Two different construals of linear order are currently entertained in generative models. Under one model linear order is defined in the syntax (Kayne 1994), while under the alternative model, precedence is not defined in the syntax, but only as part the Externalization procedure (Chomsky 2013). The Kaynian model requires (locally) total ordering of constituents by asymmetric c-command; the Chomskyan model does not. Linearization at the interface is compatible with traditional word order parameters (e.g. head to the right of head to the left). Linear order in syntax requires word order parameters to correspond to the application (or not) of movement operations. Which model is favoured by empirical evidence? A good introduction to the issue is represented by the discussion of Greenberg’s Universal 20, concerning the relative order of demonstratives, numerals, adjectives and Ns in the NP. I will briefly consider the work of Cinque (2005, 2025) and the rejoinder by Abels and Neelemann (2012).
After this introduction, the course will be devoted to the internal order (or lack thereof) of adjectival modifiers of N. We will first addresses classical ideas (ultimately due to Dixon) that adjectival modifiers are ordered by semantic class (color, shape, size etc.). Cinque (2014, 2025 and references quoted there) builds the typological sequence into a cartographic-style sequence of adjectival heads. I present evidence that this system undergenerates (it is too strong) and that the data favour a much coarser organization of modifier space than that defined by semantic classes (Manzini 2025a, 2025b, Scontras, Manzini and Mazzaggio 2026, and the literature discussed there).
The second part of the presentation focuses on Italian, where all adjectives can be freely ordered with respect to one another in postnominal position. Following insights of previous literature, Cinque proposes that unordered adjectives are reduced relative clauses; these have a separate position in the tree, corresponding to a different set of interpretations. I argue that while there is independent evidence for two different adjectival structures (such as the phenomenon of linkers), there is be no one-to-one correspondence between internal structure of modifiers and their interpretation – or in fact their position (see the refs above).

Course materials

• Abels, Klaus & Ad Neeleman. 2012. Linear asymmetries and the LCA. Syntax 15.25–74. https://doi.org/10.1111/j.1467-9612.2011.00163.x get pdf
• Cinque, Guglielmo. 2005. Deriving Greenberg’s Universal 20 and its exceptions. Linguistic Inquiry 36.3.315–332. DOI: 10.1162/0024389054396917 pdf
• Cinque, Guglielmo. 2014. The Semantic Classification of Adjectives. A view from Syntax. Studies in Chinese Linguistics, 35.1, 2014, 3-32. pdf
• Cinque, Guglielmo. 2025. The superiority of a movement approach to Greenberg’s Universal 20 (a reply to Dryer 2018). Syntactic Theory and Research 1.1. 7. pdf
• Manzini, M. Rita 2025a. Left–right asymmetries in Italian adjectives: partial order and phases. Syntactic Theory and Research 1.1.5. pdf
• Manzini, M. Rita. 2025b. Italian adjectives: why fully ordering them in the syntax does not quite work. RGG, 47(3). pp. 1–29. pdf
• Scontras, Gregory, M. Rita Manzini & Greta Mazzaggio. 2025. Italian Adjective Order Is Not Predicted by Syntactic Hierarchies and Movement. Linguistic Inquiry 1-19. pdf

Other references

• Chomsky, Noam. 2013. Problems of projection. Lingua 130. 33–49.
• Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge (Mass.): The MIT Press.