How real are adjective order constraints? Multiple prenominal adjectives at the grammatical interfaces

Abstract

Adjective order restrictions on attributive adjectives (AORs) have been subject to debate in modern linguistic research for a long time. Most generally, the question whether AORs can be located in grammar as such in rule-based fashion is still unsettled. In the current paper, we largely argue against this view and claim that several of the core data to be explained are preferences based on norms rather than rules. A pragmatic explanation is offered to account for marked or apparently ungrammatical examples. First, we demarcate AORs in the narrow sense against data based on truth-conditional differences, show the sole hard constraint to be found in a distinction between object- and kind-modification, and introduce several of the factors argued to drive AORs in the literature. A large-scale corpus study on German AAN-phrases shows a hierarchy of relative adjectives preceding absolute ones to reliably predict preferences, while temporariness and weight do not. We then illustrate that norm-based preferences can be overwritten via discourse linking and implement markedness in out-of-the-blue contexts pragmatically based on the M-principle. Speculating that AORs in the narrow sense have their origins in more general cognitive principles, our findings support approaches that locate the better part of AORs outside the realm of core grammar.

Universal linguistic hierarchies are not innately wired. Evidence from multiple adjectives

“…Both these orderings violate (4), given that they were not featured in a context that would justify lifting ordering restrictions. At the same time, Kotowski & Härtl’s (2019) overall results do show the presence of one hard constraint: in their corpus, relational adjectives (i.e., non-gradable adjectives that typically derive from nouns and express a relation between the noun on which they are formed and the one they modify; for example, material in ‘wooden toy’) are always found closer to the noun than other adjectives. Incidentally, origin and material are the two categories that occupy the bottom positions of the hierarchy of closeness to the noun in (4), hence this finding is compatible with the hierarchy put forth in the cartographic approach too.…”

“…Recent corpus studies have cast further doubt on the predictions of (4). For example, Kotowski & Härtl (2019) document naturalistic data that show size adjectives preceding subjective comment/evaluative adjectives or color adjectives preceding shape adjectives. Both these orderings violate (4), given that they were not featured in a context that would justify lifting ordering restrictions.…”

“…A number of recent approaches to adjective ordering aim to derive possible restrictions (Truswell, 2009; Kotowski, 2016; Scontras, Degen & Goodman, 2017; Simonič, 2018; Hahn et al, 2018; Kotowski & Härtl, 2019), without taking them for granted under the assumption that they are dictated by (4). Truswell’s (2009) corpus results suggest that the cartographic approach undergenerates with respect to the adjective orderings that are attested in actual language data: intersective adjective pairs allow free order (e.g., ‘square purple box’ and ‘purple square box’), while the ordering in pairs that feature an intersective and a subsective adjective of the nominal is less free (e.g., ‘big wooden table’ but *‘wooden big table’) 3…”

Looking for Cognitive Foundations of Functional Sequences

“…Asking a similar question about the universal sequence demonstrative > numeral > adjective > noun (Greenberg's 1963, generalization 20), Culbertson and Adger (2014) argue that it is rooted in general cognition. Even more related to the topic of this paper are the findings of Scontras et al (2017) and Kotowski and Hartl (2019), who argue that the universal sequence of adjectives has origins in more general cognitive principles. To motivate the approach that we are taking to address the above question, we need to have a closer look at the functional make-up of the noun phrase.…”

“…4 Naturally, our question could be restated in a theory that does not accept the existence of functional projections, namely, we would be asking what extragrammatical properties influence the universal adjective order. This is the approach of Scontras et al (2017) and Kotowski and Hartl (2019). Given our starting assumptions, we will keep talking about the functional hierarchy, but we are really interested in finding extragrammatical basis that ultimately result in cross-linguistically fixed order of adjectives.…”