds significant reuse potential as a resource for training and testing future POS tagging models tailored to Syriac. Finally, the frequency analyses of grammatical patterns in this study supports Syriac teachers and students better understand which constructions are (un)common or (im)probable, thereby improving our understanding of native and standard Syriac grammar. Show more