Experimental Evidence for the Influence of Structure and Meaning on Linear Order in the Noun Phrase, 2017-2022

Recent work has used artificial language experiments to argue that hierarchical representations drive learners’ expectations about word order in complex noun phrases like these two green cars (Culbertson & Adger 2014; Martin, Ratitamkul, et al. 2019). When trained on a novel language in which individual modifiers come after the Noun, English speakers overwhelmingly assume that multiple nominal modifiers should be ordered such that Adjectives come closest to the Noun, then Numerals, then Demonstratives (i.e., N-Adj-Num-Dem or some subset thereof). This order transparently reflects a constituent structure in which Adjectives combine with Nouns to the exclusion of Numerals and Demonstratives, and Numerals combine with Noun+Adjective units to the exclusion of Demonstratives. This structure has also been claimed to derive frequency asymmetries in complex noun phrase order across languages (e.g., Cinque 2005). However, we show that features of the methodology used in these experiments potentially encourage participants to use a particular metalinguistic strategy that could yield this outcome without implicating constituency structure. Here, we use a more naturalistic artificial language learning task to investigate whether the preference for hierarchy-respecting orders is still found when participants do not use this strategy. We find that the preference still holds, and, moreover, as Culbertson & Adger (2014) speculate, that its strength reflects structural distance between modifiers. It is strongest when ordering Adjectives relative to Demonstratives, and weaker when ordering Numerals relative to Adjectives or Demonstratives relative to Numerals. Our results provide the strongest evidence yet for the psychological influence of hierarchical structure on word order preferences during learning.

Languages can be very different from each other. For example, just focussing on the order of words, languages like English put adjectives before nouns ('red house') while languages like Thai put them afterwards ('house red'). Similarly, languages like Vietnamese put Numerals before nouns ('three houses'), while others, like the Kitharaka (spoken in Kenya), put numerals after ('houses three'). If word ordering was simply due to happenstance, we would expect to see all different orders appearing in equal proportion across languages, but we don't find that. In fact, some orders are very common, some are very rare, and some don't seem to appear at all. For example, many languages are ordered like English ('three red houses'), and many are also ordered like Thai, which is exactly the reverse ('houses red three'). But the Kitharaka order ('houses three red') is much rarer, and its mirror image ('red three houses') never seems to occur. Why is this? One of the major controversies in the language sciences is whether we need to appeal to the basic set-up of the human mind to explain the ways languages can vary, or whether these properties are instead a result of cultural differences in communication and social interaction. A great deal of recent work coming from the perspective of psychology assumes the latter: that the properties of language can be boiled down to communication, interaction and the vagaries of history, while most work in linguistics assumes the former: there must be biases in the human mind that allow us to learn languages of particular types more easily than others. This project seeks to resolve that issue. In order to do this, we test how well people learn languages of various types, to see whether their behaviour follows the general tendencies we see across real languages. Importantly, we use artificially constructed languages, rather than natural languages, in order to make sure that they only differ in the crucial respects. For example, we present English speakers with artificial languages that use word orders from Thai and Kitharaka. If Thai orders are more common across languages than Kitharaka ones because the former are easier to learn, then we should see this reflected in the behaviour of learners in our experiments. We can also see whether such patterns are always harder to learn, or if speaking a language which uses them-like Kitharaka-makes them easier to pick up in a new language. To do this, our experiments compare English, Thai, Vietnamese and Kitharaka speakers. If our learners all show the same kinds of patterns in how they learn our artificial languages that we find across real languages, that will suggest that the way languages vary is not random, nor is it entirely a product of historical facts. Rather it would suggest that there are universal cognitive biases at play. We plan to look at not just the basic question of what orders appear, but also two other well-known cases where languages don't seem to vary randomly. The first relates to how words like adjectives and numbers are placed relative to the nouns they modify: most languages place them both before or after (like English and Thai), rather than putting them on opposite sides (e.g., 'two houses red', like Vietnamese). We will test whether this type of pattern is always easier to learn in a new language. Second, we will look at whether people prefer to learn languages with suffixes (e.g., 'cat-s') rather than prefixes (e.g., 'un-happy'). Both types are present in English, but most languages have (more) suffixes. Our project we will shed light on whether there are universal cognitive biases in language learning, if such biases are at play for the particular phenomena we look at, and how people's native languages affect these biases.