Universality and the evolution of aspectual adverbials

Abstract

Templates. We argue, focussing specifically on aspectual adverbials including again, still, then, that some linguistic elements form a network and are best described as having a basic templatic definition. The template allows variation in linguistic realizations and accounts for patterns of polysemy. We show that realizations can be derived from a variety of related items historically, and that crucially the earlier forms lack any templatic component. Templatic meaning for aspectual adverbials. There are a variety of items which are amenable to a templatic treatment. Consider aspectual adverbials. Löbner 1989 and Krifka 2000 propose a system of aspectual adverbials that are related by inner and outer negation: The proposal above addresses temporal interpretation (Fred is still / already sleeping). Several adverbials, most notably still and its equivalents have usages that go beyond temporal interpretation. As dicussed by Beck 2018 a.o., German noch ‘still’ and its equivalents permit a variable range of other readings, briefly illustrated below. Reineland in still in Canada (spatial, also for already) .̱ He felt sick, but he still decided to stay (concessive) This dress is still expensive (marginal) While some readings, including , merely involve a scale distinct from the temporal scale, others (e.g. concessive, marginal involve a more significant difference (see Beck 2018, a.o.). Thus still, noch can be said to involve polysemy. Polysemy is not restricted to still. The aspectual adverbials below include repetitives, ordering then (And then he left) and its inverse. A number of items that are distinct in English have identical realizations:

	‘before			temp.		‘(not)	‘(not)
	that’	‘then’	‘again’	‘still’	‘already’	yet’	anymore’
Hindi ab tak, Nepali ahile samma				X	(X)	X
Hindi phir, Nepali pheri		X	X
Romanian mai			X	X	X		X
Italian ancora			X	X	X	X
Jamaican patois aredi					X	X
Spanish ya					X		X
Spanish todavia				X		X
Hungarian még	X			X		X
German noch	X			X		X

The polysemy of the various aspectual adverbs suggests a single underlying definition; the shared definition allows different meanings to be realized by the same element. At the same time, there is no expectation that there is a common element for all uses, as shown by the empty cells and by the first row of the table. Darker coloured cells indicate patterns which are unexpected on a Löbner-style approach. Template. We propose that the underlying meaning is a template. Here x, x′ are scalar entities (times, degrees, etc) such that x′ precedes x on scale S; P, Q are saturated predicates except for the arguments indicated, F**A is a set of focus alternatives to P(x,…) which differ in the elements under focus (times, degrees, or subconstituents). The aspectual adverbials can differ in the identity of the scale; the type of argument; whether P and Q are identical; the identity of the focused element (e.g. for repetitives (again) the time argument must be focused); and the relation (e.g. immediate precedence for still and already, simple precedence for again, ordering then and before that). . .[template] = λ**x_Sλeλ**P : ∃x_S^*∃e^*∃Q[Q(e^*,x^*,…)&Q(e^*,x^*,…)∈F**A(P(e,x,…))&x^*R**x].P(e,x,…) .̱ = λ**t_Tλeλ**P : ∃t_T^*∃e^*∃Q[Q(e^*,t^*,…)&Q(e^*,t^*,…)∈F**A(P(e,t,…))&t^*∝t].P(e,t,…) The template offers a unique underlying definition. Whenever a surface form has different aspectual interpretations, that form is more general (the unmarked form), and other aspectual adverbs are more specific, but based on the template. Various interpretations of still (e.g. marginal and concessive readings) also fit the template. Already. We reanalyze already as being similar to still, but involving an inverted time scale (so x^* follows rather than precededs x). Thus the truth of P at a preceding time is presupposed for still, but ¬P implicated for already. Focus sensitivity. The template appeals to focus alternatives to determine predicates in the presupposition. Beck 2018 argues that still, noch are not focus-sensitive. Rather, they tend to occur in environments where focus is likely to occur independently. Beck 2018 points out unlike nur/only and auch/also, noch/still does not appear to be able to associate with focus alternatives within syntactic islands. We provide examples showing that this association is possible for examples where a scale is clearly available for ranking alternatives. Historical development Crosslinguistically, aspectual adverbials often have similar etymologies. At the same time, the earlier usage lacks a templatic component, which is expected if the template is available for functional or semi-functional elements. Repetitives. Historically, we can observe a number of interesting trends in repetitives, including a recurrent pattern of elements which develop ultimately from words meaning “hinder-part” to the adverbial “back” and thence to “again”. This includes Kutchi Gujarati pacho “again (repet.& restit.) & back” (see Patel-Grosz & Beck 2014) < OIA.*paśca- “hinder part” [Turner 1966: #7990] as well as English back itself. In the case of English again, this word originally meant “back, in the opposite direction” (=OE ongean): . “He sceaf þa mid ðam scylde, ðæt se sceaft tobærst, and þæt spere sprengde, þæt hit sprang ongean.” [“He shoved then with shield so the shaft burst — the spear broke and sprang back.”](Battle of Maldon 137) Old English eft (cognate with modern English after and aft) also exhibits polysemy analysable as underspecification similar to that found in Hindi phir and Nepali pheri in their polysemous senses of “then (=after that)” and “again”: . Efterward me ssel þerne mete eft chyewe ase þe oxe þet… “Afterward one shall chew this food again like the ox that…[repetitive reading] . þone mon eft on Cent forbærnde. “That man was afterwards burned in Kent.” [*AS Chron.*685 (Parker)] In Hungarian, a non-IE language, the repetitive forms megint, ismét are etymologically related to meg ‘back’. Repetitives can also be derived from an expression meaning ‘new’ (English anew, afresh, Spanish de nuevo, Hungarian újra ‘new+onto’), though English anew carries additional pragmatics not found in de nuevo. Hindi vāpas “back” on the other hand has not (yet) developed any repetitive senses, and represents loanword from Persian, with the pās part being cognate with Old Indo-Aryan *paśca- “hinder-part” [Platts 1884:1171] (and thus is cognate ultimately with Kutchi Gujarati pacho). Hindi phir “then, again”, Nepali pheri is related to Hindi phirnā “to turn”, which derives from a reconstructed Old Indo-Aryan *phirati “moves, wanders, turns”, cp.Prakrit phiraï “goes, returns” (Turner 1966: #9078). In early Indo-Aryan we find Sanskrit púnar, ultimately underlying Nepali pani (Nepali pani derives from Sanskrit punar api “even again; again too; moreover; also” (Turner 1966:#8274)). Púnar, itself an aspectual adverbial, is of interest due to being more underspecified than many other examples, polysemous between “back; again; further; (concessive) still”. Still. English still provides an instructive view into historical developments affecting aspectual adverbials. Originally meaning “motionless” (still possible in Mod Eng), it has come in Modern English to have a great range of senses (cf.Ippolito 2007, Beck 2018). From original sense, it developed in the 14th century an additional possible meaning “always” (archaic by the 19th-c.), as in Thus haue I prov’d Tobacco good or ill; Good, if rare taken; Bad, if taken still. [1617 R.Brathwait Smoaking Age] Only from the 16th-century do we find the modern day temporal still sense, e.g.For as you were when first your eye I eyde, Such seemes your beautie still. [1609 Shakespeare Sonnets civ.sig.G2v] While the comparative sense appears consistently only from the 18th-c.: The Woodmongers Abuse..of a former Charter leaves still less Reason to fear they shou’d succeed. Concessive still likewise only appears from the 18th-century: ‘Tis true, St. Giles’s buried two and thirty, but still as there was but one of the Plague, People began to be easy. [1722 D.Defoe Jrnl.Plague Year 7] The reanalysis from “motionless” to “always” results in a temporally-associated adverb, whose denotation we can roughly formalise as: . λ**P∀ relevant times t′.P(t′) Note that unlike the aspectual adverbials this does not involve a presuppositional component. Thus the later 17th-c.re-analysis as a temporal aspectual adverbial still involves a significant change in semantic value. Thus in all of these examined cases of the development of lexical items into aspectual adverbials, a major semantic shift is involved. None of these involve a sort of gradual semantic change, but rather ‘catastrophic’ reanalyses, whose frequent and crosslinguistic occurrence strongly points to the templatic aspectual adverbial being a universally-accessible semantic chunk. Extensions. We discuss another possible use of templates, personal and demonstrative pronouns. We also explore the relationship between templates and the standard notion underspecification (Patel-Grosz and Grosz 2017 treat pronouns as involving underspecification, where demonstratives have more structure than personal pronouns). Templates give rise to a type of underspecification where individual lexical items can (a) encode particular choices for parts of the definition such as the identity of the scale involved or (b) have a more general definition, which permits a unique lexical entry with a variety of lexical meanings.