1 Introduction
Generative grammar has its beginnings in the late 1950s with the work of Noam Chomsky and emphasizes innate linguistic knowledge, or Universal Grammar. Children use their innate knowledge and, on the basis of the language they hear spoken, also known as the E(xternalized)-Language, come up with a grammar, also known as the I(nternalized)-Language (see Chomsky Reference Chomsky1986: 19–24). Generative grammar focuses on the ability of native speakers to speak and understand grammatical sentences.
Since the focus of generative grammar is on the linguistic knowledge present in the mind of a native speaker, some generativists think that it is impossible to study historical stages of a language because native speakers of, for instance, Old English are unavailable. Others argue that the historical data are rich enough to see how speakers, due to a changed input, acquire a different I-Language from older speakers.
In this coda, I provide a brief generative perspective on what features prominently in this volume as the Uniformity Principle. I show that generative views are divided but that most tacitly assume continuity of language acquisition and processing over the last millennia. I also offer some generative comments on two of the articles, framing one in terms of Extravagance.
The outline is as follows. In section 2, I sketch what the Uniformity Principle is and how a generativist would consider it. Section 3 looks at insubordination and suggests a pragmatic strengthening scenario in the cases of somewhat mirative contexts, amazement, requests and warnings. Section 4 comments on changes in the English pronoun system.
2 The Uniformity Principle
Lass (Reference Lass1980, Reference Lass1997) formulates the Uniformity Principle as the ‘likelihood [that] any linguistic state of affairs . . . has always been roughly the same as it is now’ (Reference Lass1997: 29). With this principle, which others refer to as the Uniformitarian Principle, Lass wants to emphasize that ‘there are no miracles’ (Reference Lass1997: 25) in (historical) linguistics. The examples he discusses go back to proto-languages at ‘a time-depth of 10,000 years or so’ (Reference Lass1997: 309).
Winters (Reference Winters, Winters, Tissari and Allan2010) sees this principle as evidence for the stability of principles of human cognition and Traugott (this volume) emphasizes that the linguistic processes rather than the states are uniform and therefore employs the term Uniformitarian Process Principle (hence UPP).
Most of the articles take the Uniformitarian/Uniformity Principle (hence UP) as their starting point. Traugott's contribution shows that the rise of insubordination is not a miracle but can be seen as grammaticalization and subjectification, processes that have occurred in the past as in the present, not as degrammaticalization. She shows that, although the repertoire of complementizers expands, Old and Middle English definitely use non-clausal chunks in ‘negotiated interaction’. Petré’s article on cognitive motivations through time argues that the extravagance typical of incipient grammaticalization can be measured in seventeenth-century English and functions very much the way it does now. De Smet & Van de Velde show how priming of -ly adverbs depends on their transparency both synchronically and diachronically. Cole does not explicitly mention the UP but her results are compatible with it, namely that the choice between personal pronouns and demonstratives for anaphoric use in Old English is very similar to that in Modern Dutch and German.
Thus, the data discussed in the articles go back 1200–1300 years. As Lass intends it, the issue of the UP can, of course, be extended to earlier periods for which we have evidence, e.g. writing found in Henan, China, dates back 8,000 years, Egyptian 5,300 years, and Mesopotamian cuneiform is over 5,100 years old. These languages have been shown to undergo regular processes of grammaticalization and cyclical change (e.g. Hodge Reference Hodge1970). Whether we can go back further has actually been a debate in generative grammar but not in the literature using the UP(P).
As far as I know, the generative literature has not mentioned the UP explicitly. As hinted at in the introduction, Generative Grammar is somewhat wary of work in historical linguistics because of the latter's reliance on corpora and written texts (see van Gelderen Reference Gelderen, Bowern and Evans2015). However, Hauser, Chomsky & Fitch (Reference Hauser, Chomsky and Fitch2002) discuss language evolution from animal to human communication systems. They assume that the former can acquire a language with certain characteristics, in particular recursion, whereas the latter cannot. According to this work, that difference is due to a ‘Great Leap’ that involved one change. After that initial change, not much happened. That position assumes nothing has changed in the way we acquire language which is similar to the UP and UPP. In recent generative work, there is in fact a focus on minimizing the role of Universal Grammar because our evolutionary history is relatively short. Initially (e.g. Chomsky Reference Chomsky1965), many principles were attributed to Universal Grammar (UG) but currently (e.g. Chomsky Reference Chomsky and Belletti2004, Reference Chomsky2005, Reference Chomsky, Sauerland and Gärtner2007), there is an emphasis on principles not specific to the faculty of language, i.e. UG, but to ‘general properties of organic systems’ (Chomsky Reference Chomsky and Belletti2004: 105), also called ‘third factor principles’ in Chomsky (Reference Chomsky2005).
Other generativists, e.g. Pinker & Bloom (Reference Pinker and Bloom1990) and Jackendoff (Reference Jackendoff1990), have a different view on the evolution of language, namely that ‘many aspects of language have recently evolved by natural selection for enhanced communication’ (Jackendoff & Pinker Reference Jackendoff and Pinker2005: 212). For instance, although animals and humans experience the passing of time, humans have fine-tuned this into minutes, weeks, etc. Thus, they allow for ‘capacities [to] . . . have been substantially modified in the course of human evolution’ (Reference Jackendoff and Pinker2005: 224). Van Gelderen (Reference Gelderen, Botha and Knight2009) argues that several steps are required in the evolution from pre-syntactic language (with semantic and phonological systems) to language as we currently know it (with syntax). The one is merge and the structural and thematic relations it entails to build a basic lexical layer (the VP). The others are the Economy Principles that result in what is known as grammaticalization and that build the non-lexical layers (the TP and CP).
In short, although the UP is not mentioned in the generative literature, it is compatible with Hauser, Chomsky & Fitch (Reference Hauser, Chomsky and Fitch2002) but not necessarily with Pinker & Bloom (Reference Pinker and Bloom1990), Jackendoff (Reference Jackendoff1990) and van Gelderen (Reference Gelderen, Botha and Knight2009).
3 Extravagance and insubordination
Although extravagance is tested in Petré’s contribution, it is possibly also relevant in Traugott's article. Petré (this volume) argues that the present progressive took over from the simple present through a process of pragmatic strengthening. Extravagance can be defined as ‘talking in a way that one is noticed’ (as in Petré) but can also be seen as pragmatic strengthening, e.g. in the case of negatives, or as renewal, in the case of cyclical change. Generative grammar sees these pragmatic factors as external ones, chosen by the language user but ones which the language learner responds to and, at some point, reanalyzes as the unmarked option. Insubordination may be another case of extravagance, although Traugott doesn't put it that way.
Traugott lists the characteristics of insubordination as the use of, on the one hand, a subordinator, subjunctive (i.e. dependent) mood and subordinate word order, and, on the other hand, an independent syntactic use. She distinguishes because-answers to earlier questions (e.g. her examples (7) to (10)), where ellipsis is obvious, from those where the main clause is not recoverable, as in exclamatives; the latter are the ones where pragmatic strengthening is especially relevant.
A Dutch example of insubordination is given in (1), where not only the subordinator als is present but the word order is Verb-final, typical of a subordinate clause, as the difference between (1a) and (1b) shows. Insubordinates come in many kinds, and may start as ellipsis, as (2) shows for (1a), where the main clause is assumed to be elided.
-
(1)
-
(2)
Evans (Reference Evans and Nikolaeva2007: 387–422) provides a typologically rich list of the functions that the insubordinate has. Indirection, control, polite requests, warnings and admonitions are among the typical uses of insubordinates. So, in order to warn someone, it is more effective to utter (3) than its non-elided counterpart in (4) where the meaning is the same.
-
(3)
-
(4)
Abraham (Reference Abraham2016) notes that modal particles are useful in distinguishing illocutionary independent clauses from dependent ones. That shows that, even though the word order is Verb-last, the functional head C must have illocutionary force for an utterance like (1a) but not in (2). This is shown in (5), where soms and nog express mood in (5a) but are temporal/aspectual adverbs in (5b).
-
(5)
Insubordination emphasizes degree, which is often expressed in the highest level of the clause, the CP. For instance, in Old English (6), the adverb adds degree modification to the modal.
-
(6)
The difference between (3) and (6) is that the latter has main clause word order so clearly independent, although both are extravagant in function. Generative grammar sees extravagance as external but, if enough instances of ambiguous (1) occur, there will be a change. This has occurred in (5a), which has independent illocutionary force. I'll now look at that in a minimalist way.
If the CP is an embedded CP, as in (2), the C needs an uninterpretable feature to function as C to its own clause (u-Q or u-T), but it will also need a feature that values something the higher verb selects. Roussou (Reference Roussou2010: 582) puts the function of the conjunction as having the ‘dual capacity of being selected by a matrix predicate and of selecting a clause’. In a model that uses features, such as Chomsky (Reference Chomsky1995), the valued interrogative features on C would be selected by the higher verb, as in (7a), where Q represents a condition. The reanalysis would involve a simplification to one set of features in C, as shown in (7b)
-
(7)
In short, insubordinate sentences, such as (1a) and (3), are ambiguous between dependent and independent clause. An utterance like (5a) is not, if we use Abraham's argument about modal particles, and has been reanalyzed as an independent clause where the complementizer als blocks Verb-second. Future work could look at modal particles in Old and Middle English insubordinate clauses.
4 Pronouns
Cole (this volume) proposes that the Old English use of anaphoric personal pronouns and demonstratives for anaphoric use is very similar to that in Modern Dutch and German, namely the latter are predominantly topic-switching, as in (8).
-
(8)
She expands on earlier literature and provides an interesting picture of anaphora in West Saxon religious, legal and medical texts.
Cole reminds us that only (third-person) demonstratives can be modified by a relative clause, not third-person pronouns. Personal pronouns can be used reflexively and, in that function, are optionally modified by ‘self’. Nominative pronouns can have an intensifying ‘self’. She provides tables for the information structure of pronouns and demonstratives, namely that demonstratives are rare in discourse-old contexts. Nominatives with an intensifying ‘self’ generally refer to discourse topics.
A generative model might ask certain questions about the features of the demonstratives and also add the role of null subjects, which Cole doesn't do. Corpus-based research (e.g. Walkden Reference Walkden2013, Reference Walkden2014) shows that only third-person null subjects can be null in Old English, also in some of the texts that Cole considers. Verbal agreement could license these null subjects since it distinguishes between third-person singular and plural. Coupling that with insights regarding the structural make-up of the different pronouns, we can see that Old English has three kinds of third-person subject: the null subject; the third-person h-pronoun, with the same person features as the null subject; and the demonstrative, with person and definiteness features. According to van Gelderen (Reference Gelderen and Lohndal2013), this system is reanalyzed around 1200 into one that no longer uses the demonstrative or the null subject but strengthens its h-pronoun, which loses its reflexive function. The changes that are most relevant are the shift towards the use of demonstratives before a noun, i.e. the introduction of an article, and the introduction of new personal pronouns for which, as Cole (Reference Cole2014) has shown, it isn't clear whether language contact is responsible or it is due to internal change. I have listed them in table 1.
Changes in Late Old English
Assuming a minimalist analysis (e.g. Chomsky Reference Chomsky1995, Reference Chomsky, Sauerland and Gärtner2007), we can see articles as clear probes located in D, as in (9a), with uninterpretable features probing the phi-features of the noun. Since the has [u-phi], it cannot occur on its own (*I saw the). The demonstrative can occur on its own (I saw that) and I therefore assume it has interpretable person features ([i-ps]), as in (9b), or interpretable person and deictic features ([i-loc]), as shown in (9c). Number in (9bc) has to be checked, as I have indicated, but the exact probe on the demonstrative still has to be determined. That probe might be [u-#] in the NumP.
-
(9)
As for the difference between first- and second-person pronouns on the one hand and third-person ones on the other, Kinn (Reference Kinn2016) claims that a structural constraint on the actual pronoun explains the person split in Old Norwegian null-subjects: third-person pronouns have less structure, because they lack a D and are merely Phi, and can therefore be left unpronounced. This system is applicable to Old English as well with the added licensing of the Phi through agreement. So, Old English demonstratives are DPs but third-person h-pronouns are not. This is different from what I claim in van Gelderen (Reference Gelderen2000: 100–1) on the basis of sequences of personal pronouns and names. There, I assume that sentences such as (10) show that the third person can be a D, but in fact, other pronouns do this as well and do so with common nouns, as in (11), rather than only with proper names. This picture is more compatible with first and second person being definiteness markers, as Kinn (Reference Kinn2016) claims for Old Norse, and the third-person cases being appositives.
-
(10)
-
(11)
One caveat is in order about first- and second-person reflexives which remain in use till Middle English. These are object pronouns and would function differently. Again, this remains for further study.
In short, Cole contributes to the pragmatics of demonstratives and to their occurrence. A generative perspective would consider this in terms of structural changes, such as those in (9).
5 Conclusion
In this coda, I have added a few remarks on the volume from a generative perspective. First, I have sketched two different generative responses to the UP. Then, like Evans and Traugott, I argue that insubordination can be thought of as originating from ellipsis and as having reanalyzed its C. Finally, I review some approaches to changes in demonstratives and pronouns using features.
