This paper reconstructs the pronunciation of the English Romantic poet John Keats — and in particular the likely original sound of one of his final poems, Bright Star — just over two hundred years after his death (1821). Keats makes a particularly fascinating phonological study due to Lockhart’s (1818) denunciation of him as a ‘Cockney poet’. I use evidence from contemporary pronouncing dictionaries and several other treatises and orthoepical works from the early 19th century to narrow down likely pronunciation variants in 1819. I then employ evidence from Keats’s own rhyme schemes (more concrete) and biographical details relating to his location, education, and social and literary milieu (more speculative) to construct a likely rendition of Bright Star in the poet’s own voice.
This paper presents the results of a study on the possible diachronic pathways that lead to so-called rhotic or retroflex vowels. Based on a historical-comparative examination of several unrelated areal clusters and individual languages which contain retroflex vowels, in particular in the Hindu Kush area, China and California, a variety of possible sources for retroflex vowels could be identified. Ultimately these can be reduced to three separate in-paths with two different underlying mechanisms.
The paper presents a pilot study investigating the extent of sense sharing between loanwords in Middle English and their etyma in French and Latin, undertaken as part of a new three-year project. Fifty loanwords have been examined, with all senses in the borrowing and donor language(s) recorded using the major historical dictionaries. These have been standardised and mapped onto a framework using the categories of the Historical Thesaurus of English ( Kay et al. 2022 ) to allow comparison. Our main results show that polysemy is nearly always mirrored but complete semantic overlap is rare; that senses unique to Middle English are common but short-lived; and that figurative senses are almost never borrowed into Middle English without the original literal senses.
On the general level, the paper evaluates analytical methods for combined linguistic, archaeological and historical research. Specifically, the paper takes up the controversially discussed, but never fully resolved, history of the g > ɣ > h lenition in Bohemia, a core migration area of medieval Slavic, investigates the distribution of lenition in space and time and establishes a clear distinction between the phonetic phase of lenition, preceding yer loss, and the phase of phonological reanalysis following it. Kernel Density Estimation points to the beginning of the phonological phase between the archaeological area of Prague-type ceramics and Germania Slavica with northeastern Bavaria, from where it spread to neighboring areas. Potential indicative value of toponymy in Winden - and windisch - is evaluated relative to a hypothesis about language contacts.
This paper examines the correlation between the position of resumptive demonstratives and personal pronouns in left-dislocation and their functions in the discourse. It shows that these structures typically introduce new topics, hence two thirds show no antecedent and topic continuity. When antecedents are present, they are fairly common in percentage terms in Ælfric’s Catholic Homilies I. They are non-nominative and linked with object demonstratives occurring high. By contrast, personal pronouns show a more general tendency to persist as topics both in Ælfric’s Catholic Homilies I and in Cura Pastoralis . Moreover, their function does not seem to be directly linked with low position.
Predicate-specific oblique subjects have emerged throughout the history of Icelandic. The novel contribution of this paper is spelling out the precise mechanisms of the changes. We focus on three general processes: Oblique-Case Substitution (OCS), Case-Preserving Anticausativization (CPA) and Argument Swapping (ARS). OCS involves morphosyntactic leveling targeting experiencer predicates, causing nominative experiencer subjects to be replaced by oblique ones. CPA is a special kind of anticausativization, where the oblique case of the object of a transitive verb matches the case of the subject of the corresponding anticausative, resulting in an oblique-subject construction. Finally, ARS involves a reanalysis of an object as a subject. The oblique subject in dative-nominative structures, in particular with st -predicates (‘middles’), arose through ARS from earlier nominative-dative structures, motivated by the animacy of the dative.
The last broad assessment of Algonquian internal relationships was over a quarter century ago, by Goddard (1994) , and most research since adopts his view that the relationships almost entirely reflect an areal cline rather than subgroups, though that consensus may be beginning to break down. Such work, though, has seldom paid explicit attention to basic principles of subgrouping, including the foundational one of synapomorphy. Using a character-based approach to phonological history, I explore evidence for a new subgroup involving languages of the Great Plains: Cheyenne, Arapaho, and Atsina. This is a small step toward a fuller reorganization of the Algonquian family tree.
This paper addresses a classic problem in historical linguistics — ‘drift’, i.e., identical or similar changes occurring in related languages after separation from their common ancestor — from the perspective of theoretical historical phonology. I endorse Joseph’s (2006) view of the basic mechanism of drift: it arises when synchronic variation, inherited from the proto-language, narrows in range to produce similar outcomes. This paper elaborates on this proposal by identifying the mechanism that ensures inherited variation is narrowed down in similar ways, framing it within the theory of the life cycle of phonological processes. As proposed by Sapir (1921) , ‘drift’ is often criticized for invoking the spectre of unexplained teleological factors (‘submerged form-feeling’) driving change. It is reasonable to ask whether parallel developments involved in ‘drift’ are ever distinct from ‘ordinary’ change. After all, change is subject not only to random contingencies but also to universal forces, which can conspire to produce parallel developments. This is especially pertinent for phonological change, widely agreed to be grounded in substantive biases related to the human articulatory and auditory systems ( Blevins 2004 ). If we wish to disentangle the mechanism that might account specifically for drift from such universal tendencies, it is particularly valuable to consider cases of drift with no ready universal explanation — those that involve typologically unusual phenomena. Here, I discuss the consonant gradation systems of the Uralic languages, with a particular focus on the Finnic and Sámi branches. I offer a reconstruction of its rise and development within the life-cycle framework. In doing so, I highlight both the explanatory power of the theory in accounting for Sapirean drift and its heuristic value in opening up new approaches to long-standing empirical controversies.
Approximative adverbs such as almost, nearly, approximately and circa have received relatively little attention in historical linguistics. This paper documents how approximative expressions become more frequent as well as more lexically diverse in two European languages, to wit English and Dutch. We argue that this change is not simply a language-internal matter, comparable to, say, the rise of modal verbs in Germanic, but reflects a cultural change as well, one having to do with an increasing concern with precision when it comes to numbers and measures (such as weights or temperatures). We make a distinction among various classes of modified expressions, including numerals, quantifiers and absolute adjectives, and irrealis uses ( I almost made a mistake ), and show how diachronic developments point toward increasing specialization.
This paper investigates the development of sentence adverbs (SAdvs) and modal particles (MPs) by especially focusing on the German language. By exploring their common grammaticalization paths, we claim that these are much more restricted than previously assumed. In particular, we argue that two pathways seem to play a major — if not exclusive — role, namely one starting from lower adverbs and one from parenthetical clauses, and that syntactic ambiguity in the middle field plays a fundamental role in this process. In addition, we show that there is no evidence of secondary grammaticalization occurring within the class of sentence adverbs and modal particles.
The paper examines contact-induced morphosyntactic change in Swahili, where material which had historically been lost is ‘reintroduced’ through contact with closely related languages which have retained the original feature. The paper discusses three examples of these morphosyntactic ‘false cognates’: diminutive marking, habitual marking and demonstrative forms, and shows that if it were not from the evidence from different diachronic stages and varieties of Swahili, these forms could well be analysed as inherited from Proto-Bantu. The paper contributes to our understanding of the historical development of Swahili, patterns of variation found in Swahili and Bantu languages more widely, as well as the importance of comparative evidence for the unravelling of historical and contemporary relations between closely related linguistic varieties.
I argue that Contrastive Hierarchy Theory (CHT) makes it possible to implement Roman Jakobson’s 1931 program for diachronic phonology. CHT computes segmental contrasts by language-particular feature hierarchies and incorporates the Contrastivist Hypothesis, which holds that only contrastive feature specifications can be active in the phonology. This empirical hypothesis is susceptible to be falsified by the ‘Oops, I Need That’ Problem. I review two case studies: the evolution of the modern Manchu vowel systems from Classical Manchu, and changes in i -umlaut from early Germanic to Old English. Both cases adhere to the Contrastivist Hypothesis and show connections between diachronic change and synchronic analysis as proposed by Jakobson, and between contrast and activity as predicted by CHT.
Examining the distribution of suppletive stems over inflectional paradigms, a constraint on their complexity, call it Crossover Constraint, is shown to be untenable. There are paradigms, in some languages at some stage of their development, where suppletive distributions are maximally complex — like one stem of a verb being used for 1st Person Singular and 3rd Person Plural and another stem for all other person and number forms, which in terms of paradigm geometry is a crossover distribution. Rather than unceremoniously dumping this universal it is argued that constraints on states must be distinguished from constraints on transitions. There are two main routes to suppletion, the harnessing of distinct stems for one lexeme and the phonological separation of once unitary stems; their respective results are synchronically indistinguishable. When stems are combined, they will divide up paradigms in ways that follow morphological patterns, which outlaws crossovers, while phonological differentiations are under no obligation to respect morphological groundplans.
This paper explores the differential marking of objects (DOM) in two southern early Italo- Romance vernaculars, namely old Sicilian and old Neapolitan, and in old Sardinian texts from two different areas, Logudoro and Arborea. We investigate the constraints on the marking of objects, whether semantic, encoding the Individuation Hierarchy, syntactic, determined by verbal valency and/or the position of the O argument in relation to the verb, or pragmatic, resulting from the topicality of the object. We show that, unlike in old Neapolitan and old Sicilian, in old Sardinian Topicality and the Person Hierarchy do not appear to play a role in the occurrence of the object marker a ( d ).
The South Slavic dialect continuum formed by Serbian and Bulgarian provides a fine environment to investigate case loss. Different points on this continuum show varying degrees of case loss: standard Serbian represents a conservative variety with six cases, and standard Bulgarian represents an innovative variety with no nominal case inflection. The continuum between these two extremes shows a transition from larger to smaller case systems. We focus on three functions: instrument, transport, and trajectory, which were originally conveyed by the non-prepositional instrumental, and model what happens to them when the case form is lost. We demonstrate how they lose their original unity and are distributed over different constructions.
The paper discusses a shift in the number system of Proto-Tanzanian Cushitic from the word for “three” to mean “four”, and the one for “two” to mean “three”. This shift was brought about by contact between an incoming group of northern Cushitic speakers and the already present group of Tanzanian Cushitic speakers due to similarity in form between the words for “four” in Tanzanian Cushitic and “three” for in the incoming group. This shift pulled the word for “two” to fill the gap of the old “three” and triggered the innovation of a new internally derived word for “two”. This scenario provides a more intricate view on the Cushitic migration from Ethiopia to Tanzania. Kw’adza joins the rest of Tanzanian Cushitic undergoing this complex and unique set of shifts, but Aasa does not.
We assess the depictive and descriptive character of sound ideophones and their historical evolution in Emai of West Africa. Their initial depictive nature remains evident in phonological similarity to extant verbs with related meanings and retention of a perfective aspect mark. In their descriptive mode ideophones appear as synchronic adverbs that impact tonal marking within a clause. To account for their evolution from expressive acategorial forms to verbs, then to adverbs we rely on two constructions of cause-effect, one being resultative and the other aspectual extent. Via analogy each interacts with clausal expression of ideophonic forms leading to their recategorization and morphosyntactic integration into the contemporary grammatical system, particularly as it applies to information flow in traditional narrative discourse.
Inspired by discussions of privatives as a ‘previously unidentified source’ for verbal negation in main clauses, with evidence from Arawakan languages ( Michael 2014 ; Rybka & Michael 2019 ), this paper builds on earlier observations on the relation of privative nominal suffixes to verbal negative inflections in Australian languages ( Dixon 1980 ; Phillips 2023 ). Using data from selected languages, I attempt to trace the diachronic processes involved, specifically the mechanisms (reanalysis and extension) and motivations (especially pragmatic) behind the insubordinations which lead via somewhat different pathways either to indicative negatives (standard negation) or imperative negatives (prohibitives). I hope thereby to have contributed to a fuller appreciation of insubordination as factor in diachronic change (cf. Evans 2007 ; Mithun 2008 ; Cristofaro 2017 ).
This study presents evidence that dissimilatory constraints act as a discriminating factor in the selection of models for analogical change within inflectional paradigms. The study compares the lexical extension of root-final /ɡ/ in preterite forms in two groups of Occitan varieties: one group with established preterite desinences in -/εɡ/-, where introducing /ɡ/-final roots creates a sequence -/ɡεɡ/-; and another group with established preterite desinences in -/εɾ/-, where introducing /ɡ/-final roots creates a sequence -/ɡεɾ/-. Incidence of root-final /ɡ/ is lower in the velar desinence group than in the rhotic desinence group, indicating that dissimilatory pressure contributes to speaker choice between available roots, alongside other factors such as existing paradigmatic relationships and lexical gang effects.
Although there are many studies on translational language, they present contradictory results. This study investigates the syntactic and typological properties of translational language by adopting two main indices of dependency grammar with an aim to reveal one probabilistic translation regularity: the law of interference. After that, machine learning methods were adopted to classify translated and non-translated texts, and to further verify translation regularity on larger datasets by checking classification performance. The results show that (1) the MDD of translated English abstracts is significantly longer than that of non-translated English abstracts; (2) there are more head-final and fewer head-initial dependencies in translated than in non-translated English abstracts; and (3) compared with K -nearest Neighbor and Support Vector Machine, Random Forest achieves satisfying classification performance. These findings show the interference effect in translation and suggest that the framework of dependency grammar is helpful to reveal translation regularity.
