Grammaticalization, the historical emergence of new items with grammatical function from earlier lexical items, is generally considered to be a unidirectional process. Much recent interest has, however, focused on degrammaticalization... more
Grammaticalization, the historical emergence of new items with grammatical function from earlier lexical items, is generally considered to be a unidirectional process. Much recent interest has, however, focused on degrammaticalization changes that run counter to this general direction. This paper considers three cases of degrammaticalization from Bulgarian and Welsh, involving shifts from pronoun to noun, and from preposition to verb. These cases exhibit a common set of properties, such as the central role played by syntactic reanalysis and pragmatic inferencing, that justify viewing them as examples of a new type of degrammaticalization. Degrammaticalization via syntactic reanalysis appears to be cross-linguistically rare, because it is constrained by two factors: the requirement that the item undergoing degrammaticalization should have become grammatically or semantically isolated; and the requirement that it should match a possible morphological pattern for the lexical category that it is to join.
This paper examines the evidence from Middle Welsh for the emergence of the Modern Welsh marker of clausal negation ddim. It considers cases where the indefinite pronoun dim ‘anything’ appears not to be an argument of its verb, and... more
This paper examines the evidence from Middle Welsh for the emergence of the Modern Welsh marker of clausal negation ddim. It considers cases where the indefinite pronoun dim ‘anything’ appears not to be an argument of its verb, and therefore can be considered to be a ‘pseudoargument’ with an adverbial function ‘at all’. Isolated examples of this use can be found from the thirteenth century onwards, with robust attestation in a number of texts from the fourteenth century. It is argued that the pattern of attestation is consistent with pseudoargument dim being a late-thirteenth-century innovation, found only in texts first committed to writing from that time onwards. The primary factor in its emergence is argued to be the potential ambiguity in the analysis of optionally transitive verbs. Syntactic differences between Middle Welsh pseudoargument dim and the present-day Welsh clausal negation marker suggest that the former is not the direct ancestor of the later, but rather Middle Welsh pseudoargument dim instead survives in semi-fossilised form as a sentence-final adverbial today.
Verb-second orders are only found in the Middle Welsh period: Old and Modern Welsh mainly exhibit verb-initial patterns. In this paper I show how these V2 orders developed by carefully reconstructing their syntactic history from earlier... more
Verb-second orders are only found in the Middle Welsh period: Old and Modern Welsh mainly exhibit verb-initial patterns. In this paper I show how these V2 orders developed by carefully reconstructing their syntactic history from earlier patterns with hanging topics and focussed cleft constructions in Old Welsh and related Celtic languages. I provide a syntactic reconstruction of the V2 structures with preverbal functional particles a and y. These C-particles played a pivotal role in relative clauses as well and can be traced back to pronominal elements in Proto-British, the predecessor of Welsh, Breton and Cornish (cf. Schrijver 1997). I argue that these relative particles in the C-head are the result of Spec-to-Head reanalysis of pronominal phrases and that a similar reanalysis of the adverbial phrase *ed ‘thus’ yielded CVSO orders. During the next stage, the relative clauses in clefts were reanalysed as matrix clauses with V2 order and a process of rebracketing integrated hanging and dislocated topics followed by CVSO into the matrix CP with V2 order as well. These developments resulted in the extension of IS functions for the sentence-initial constituents (beyond original contrastive focus) and a generalised Edge Feature on the C-head.
This paper focuses on verb-second main clauses in Middle Welsh and their interaction with co-ordination. It argues that the only empirically-adequate analysis of Middle Welsh coordination patterns requires that the conjuncts be analysed... more
This paper focuses on verb-second main clauses in Middle Welsh and their interaction with co-ordination. It argues that the only empirically-adequate analysis of Middle Welsh coordination patterns requires that the conjuncts be analysed as full CP-clauses. Apparent gaps in those clauses require the postulation of an empty category. Two candidates for such a category are considered, pro and an empty operator. The former analysis is rejected because the empty category fails to obey the licensing conditions on pro in Middle Welsh. A discourse-licensed empty operator on the other hand provides a good account of the environments in which these gaps are found. Cross-linguistic and internal diachronic evidence show that such an analysis of Middle Welsh as having an empty operator in coordinate structures fits well with a typology of such structures according to which languages may adopt and move historically between a highly restrictive ‘syntactic’ system of coordination and a much less restrictive ‘pragmatic’ one.
A frequent development is for languages to replace a preverbal marker of negation with a newly innovated postverbal one (‘Jespersen’s Cycle’), as with French ne verb > ne verb (pas) > ne verb pas > (ne) verb pas. Although it has long been... more
A frequent development is for languages to replace a preverbal marker of negation with a newly innovated postverbal one (‘Jespersen’s Cycle’), as with French ne verb > ne verb (pas) > ne verb pas > (ne) verb pas. Although it has long been known that Welsh has undergone Jespersen’s Cycle, the Welsh development has never been traced in any great detail. This chapter provides documentation of the various stages of Jespersen’s Cycle in Welsh and an account within minimalism. It shows that the postverbal negative marker (d)dim developed via a series of staged reanalyses, from a noun ‘small thing’ to become an indefinite pronoun ‘anything’, itself a negative-polarity item. This was then reanalysed as a negative polarity adverb ‘at all’, which became an optional then compulsory marker of negation. I interpret these stages in minimalist terms as a change in the features on (d)dim: first from noun (N- head) to indefinite pronoun (D-head marked as a negative-polarity item); then to adverb (AP marked as a negative-polarity item); then to a specifier of NegP with an uninterpretable Pol [Neg] feature; and finally as a specifier of NegP with an interpretable Pol [Neg] feature. The Welsh evidence also bears on broader issues of the nature of syntactic change and grammaticalisation. This chapter shows that the historical development can be conceived of as a series of distinct reanalyses, with (d)dim successively splitting into two items with distinct syntactic properties at each stage.
In the article I discuss the quite frequent Middle Welsh construction exemplified by y wreic uwyaf a garaf "the woman I love most". I propose that it can best be understood as the result of some processes of reanalysis and blending of... more
In the article I discuss the quite frequent Middle Welsh construction exemplified by y wreic uwyaf a garaf "the woman I love most". I propose that it can best be understood as the result of some processes of reanalysis and blending of different source constructions.
The Middle Welsh abnormal sentence, with its subject-first structure, has long been noted as anomalous amongst traditionally verb-initial Celtic languages (Evans 1964). Curiously, this subject first structure without emphasis indicative... more
The Middle Welsh abnormal sentence, with its subject-first structure, has long been noted as anomalous amongst traditionally verb-initial Celtic languages (Evans 1964). Curiously, this subject first structure without emphasis indicative topicalization is unique to Middle Welsh prose (Mac Cana 1973, Fife and King, 1991), and has subsequently disappeared in the modern language. However, it still begs the question of why and how this formation developed and why it is such an anomaly.
The development of the abnormal sentence seems to have happened concurrently with the Norman invasion of Wales in 1067, a time of heightened internal political strife within the Welsh borders and the impetus for the introduction of French as a prestige language among the elite of England and Wales (Mac Cana 1979). The theory of contact induced language change indicates that linguistic shift is driven by societal factors rather than simply linguistic ones, meaning that the speakers of a language cannot simply decide that a specific feature of a contact language would work well within the confines of the grammatical structure of the receptive language and expect it to subsequently appear in the language as a whole (Thomason 2001). An examination of a literary exchange between Medieval Wales and Norman France indicates a deep cultural impact on Medieval Welsh literature, which lead to a noticeable shift in the linguistic and structural content of the literary products of that period.
This background provided the foundation to examine the development of the abnormal sentence through a sociolinguistic lens. I used an analysis of the underlying reason for the creation of the Mabinogion (Fulton 2005) and the Celtic substrate in many French poetic romances as examples of the power Normans had on the literary culture of Medieval Wales (Short 2003). Expanding on the claim that the abnormal sentence is exclusively found in the literature (Mac Cana 1979, Fife and King 1991), I posit that the development is a move to emulate the subject first nature of the prestige language, Norman French.