In this chapter we investigate the role of socio-psychological motivations in accounts of grammatical change. Laboratory and corpus evidence is presented to substantiate the impact of dynamic prestige meanings (associated with non-posh media cool) on the diffusion of the object pronoun hun 'them' as a subject in Netherlandic Dutch. In a speaker evaluation experiment, 185 listener-judges rated speech stimuli with standard and non-standard pronouns on pictures which were the best instantiations, according to a preceding norming task, of the evaluation dimensions old prestige (superiority), new prestige (dynamism), and disapproval. While subject-hun was found to be significantly less superior than the standard pronoun, it was perceived to be no less dynamic. The impact of this dynamic prestige meaning was further investigated on the basis of a dataset of tweets. Regression analysis demonstrated that the preference for hun could be adequately predicted on the basis of production proxies of hun’s social meaning. Taken together, all the available data suggest that the social meaning of hun is a pivotal determinant of its diffusion, viz. its use as a consciously deployed 'stylizer', but also the internal conditioning of its non-conscious use as a pronoun alternative.

Chapter 2 examines linguistic changes that can be accounted for by solving labeling paradoxes. In Chomsky (2013, 2015), merging a head to a phrase no longer automatically results in the projection of that head into a label and labeling paradoxes arise when two items merge that are (too) symmetric. These paradoxes can be resolved in several ways, namely by having one of the XPs move or by feature-sharing. The resolution discussed in this chapter involves the change from phrase to head, a possibility not discussed by Chomsky. The changes discussed involve pronouns reanalyzing as functional categories, i.e. as T or v, and demonstratives reanalyzing as articles and complementizers. In the changes, a third factor resolution to the labeling problem can be observed: a change from feature-sharing and agree to Minimal Search. The changes also show other factors involved, e.g. the difference between <Q,Q> and <phi,phi> sharing. The wh-elements whether and how are specifiers and show no reanalysis to head, which indicates their feature-sharing is stable.

Chapter 1 provides some background on the shift in emphasis from Universal Grammar (UG) to third factors and gives a description of selected third factors, e.g. the Inclusiveness Condition and the Extension Condition. The main emphasis is on the Labeling Algorithm and the Principle of Determinacy. Generative models focus on the faculty of language as represented in the mind/brain. UG is the “system of principles, conditions, and rules” that all languages share through biological necessity. However, although UG received a lot of attention, recently principles “grounded in physical law” and the general “capacity to acquire knowledge” have been emphasized more. This chapter also introduces two main reasons of language change that are responsible for the linguistic cycle: those caused by economy and those by innovation.

Chapter 5 examines the tension between determinacy and labeling. Due to determinacy, if there is a TP, Verb-second (V2), i.e. V to C, is not possible but TP expletives are. Conversely, if there is no TP, V2 is possible but TP expletives aren’t. I will argue that older stages of English lack a TP and that this enables both V2 and movement of the subject from the specifier of the v*P to the specifier of the CP. It also makes the grammatical subject position and the expletive optional. Later stages of English introduce a TP, which enables expletives in the TP but bars V2. The loss of V2 and introduction of expletives has not been linked before and this offers a new perspective both on the data in English and in V2 languages and on the tension between the two third factor principles.

Chapter 3 considers how the Principle of Determinacy disallows certain constructions and what options are available to ‘fix’ these ungrammatical structures. Chomsky, Gallego, and Ott (2019) rule out having more than one choice in the workspace/phase, i.e. the workspace must be determinate. If a phrase moves from one position to another in the same phase, i.e. without being transferred/eliminated from the workspace, merge will face the dilemma of which of the two copies will move to a higher position. Determinacy is a third factor formulation of anti-locality and accounts for the Subject Island Condition and the ban on topicalization of the subject. The chapter focuses on how the CP–TP complex makes it hard for syntactic objects to move from the specifier of the v*P to the specifier of TP and then higher, to positions where they check the Q-features. Such a movement results in a that-trace violation. Deleting the C solves this problem and also explains subject-less relative clauses and complementizer-less object clauses in English. Other languages don’t posit a TP and solve the anti-locality problem that way. These languages lack that-trace effects but don’t delete the complementizer.

In this book, I have argued that efficient computation is key to language change. Third factor principles such as Minimal Search and Determinacy are implicated in the way languages change and in how paradoxes are resolved. For instance, the change from phrase to head, chronicled in Chapter 2, is a move towards using Minimal Search.

Chapter 4 provides examples of reanalyses, rather than different choices, that are prompted by the Principle of Determinacy. The first change involves the reanalysis of a loosely adjoined phrase as a subject argument because a topicalized subject does not result in an optimal computation. The principle also accounts for changes involving copulas, both the change from demonstrative to copula and from topic to subject. Auxiliaries and quantifiers in English provide fertile ground to investigate determinacy, because these move from lower to higher heads and specifiers, respectively. It is shown that auxiliary movement indeed violates determinacy and that options exist to circumvent it, e.g. skipping T and reanalyzing as a higher functional head. Floating quantifiers do not violate determinacy because they first move as QPs and subsequent moves are of DPs.

Chapter 6 discusses another issue in efficient computations that language change casts some light on, namely through changes affecting adjuncts. Chomsky (2000, 2001) distinguishes between arguments (subjects and objects) and adverbials in terms of ordered pair-merge and unordered set-merge, respectively. I examine changes of VP and NP adjuncts to specifiers positions of functional categories (ASPP and DP, respectively) and of adjuncts to arguments. These changes show that pair-merge can be avoided. Adjuncts that are in specifier positions of functional categories in their turn reanalyze as heads, driven by labeling pressures. I also address the question of whether subordinate and insubordinate adjunct clauses change in unidirectional ways, and conclude that they don’t.

Acceptability and grammaticality are clearly closely related, but the relationship is not always straightforward. Sometimes, sentences that are thought to be ungrammatical are perceived as acceptable, leading to an illusion of grammaticality, or grammatical sentences are perceived as unacceptable, leading to an illusion of ungrammaticality. Such cases occur with morphological ambiguity, attachment ambiguity, agreement attraction, and negative polarity items, among others. Processing difficulty is one of the factors that can lower the acceptability of a grammatical sentence, as may be seen in the effects of constituent length and dependency length on acceptability. In some cases, such as superiority violations and island violations, it has been argued that these may actually be grammatical, but unacceptable, though this is the topic of much ongoing research involving cross-linguistic work and studies on repeated exposure (satiation) and memory capacity. Having better models of acceptability and better ways of directly measuring grammaticality would be desirable.