ABSTRACT. I use findings from recent research in Australian languages to discuss three fundamental areas of historical linguistics: 1) exceptionalism and universalism in language change; 2) the role of phylogenetics and evolutionary models in studying language change, and 3) the role of non-linguistic factors in describing and explaining language change. Australian languages have long been used as testing grounds for various claims of 'exceptionality', from the lexicon to phonology and syntax, to claims of social relationships and language change (for an overview, see Bowern 2010). In some ways, the prototypical Australian language community is very different from those for which core principles of historical linguistics were originally established: small, mobile, multilingual populations. Yet, as we have seen again and again over the last 20 years, Australian languages are describable by the same descriptive and diachronic theoretical vocabulary that linguists use for other parts of the world. Indeed, Australian languages are an excellent natural laboratory for studying language change, because of the large number of languages, the relative uniformity in populations, and the wide differences in language contact phenomena.

In the second part of the talk, I turn to phylogenetics (and other computational or quantitative approaches to historical linguistics). Recent work has discussed both the appropriateness of algorithmic approaches to historical problems, and specifically biological (or evolutionary) approaches to linguistic data. I use data from Australian languages to make the case that an evolutionary approach to language change has advantages for how we model language more generally; in short, that it leads to a more consistent approach to our data and a more tenable set of assumptions about how languages change. I discuss criticisms of phylogenetic methods and show how some problems can be overcome.

I conclude the talk with some remarks on the role of non-linguistic factors in language change. An evolutionary approach to language lets us see language as something that speakers do: language changes not of its own accord, but (most broadly) as a result of how speakers interact with one another. Evolutionary approaches make clear the differences between features of language that are the properties of individuals, and language as a system, that changes as features fluctuate across populations.

ABSTRACT. The current study presented a constructional and diffusional account for the formation of a Chinese connective biéshuō deriving from the prohibitive expression with the meaning of ‘don’t say’. Specifically, this diffusional change was found to be intimately related to the diachronic change of each element (i.e. a prohibitive marker and a SAY verb) constituting the connective. The findings of the current study lend support to the usage-based view that language is a dynamic and organized system based on similarities (e.g. Bybee 2010).

ABSTRACT. Sinitic languages (i.e. so-called “Chinese dialects”) are genetically and areally closely related. It is often observed that grammatical innovations are extensively shared by Sinitic but constructions of a specific grammatical function can be derived from different sources and paths. In Sinitic, a semi-schematic verb complement construction (VCC), usually termed as the “potential complement construction” (PCC, see Wu and He, 2015), is used to encode “situational possibility” (van der Auwera and Ammann, 2005). But PCC varies in historical sources, grammaticalization paths, substantive elements, morphosyntactic bondedness, and further word order specifications across Sinitic. This paper aims to investigate the causes and constraints of constructional change and variation in an areal perspective by examining the history of PCC in Min, a Sinitic language mainly spoken in southeastern China. I will first outline the morphosyntactic variations of the Min PCC, and then trace its developments using data drawn from historical texts. Afterwards, the interaction of multiple causal factors in the evolution of the Min PCC will be further explored against the historical backdrop of Sinitic grammars. Unlike in other Sinitic languages, PCC in Min is characteristically substantiated by a pair of epistemic modals 會e7 ‘can’ and袂be7 ‘cannot’ in the generalized form “V-會e7/袂be7-C” (Lien, 2015). As attested in Min texts, the Min PCC is derived from a kind of serial verb construction (SVC) with the earliest documented usages only in its proto negative form “V-(O)-袂be7-C” before its proto positive form “V-(O)-會e7-C” analogically emerged in the early 19th century. Meanwhile, like in southern Sinitic, PCC in Min is also in an ongoing coalescence process whereby “V-會e7/袂be7-C” becomes fused as a VCC or verb compound (Li and Thompson, 1981:54-57). This process can be synchronically attested in the morphosyntactic variations of PCC regarding the position of its object in Min dialects (Table 1).

Table 1. Morphosyntactic variations of PCC regarding the position of its object in Min dialects _________________Proto_______________Loan__________________Coalesced Positive_________V-O-會e7-C________V-會e7-O-C___________V-會e7–C-O Negative________V-O-袂be7-C_______V-袂be7-O-C_________V-袂 be7-C-O

I will argue that the proto forms of the Min PCC are recruited as a subtype of the Sinitic PCC, and then reanalysed and constrained as a VCC for two main reasons, one being its form-function parallel with the major type of the Sinitic PCC (i.e. “V-得de/不bu-C-(O)”), and the other being a Sinitic-wide coalescence tendency imposed on the verb and its resultative element. The loan forms, rather than representing an intermediate stage, are purely contact-induced by the neighboring Sinitic languages. To conclude, the developments of the Min PCC, as a subtype of the Sinitic PCC, are subject to both internal and external causal factors.

References: Li, Charcles N., Thompson, Sandra A. 1981. Mandarin Chinese: A Functional Reference Grammar. University of California Press, Berkely and Los Angeles. Lien, Chinfa. 2015. Min languages. In: Wang, William S-Y, Sun, Chaofen (Eds.), The Oxford Handbook of Chinese Linguistics. Oxford University Press, Oxford, pp. 160-172. van der Auwera, Johan, Ammann, Andreas. 2005. Situational possibility. In: Haspelmath, Martin, Dryer, Matthew S., Gil, David, Comrie, Bernard (Eds.), The World Atlas of Language Structure. Oxford University Press, Oxford, pp. 302-305. Wu, Fuxiang, He, Yancheng. 2015. Some typological characteristics of Mandarin Chinese syntax. In: Wang, William S-Y, Sun, Chaofen (Eds.), The Oxford Handbook of Chinese Linguistics. Oxford University Press, Oxford, pp. 379-392.

ABSTRACT. Linguistic construal of time lies at the center of language; it is also one of the cognitive foundations of culture. Construal of time categories can frame the situation semantics so as to either focus on boundedness (in the sense of Verkuyl 1972) or blend it out; it enables different conceptualizations of the inner temporal constituency of situations (in the sense of Comrie 1976) that constitute the category of aspect. In addition to aspect (as situation-internal time), there is also situation-external time, conceptualized as relatively ordered temporal intervals, yielding the category of tense. The essential similarity of aspect and tense lies in the temporal framing of situations by aspect and the relative ordering of these frames by tense. Historical developments have reflected this similarity in transfers of formatives (such as the -s- aorist ending becoming a marker of perfectivity in a part of early Indo-European) and constraints on aspect-tense formations. This workshop aims at going one step further than the usual research by analyzing the systematic and functional details of such processes in the context of the semantic, syntactic and pragmatic properties of the discussed languages. For the development of aspect, the following major strategies are usually mentioned: univerbation of preverbs to yield prefixes; stem modification (including reduplication e.g. in Indo-Aryan); and paraphrasis with verbs of movement, etc. However, all these strategies affect lexical aspect, whereas grammatical aspect as a (possibly hierarchically branching) binary temporal category does not seem to be captured by them. The workshop will discuss processes which can lead – and have led – to the development or simplification of aspect and tense systems, including their mutual interaction. Based on a clear definition of lexical aspect as a property of the predicate and of grammatical aspect as a temporal framing of the situation that overlays lexical aspect (cf. Gvozdanović 2012) with a propensity to participate in higher language levels, this workshop will also discuss the influence of scoping sentential elements (e.g. temporal adverbials) on aspect and tense. Another possible

topic is temporal structures in narratives and their semantic, cognitive and informational properties. Special attention will be paid to typology. Even genetically closely-related languages exhibit differences in the construal of temporal categories, especially of aspect (e.g. Dickey 2000 concerning Slavic). The emergence of radically different aspect and tense systems, as a rule, never stands alone, but occurs as part of integrated space-time construals (cf. also Gvozdanović 2015) in typologically different systems (such as e.g. Celtic vs. Germanic or Romance). Development of temporal categories in a typological context will be another major focus. The workshop aims to cover developments in different language families, but its organization will hinge on categories and problems of identification. In this sense, it will contribute to theory and methodology in addition to language description. Language contact and typology will play a major role throughout, not only for understanding sources of the relevant developments, but also for gaining more general insights into the formative properties of language types. Initial references Bybee, Joan & Östen Dahl. 1989. The creation of tense and aspect systems in the languages of the world. Studies in language 13-1, 51-103. Binnick, Robert (ed). 2012. The Oxford Handbook of Tense and Aspect. Oxford: OUP. Comrie, Bernard. 1976. Aspect. Cambridge: CUP. Dickey, Stephen. 2000. Parameters of Slavic Aspect. Stanford: CSLI Publication Gvozdanović, Jadranka. 2012. Perfective and Imperfective Aspect. In: Binnick, Robert (ed.), The Oxford Handbook of Tense and Aspect, 781-802. Oxford: OUP. Gvozdanović, Jadranka 2015. Evaluating prehistoric and early historic linguistic contact. In: Haug, Dag T.T. (ed.), Historical Linguistics 2013: Selected papers from the 21 International Conference on Historical Linguistics, Oslo, 5-9 August, 2013, 89-108. Amsterdam: John Benjamins. Verkuyl, Henk J. 1972. On the Compositional Nature of the Aspects. Dordrecht: Reidel.

ABSTRACT. While our understanding of the creation of verbal tense–aspect–mood morphology has advanced considerably in the last few decades (Bybee & Dahl 1989; Heine & Kuteva 2002; Hopper & Traugott 2003, inter alia), we know much less about what happens to verbal morphology after it is created. That is, our understanding of how a rich TAM system will tend to change is still incomplete. This shortcoming is even more pronounced when it comes to Papuan languages, which often possess very rich TAM systems. 
This paper begins to fill that gap by describing how the TAM system has been reshaped in the Sogeram language family, focusing specifically on the complex tense system possessed by many Sogeram languages. The Sogeram languages are a group of ten Papuan languages spoken in Papua New Guinea, and Proto-Sogeram has been reconstructed in some detail (Daniels 2015). Many Sogeram languages possess complex tense systems that make several remoteness distinctions in the past, and fortunately many of the forms involved in the expression of different tense meanings are cognate. This has allowed for a fairly accurate reconstruction of the tense system of Proto-Sogeram, which possessed five past tenses (one of which was also used for events with present time reference) as well as a future tense. This detailed reconstruction allows us to examine how the system has behaved diachronically. 
The paper presents an overview of these systemic changes, highlighting both the predominant patterns as well as the divergent developments. This perspective offers numerous noteworthy findings. In general, the tense with present time reference and the most remote past tense have been the most diachronically stable, perhaps due to their cognitive salience, while the future tense and the “middle” past tenses are less likely to be retained. This sometimes creates typologically unusual tenses, as with the Aisi immediate past, which can refer to events from both the day of the speech act and the day prior (Andersen & Roberts 1991). Tenses are not observed to change time reference with regard to one another—that is, tenses do not “jump over” one another. It is also interesting to note how new tense morphology is inserted into the existing system; a particularly surprising development has occurred in Sirva, where a new tense created with immediate-past morphology has remote-past time reference, even though the Proto-Sogeram remote past tense is retained (and given hesternal time reference). 
Because there has been so little work on the cross-linguistic development of complex tense systems, this paper offers a particularly valuable contribution to our understanding of language change in general, and of the diachrony of tense in particular. It also expands our knowledge of Papuan historical linguistics, a sorely under-studied topic. 

References 
Andersen, T. David & John R. Roberts. 1991. An exception to the hodiernal:non-hodiernal  distinction. Studies in Language 15(2). 295–299. doi:10.1075/sl.15.2.03and. 
Bybee, Joan & Östen Dahl. 1989. The creation of tense and aspect systems in the languages of  the world. Studies in Language 13(1). 51–103. 
Daniels, Don. 2015. A Reconstruction of Proto-Sogeram: Phonology, Lexicon, and  Morphosyntax. University of California, Santa Barbara Ph.D. dissertation. 
Heine, Bernd & Tania Kuteva. 2002. World Lexicon of Grammaticalization. Cambridge  University Press. 
Hopper, Paul J. & Elizabeth Closs Traugott. 2003. Grammaticalization. 2nd ed. Cambridge: Cambridge University Press.

ABSTRACT. The aim of this study is to illustrate the change of IRASSHARU in modern Japanese society. IRASSHARU was a common honorific form of the three verbs:IKU(to go), KURU (to come), and IRU(to stay). However, over the past few decades, the usage of the common form has not only declined, but separated into three individual honorific forms. Diversification in Japanese society could be one of the reasons why the language change occurred in the late 20th Century.

ABSTRACT. One of the intriguing facts about the emergence of Modern Hebrew is early formation of a distinct, relatively stable colloquial register only a few years after the speech community first started to form. The aim of this paper is to suggest that this may be accounted for by the incorporation of pre-existing linguistic habits of the first L2 speakers based on their former familiarity with Hebrew as a language of culture. This unique phenomenon, which seems to be peculiar to Hebrew due to its unconventional history, was not recognized by research so far. Based on an extensive examination of pre-modern non-belletristic texts, I present data that indicate that some of the most conspicuous features of colloquial usage as we know it today were not created after speech revival due to the natural dynamics of speech, as has been often assumed in research, but preceded it. The availability of a certain portion of non-canonical options in the linguistic inventory already accessible to prospective speakers enabled the swift formation of register differentiation, based not solely on processes of linguistic change, but at least to a certain measure also on continuity with pre-existing linguistic habits created within the written modality. The linguistic data available to L1 children included simultaneously contact-induced substrate phenomena, superstrate classical Hebrew phenomena, and a certain portion of phenomena incompatible with traditional Hebrew grammar that were created in writing throughout the history of Hebrew. L1 speakers could readily identify those features common in daily speech that were considered unacceptable by normative standards. While they didn’t necessarily reject such features, but adopted them while allocating to them a colloquial tint. This set of features, common to pre-modern texts and to contemporary colloquial usage, is attested in various areas of grammar and consists of a wide range of linguistic phenomena in each area. The presentation clusters the relevant phenomena in four groups: (1) grammatical agreement (e.g. the employment of the masculine singular verbal form in existential constructions, regardless of the gender or number of the subject); definiteness (e.g. the occurrence of the definite article prior to the first element in construct state constructions; (3) morphological phenomena (e.g. the conjugation of certain prepositions, such as 'otxem rather than 'etxem); and the use of function words (i.e. the employment of certain lexical items, such as texef 'immediately', 'od ha-pa'am 'again' and more).

ABSTRACT. [Note: PDF version also uploaded for ease of reading]

This presentation focuses on cases of apparent paradigm leveling found in varieties of two related sign languages, namely Japanese Sign Language (JSL) and Taiwan Sign Language (TSL). Paradigm leveling, or the equivalent of it in sign languages, has not been documented, part of the reason being that it is not common for grammatical features of sign languages to be described as paradigms. However, in some semantic domains where the number of the constituent lexical items is limited, it is sometimes found that a phonemic change takes place across multiple members, sometimes in all members belonging to the same domain, such as kinship terms, number expressions and days of the week. Such changes can be perceived as parallel to morphological changes that occur in paradigms in spoken languages, such as pronouns, in the sense that a change is associated with an abstract notion shared by the members in common.

In this presentation, we will try to capture patterns of identifiable changes in some paradigms in JSL and TSL. Then we will then examine factors that trigger the changes and the causes for limiting (or non-limiting) the spread of the changes. Documentation of old forms of JSL is limited, and the examination is based on a comparison between an old form(s) appearing in available sources (Sadobaru 1902, Matsunaga 1937, and Peng 1976) and the corresponding forms found in modern varieties of JSL and TSL. In the following description, the notations follow those proposed in Kikusawa and Sagara 2015, developed for describing phonological changes in JSL and related languages under examination. In brief, an index finger to a little finger is referred to as 1 through 4 respectively and the thumb is referred to as 5. Symbol “S” indicates that the finger is straight, “B” bent, and “P” in pinching position. The fingers that are not described are in the default (fist) position and the description applies to the dominant hand unless specified otherwise. 　　 The extent of the changes discussed ranges from partially shared to fully shared within each paradigm. In basic kinship terms (documented in 1976 by Peng), for example, all contained an expression supposedly expressing ‘biologically related,’ articulated by the index finger of the dominant hand touching the cheek (_S1touch.cheek_) as in Table 1. However, in JSL, this expression is lost in the words for siblings, thus simply _S2up_ for ‘older brother’ simply and _S3down_ for ‘younger sister.’ In TSL, the finger for touching cheek has been assimilated to the following articulator. Now the same finger is used for both touching cheek and the brother/sister indication, thus, _S2touch.cheek^S2up_ for ‘older brother’ and _S3touch.cheek^S3down_ for ‘younger sister’. On the other hand, the expressions for ‘parent’ have been retained in both JSL and TSL without change, while the situation for ‘grandparent’ considerably differ from the others. There are various possibilities to be examined as to why these were retained while others have changed, including the ease or non-ease of articulation, token frequency, potential conflict with existing forms and others.

Table 1: Kinship Terms in Tokyo JSL as Documented in Peng 1976

GRANDPARENT PARENT SIBLING (OLDER) SIBLING (YOUNGER) MALE S1touch.cheek^B5 S1touch.cheek^S5up S1touch.cheek^S2up S1touch.cheek^S2down FEMALE S1touch.cheek^B4 S1touch.cheek^S4up S1touch.cheek^S3up S1touch.cheek^ S3down (Consecutive signs are connected by the symbol “^”.)

The nature of the “paradigms” looked at in this study may appear to be different from those that are commonly investigated and the time depth of the changes investigated is about 120 years and much shallower than that commonly looked at in spoken languages. However, various relevant patterns are found in partial changes in number systems, which are not necessarily the same across dialectal varieties. Changes in days of the week show partial replacement in JSL varieties, while full replacement, which is obviously contact induced, took place in each of the two TSL varieties. We believe that such specific cases will contribute to the cross-modal understanding of paradigm leveling.

References Matsunaga, Tan. 1937. Rooa temane jiten (Lexicon of Japanese Sign Language). Rooakai 78-82.

Sadobaru, Sue. 1909. Rooa-kyooju shuwa-hoo (Sign Language Expressions in Deaf Education). Kagoshima: Kagoshima Deaf School. [Reprint of 1902]

Peng, Fred C.C. 1976. Kinship signs in Japanese Sign Language. Sign Language Studies 5:31-47.

Kikusawa, R. and K. Sagara. 2015. Towards Historical Sign Linguistics: A Preliminary analysis based on number expressions. Paper presented at the 4th Meeting of Signed and Spoken Language Linguistics, National Museum of Ethnology, Osaka. September 20, 2015.

ABSTRACT. The Aramaic language family has survived into modern times in pockets across the Middle East. The remaining branches are Western Neo-Aramaic, Ṭuroyo, North-Eastern Neo-Aramaic and Neo-Mandaic. The largest branch, North-Eastern Neo-Aramaic, contains many dialects, with a high degree of mutual incomprehensibility between many, despite being indigenous to a relatively small region in the borderlands of Iraq, Turkey, Iran and Syria.

            War, persecution, and ethnic cleansing have driven most dialects into near-extinction, with some already extinct. In the last few decades, however, scholars have made a concerted effort to document them, and today we have a great deal of data on a wide variety of dialects, though much remains to be done.

            North-Eastern Neo-Aramaic is also blessed with an unusually long written history. While many Aramaic-speaking communities continued to write in the classical languages, long after the spoken language had moved on, certain Christian and Jewish communities in Iraq began to write religious literature also in their local vernacular. The earliest texts date as far back as the late 16^th century: Christian manuscripts from the region around Mosul and Jewish manuscripts from the town of Nerwa near the border with Turkey. In addition to these we have documentation by missionaries active in the region in the 19^th century.

            We are thus in an enviable position of being able to compare the dialects reflected in the early texts with our data on the dialects spoken in the same areas. We find that the manuscripts clearly reflect the regional variety, but, at the same time, they retain archaic features and lack certain constructions (such as new perfects) which must have emerged in the intervening time (see, e.g. Mengozzi 2012, Coghill 2016: 280–83, 285–86).

            Nevertheless, only a small minority of dialects are documented before the 19^th or 20^th centuries. Thus, in establishing diachronic developments in North-Eastern Neo-Aramaic, we depend to a very large extent on synchronic dialectal variation. That is, for a given construction or form, by comparing its form, function and distribution in various dialects, while drawing on cross-linguistic parallels, we can reconstruct its emergence and development.

            This has proved very fruitful, as I have shown in my work on the recent development of a new future tense (Coghill 2010, 2012), which drew upon my own fieldwork, and the development and decline of ergativity in eastern Aramaic (Coghill 2016), which would not have been possible without the documentation carried out by my colleagues, in particular Geoffrey Khan.

            In this talk I will discuss another construction found in many North-Eastern Neo-Aramaic dialects and its possible diachronic development, and show how detailed dialectal documentation, together with cross-linguistic parallels, enables us to reconstruct an unusual grammaticalization path: that of a verb of movement to a verbal prefix marking the past perfective.

            The verbal prefix concerned takes various forms, according to dialect: qam-, kəm-, qəm-, gəm- etc. It is attached to the so-called Present Base, a stem derived from an old Active Participle. When unprefixed, this stem has an irrealis function, but it may take a prefix to express the indicative present and another to express the future. For past perfective forms, another stem, the Past Base is used. Yet qam- with the Present Base may be used with an equivalent past perfective function, except that it is normally restricted to transitive verbs, and in some dialects is the only past perfective form to allow a full set of object suffixes.

            There have been various theories as to the origin of qam-: an origin in earlier Aramaic qḏām ‘before’ (Nöldeke 1868: 296–97), in the earlier Aramaic verb qaddem ‘he was/did before’ (Maclean 1895: 82), and in the verb qym I ‘to get up’ in the form qāʾəm ‘he gets up’ (Pennacchietti 1997). Most recently, Fassberg (2015), finding these theories lacking, has suggested it results from the metanalysis of the present indicative prefix k- with an initial m- of some Present Base stems. I will argue, however, that dialectal data showing qym I as an auxiliary verb with a similar function to the qam- prefix make Pennacchietti’s theory far more plausible, and that Fassberg’s arguments against it can be rejected. Moreover, as Pennacchietti showed, this is backed up by cross-linguistic parallels, in particular with the Catalan/Occitan go-past (Detges 2004, Jacobs 2011).

Coghill, Eleanor (2010a). ‘The grammaticalization of prospective aspect in a group of Neo-Aramaic dialects’, Diachronica 27:3, 359–410.
Coghill, Eleanor. 2012. ‘Parallels in the grammaticalisation of Neo-Aramaic zil- and Arabic raḥ- and a possible contact scenario’, Grammaticalisation in Semitic, Journal of Semitic Studies Supplement Series. Oxford University Press, 127-144.
Detges, Ulrich. 2004. ‘How cognitive is grammaticalization? The history of the Catalan perfet perifràstic. In Olga Fischer, Muriel Norde and Harry Perridon (eds) Up and down the Cline: The Nature of Grammaticalization. Amsterdam/Philadelphia: John Benjamins.
Fassberg, Steven E. ‘The Origin of the Periphrastic Preterite kəm/qam-qāṭəlle in North-Eastern Neo-Aramaic.’ In Geoffrey Khan and Lidia Napiorkowska (eds), Neo-Aramaic and its Linguistic Context. Piscataway, NJ: Gorgias Press.
Jacobs, Bart. 2011. ‘Present and historical perspectives on the Catalan go-past’, Zeitschrift für Katalanistik 24, 227–255.
Maclean, Arthur J. (1895). Grammar of the Dialects of Vernacular Syriac. Cambridge: Cambridge University Press.
Mengozzi, Alessandro (2012). ‘The contribution of early Christian vernacular poetry from northern Iraq to Neo-Aramaic dialectology: Preliminary remarks on the verbal system’, ARAM Periodical 24 (Proceedings of the Twenty-Sixth International ARAM Conference on Neo-Aramaic Dialects, Oxford, 6–7 July 2
Nöldeke Theodor. 1868. Grammatik der neusyrischen Sprache am Urmia-See und in Kurdistan. Leipzig: T.O. Weigel.
Pennacchietti, Fabrizio A. (1997). ‘On the etymology of the Neo-Aramaic particle qam/kim’ [in Hebrew]. In Moshe Bar-Asher (ed.), Gideon Goldenberg Festschrift (Massorot: Studies in Language Traditions and Jewish Languages 9–11). Jerusalem: Magnes, 475–82.

ABSTRACT. The Yeniseian family of central Siberia contains several extint languages (Yugh, Kott, Assan, Arin, Pumpokol) and one surviving language, Ket, spoken fluently by fewer than 50 elderly people today. Information gathered on the extinct languages, as well as extensive fieldwork over the past 65 years conducted on the three surviving Ket dialects, has yielded valuable insights into the prehistory of North Asia that impact our understanding of both areal and genetic linguistics. Through centuries of influence from the surrounding Turkic, Uralic, Tungusic and Mongolic families, Ket noun and verb morphology show a typologically rare shift from strongly prefixing to predominantly suffixing, while retaining the original prefixes as well. Yeniseian also appears to be genetically related to the Na-Dene (Athabaskan-Eyak-Tlingit) family of North America, so that Ket language data can serve as external comparanda that illuminate the origin of several otherwise inexplicable morphological in Na-Dene. Had the Yeniseian languages vanished without any documentation, these insights could never have been gained.

ABSTRACT. This paper will explore the relationship between documentary work on endangered languages and historical linguistic research from two perspectives: 1) how new data produced through language documentation can revolutionize our understanding of linguistic history, and 2) how the history we infer through the comparative method and historical linguistics can impact the endangered language communities with which we work.

It is no great surprise that new data often force analysts to revisit earlier assumptions. Mayan languages have, in the last three decades, been ably described and documented, many for the first time, and in several cases by linguists who are themselves native speakers of these languages. This wave of new data on Mayan languages has led to many major and minor revisions to our understanding of Mayan language history. In this paper, I focus on the history of grammatical aspect in Mayan languages and trace how new insights from improved descriptive data for contemporary Mayan languages have led to changes in our reconstruction of the development of aspect in Mayan. Newly available data for two extinct Mayan language, Ch’olti’ and the hieroglyphic language known as Classic Mayan, have also contributed significantly to our understanding of aspect in Mayan.

Finally, drawing from ongoing language documentation efforts for Ixil Mayan, this paper will discuss how historical linguistic research produces historical insights that can be made meaningful to speakers of endangered languages. The same marginalizing processes that lead to language endangerment also frequently work to erase the histories of communities of speakers of endangered languages. Because of this, even the fairly coarse-grained historical view that insights about language relationships and historical patterns of linguistic contact bring, can fill a gap in the history of indigenous peoples that may not be taught or valued in the dominant society in which they live. I will discuss how scholarly work on the history of aspect, and related historical linguistic details, has been communicated to, and interpreted by community members. While communicating historical linguistic findings to community members requires sometime substantial effort to make that work generally accessible, such community engagement puts historical linguists in a position to change history, not only in terms of linguistic reconstruction, but also in terms of how marginalized communities understand and connect to their own past.

ABSTRACT. A number of different social category systems are found in most parts of the Australian mainland (excluding Tasmania and the Torres Strait Islands). These include two-termed (matri- or patri-)moiety systems, four-termed section systems, eight-termed subsection systems, four-termed semi-moiety systems, as well as two-termed “generational moiety” systems. The distribution, specific names, and historical relations between these systems has puzzled ethnographers (and to some extent linguists) since the late nineteenth century. The names tend to be shared over large geographic areas, which do not match very accurately the distribution of linguistic families and subgroups established in classifications by historical linguists. Hence it appears that the distribution of terminological systems, if not that of the social categorisation systems themselves, is prevailingly the result of areal diffusion.

Our presentation is confined to section systems. One widespread system is illustrated by the Ngarluma system shown below: A and B, C and D are intermarrying categories; A and C, B and D are mother-child links.

A Panaka B Purung C Karimarra D Paltyarri

We illustrate the distribution of the main sets of section names—which form an arc spanning from the west to the east coast via a large north-central region. Using the methods of areal and historical linguistics, including borrowing and etymological analysis, we try to establish the scope, direction of spread and likely place of origin of the main terminological sets. We also explore possible mechanisms by which such systems were spread, including: • borrowing of terms with and without phonological and morphological adaptation (e.g. adding gender markers) • translation of terms (expected but hard to find examples) • substitution of a foreign term by a native one • borrowing of terms with pragmatic shift of categories (i.e. differences in marriage or descent relations from that of the borrowed system) • stimulus diffusion (i.e. imitation of a foreign system by creating one using native terminology We question what evidence there might be to determine whether the different attested systems were independently innovated or whether they all stem from a common origin.

This study, based on the Austkin research project (http://austkin.net), is an exercise in the application of historical linguistics to the reconstruction of prehistoric social change on a continent-wide scale.

References Dousset, Laurent. 2005. Assimilating identities: Social networks and the diffusion of sections: (Oceania Monographs 57) Sydney: University of Sydney. Koch, Harold. 2015. Patterns in the diffusion of nomenclature systems: Australian subsections in comparison to European days of the week. In Dag T.T. Haug (ed.), Historical Linguistics 2013: Selected papers from the 21st International Conference on Historical Linguistics… (CILT 334) Amsterdam: John Benjamins. 109-132. McConvell, Patrick, Ian Keen, and Rachel Hendery (eds), 2013. Kinship systems: change and reconstruction. Salt Lake City: University of Utah Press. McConvell, Patrick, Piers Kelly and Sebastien Lacrampe (eds). In press. Skin, kin and clan: The dynamics of social categories in Indigenous Australia. Canberra: ANU Press.

ABSTRACT. Research question: European languages constitute a linguistic area where cross-linguistically rather exotic linguistic features are concentrated and shared among genetically only loosely related languages (Standard-Average-European, henceforth SAE). Whereas relevant SAE-features have been identified by areal typologists already (Haspelmath 2001, Stolz 2006, Cysouw 2011), the age and origin of SAE are still mysterious. Most likely, there are several historical sources for the different SAE features (van der Auwera 2011). We argue that study-ing the distribution of SAE features across varieties of individual languages, both historical and present-day, substantially enhances our understanding of the emergence and spread of SAE.

Motivation: The idea of a European linguistic area has been established mainly on the basis of codified modern standard varieties. As a consequence, previous research has left two empiri-cal desiderata. The first desideratum concerns the diachronic perspective: Following Haspel-math (2001), the most relevant factor giving rise to SAE was language contact during the pe-riod of the Great Migrations. Whereas this is certainly correct for a significant proportion of SAE features, some features might have spread only later, e.g. due to Latinate influence dur-ing the Middle Ages, and/or due to certain codification strategies shared among SAE lan-guages. The Post-Migrations development of SAE features can be studied in those SAE lan-guages with a richly documented language history by analyzing earlier stages of the languages with regard to their SAE profile. The second desideratum concerns synchronic variation: Whereas Heine & Nomachi (2010) strongly emphasize that nonstandard varieties must be taken into consideration by areal-typological research, too, Murelli & Kortmann (2011) even go a step further: they suspect that SAE-ness is less articulate at the level of spoken vernacu-lars. If this turns out to be correct, then including nonstandard varieties into typological com-parison is not just a matter of granularity, but it may create a qualitatively different, perhaps more realistic picture of Europe’s position in typological space.

Method and data: As a starting point, we selected German for a pilot study because standard German is a core SAE language, has been attested in diachronic depth, has a long tradition of codification, and facts about the structure of dialects are relatively easily accessible. We tested 33 phonological and morphosyntactic SAE features (taken from existing collections by Haspelmath 2001, Stolz 2006, Cysouw 2011) for their presence in Modern Standard German, Old High German, and Low-Alemannic (south-western) dialects. We distinguished between obligatoriness (1.0), optionality (0.5) or complete absence (0) of the respective features. We used existing grammatical descriptions as well as a large electronic corpus of spoken Ale-mannic.

Results: Overall, standard German turns out to be the most typical SAE language of the three varieties under investigation, followed by Low-Alemannic, and then Old High German. We categorized SAE features according to their (potentially) differential presence in the three varieties: • ‘Constant’ features are those that are present in Old High German already and persist in later periods (14/33 features). • ‘Expanding’ features are those that are attested more pervasively or even exclusively in the modern varieties. They appear in three flavors: (i) ‘expanding-indifferent’: The feature expands after Old High German without any difference between standard and vernacular varieties (3/33); (ii) ‘expanding-from-above’: The feature is more prominent in Modern Standard German than in both Old High German and Low-Alemannic (5/33); ‘expanding-from-below’: The feature expands predominantly in the vernacular but less so or not at all in the standard variety (4/33). • ‘Reducing’ features are those whose presence is reduced (or eliminated) since Old High German times. Again, reduction may be ‘indifferent’ (1/33), ‘from-above’ (not attested), or ‘from-below’ (3/33).

Discussion: As for the expansion of SAE features in German, we are now able to infer four different ages: The oldest SAE features of German are those that were present already when the historical attestation in written records begins in Old High German times and have persist-ed since then (‘constant’ features). Younger features appear only after Old High German but were present already before the codified variety bifurcated from the vernacular (‘indifferent’ features). Features that are related to the codification process are even younger and appear only in the standard variety (‘expanding-from-above’ features), e.g. obligatory relative pro-nouns, obligatory subject pronouns, or the lack of negative concord (cf. Jäger 2008). The youngest SAE features in German are those that expand in the vernacular but have not (yet?) reached the standard variety (‘expanding-from-below’), e.g. case syncretism or ‘preterite de-cay’, i.e. the semantic shift of the perfect from anterior to preterital meaning (cf. Drinka 2013). Unexpectedly in the light of Murelli & Kortmann (2011), the expansion of SAE features is quite balanced with regard to the standard vs. vernacular distinction. However, reduction of SAE-ness is asymmetrical: Once established, SAE characteristics are maintained in the standard variety whereas they are given up more easily in the vernacular. In sum, the majority of SAE features is affected by Post-Migrations change, and most of those later changes are sensitive to variety type.

References:

Cysouw, Michael (2011): Quantitative explorations of the worldwide distribution of rare characteristics, or: the exceptionality of northwestern European languages. In: Simon, Horst J. / Wiese, Heike (eds.): Expect the Unexpected: Exceptions in Grammar. (Trends in Linguistics. Studies and Monographs 216.) Berlin / New York: de Gruyter. 411-32. Drinka, Bridget (2013): Sources of Auxiliation in the Perfects of Europe. In: van de Velde, Freek / de Smet, Hendrik / Ghesquière, Lobke (eds.): On multiple source constructions in language change. = Special issue of Studies in Language 37: 599- 644. Haspelmath, Martin (2001): The European linguistic area: Standard Average European. In: Haspelmath, Martin / König, Ekkehard / Oesterreicher, Wulf / Raible, Wolfgang (eds.): Language Typology and Language Universals, vol. 2. (Handbooks of Linguistics and Communication Science.) Berlin / New York: de Gruyter. 1492-1510. Heine, Bernd & Motoki Nomachi (2010): Is Europe a Linguistic Area? In: Nomachi, Motoki (ed.): Grammaticalization in Slavic Languages: From Areal and Typological Perspec-tives. Hokkaido: SRC. 1-26. Jäger, Agnes (2008): History of German Negation. Amsterdam / Philadelphia: Benjamins. Murelli, Adriano / Kortmann, Bernd (2011): Non-standard varieties in the areal typology of Europe. In: Kortmann, Bernd /van der Auwera, Johan (eds.): The Languages and Lin-guistics of Europe: A Comprehensive Guide. Berlin/New York: Mouton de Gruyter. 525-544. Stolz, Thomas (2006): Europe as a linguistic area. In: Brown, Keith (ed.): Encyclopedia of Language and Linguistics, Vol 4, Oxford: Elsevier. 279-295. van der Auwera, Johan (2011): Standard Average European. In: Kortmann, Bernd & van der Auwera, Johan (eds.): The Languages and Linguistics of Europe: A Comprehensive Guide. Berlin/New York: Mouton de Gruyter. 291-306.

ABSTRACT. In creole studies, it has traditionally been assumed that creole languages were the product of adult acquisition of a language, such that either a pidgin variety or, at least, one starkly impacted by the subtractive effects of second-language acquisition, was expanded by future generations into a new language. Under this assumption, creolization was a distinct process from language mixture alone, which can be argued to have affected all of the world’s languages to various degrees. This hypothesis, proposing that creolization is a particular kind of language change yielding languages both synchronically and diachronically distinct from older languages, can be termed the Creole Exceptionalism (CE) hypothesis.

Various creolists (Mufwene 2001, Ansaldo, Matthews and Lim 2007, Aboh 2015) have proposed, however, that to treat creolization as a distinct process is a scientific mistake, and that in fact, the languages traditionally termed creoles emerged from language mixture alone, and not pidgins or other varieties impacted by second-language acquisition to any significant degree. As such, there is proposed to have been no qualitative difference between the emergence, for example, of modern English and Gullah Creole English, or Haitian Creole and Romanian. This hypothesis, under which a language’s creators select features from a pool analogous to the kind posited under the study of population genetics, can be termed the Feature Pool (FP) hypothesis.

In this presentation, I will test these two competing hypotheses via an examination of the creole language Palenquero, with a Spanish lexifier, created by speakers of the Bantu language Kikongo. Under the FP conception, since both Spanish and Kikongo are richly affixal languages, Palenquero should be a highly synthetic and inflected language, when in fact, it is highly analytic. Also, the FP proposes that languages retain features shared by all languages in the context. However, Palenquero lacks many features that Spanish and Kikongo share, such as aspects of the marking of plurality and gender, marking of specificity, and Differential Object Marking.

The data suggest that Palenquero is not only a mixed language, but also one in which the highly subtractive effects of acquisition beyond the Critical Period of acquisition had a decisive effect upon its grammar. That is, social factors beyond hybridicity determined the typology of the creole. I will briefly indicate that the facts are similar for most creole languages, and review some pan-creole features that are diagnostic of pidginization rather than step-wise grammar-internal change.

Aboh, Enoch, 2015. The emergence of hybrid grammars. Cambridge: Cambridge University Press.

Ansaldo, Umberto, Stephen Matthews, and Lisa Lim, eds. 2007. Deconstructing creole. Amsterdam: John Benjamins.

Mufwene, Salikoko. 2001. The ecology of language evolution. Cambridge: Cambridge University Press.

ABSTRACT. This paper explores how different types of markedness relations interact in the development of aspect systems and how they can be captured theoretically. The Indo-European languages Vedic and Latin represent a convenient point of departure for an exploration along such lines. At the beginning of their attested traditions, both have typologically similar complex inflectional verbal systems where aspect plays an important role; however, in the course of time, Vedic gradually develops a remoteness-based tense system where aspect plays little or no role, while the Latin aspect system remains largely intact through its attested history, and forms an important part of the tense/aspect system of present-day Romance languages. Furthermore, in both of these languages different verbal stems represent the primary morphosyntactic means for expressing aspectual distinctions, which on the face of it are rather similar. On closer scrutiny, however, some significant differences appear. According to a recent study (Dahl 2016[2015]), Early Vedic has a three-way morphological distinction in its past tense system between the perfective Aorist, the anterior/retrospective Perfect and the neutral Imperfect. Latin, on the other hand, is generally taken to have a two-way distinction in its past tense system between the imperfective Imperfect and perfective Perfect (cf. e.g. Devine and Stephens 2015). An intriguing possibility arising from these observations is that the Early Vedic past tense system may be interpreted in terms of privative aspectual viewpoints, with the so-called Imperfect as a kind of default past tense, whereas the Latin system is structured as an equipollent opposition between mutually exclusive aspectual viewpoints. Under this assumption, it would be reasonable to assume that the different paths of developments the two languages follow is somehow linked to this fundamental difference. The notion of neutral aspect, deriving from Smith (1991), is controversial. In Dahl’s (2010, 2016[2015]) approach this aspectual type is defined in terms of an overlap relation between reference time and event time (cf. e.g. Demirdache and Uribe-Etxebarria 2008, 2014 for analogous analyses). This relation is taken to be semantically more general than those characterizing the imperfective and perfective aspect, which are defined in terms of opposite inclusion relations between event time and reference time. Recently, Altshuler (2014) has suggested an alternative approach to aspectual interpretation, where viewpoint operators are taken to be partitive in nature, and to differ with regard to a restricted number of properties (whether the extension of the predicate they select needs to contain proper parts of the event it denotes or not, whether they select singular events or not, and whether they obligatorily include the culmination or maximal part of the event or not). He proposes the following hypothesis about (im)perfectivity (2014: 771): a. An operator is perfective if it requires a maximal stage of an event in the extension of the VP that it combines with. b. An operator is imperfective if it requires a stage of an event in the extension of the VP that it combines with, but this stage need not be maximal. An immediate advantage of an approach along such lines is that it renders the assumption of a separate neutral viewpoint unnecessary and allows for distinguishing different types of perfective and imperfective operators. The two approaches to aspectual semantics outlined above provide complementary perspectives on the development of aspect systems. For instance, Dahl (2016[2015]) suggests that the marked aspect categories in Vedic, the perfective Aorist and the anterior/retrospective Perfect, gradually lose their various lexically and contextually determined readings, ultimately showing a neutral aspectual meaning. On this view, the development of the Vedic aspect system is understood as a gradual process of semantic erosion of the two aspectually marked past tense categories, while the unmarked category remains stable. On the other hand, the analysis of the development of the Latin tense/aspect system given in Haverling (2010) is based on the assumption that the Perfect originally represented a general (perfective) past, which developed a more markedly perfective meaning in Late Latin. The Imperfect behaves as a markedly imperfective past tense in Early and Classical Latin but shows an increasing tendency in Late Latin to be used in contexts previously characteristic of the Perfect. This development appears to represent something like a markedness shift in the aspect system and seemingly may be elegantly captured by Altshuler’s (2014) model. The question arises whether the two theoretical approaches can account equally well for pertinent diachronic data from the two languages under consideration, and this will be addressed in the present contribution. References: Altshuler, Daniel. 2014. A typology of partitive aspectual operators. In Natural Language and Linguistic Theory 32:735–775. Dahl, Eystein. 2010. Time, Tense and Aspect in Early Vedic Grammar. Exploring Inflectional Semantics in the Rigveda. Leiden: Brill. Dahl, Eystein. 2016[2015]. ‘Toward a formal model of semantic change: A neo-Reichenbachian approach to the development of the Vedic past tense system.’ In Lingua Posnaniensis Vol 57(1), 41-76. Demirdache, Hamida and Myriam Uribe-Etxebarria. 2008. Scope and anaphora with time arguments: The case of ‘perfect modals’. In Lingua 118 (2008) 1790–1815. Demirdache, Hamida and Myriam Uribe-Etxebarria. 2014. Aspect and temporal anaphora. In Natural Language and Linguistic Theory 32: 855–895. Devine, Andrew M. and Laurence D. Stephens. 2013. Semantics for Latin. An Introduction. Oxford: Oxford University Press. Smith, Carlota. 1991. The Parameter of Aspect. Kluwer Academic Publishers, Dordrecht/Boston/London. Haverling, Gerd. 2010. Actionality, tense, and viewpoint. In Philip Baldi and Pierluigi Cuzzolin (Eds.) New Perspectives on Historical Latin Syntax 2. Constituent Syntax: Adverbial Phrases, Adverbs, Mood, Tense. Berlin: De Gruyter Mouton, 277–524.

ABSTRACT. In Old Indic and Iranian languages (Sanskrit, Old Persian and Gathic)the Perfect stem is formed by reduplication of the initial portion CV of the root and a special set of endings: Sanskrit da-dar-s'a 'I have seen', Young Avestan da:-dars-a 'I have seen'. During the late Old Iranian period we witness the rise of analytic formations whereby the monolectal aspectual categories (Imperfect, Aorist, Perfect) were lost and subsequently were renewed on an analytic basis by combining the past passive participle (the form in -ta) with the copula. During the Middle Persian period the Old Persian construction uta:=maiy ... kartam was reduced to u=m ... kard '(and) I did' and reinterpreted as an ergative construction by moving into the earstwhile domain of the active perfective category (Aorist). In later development the possessive clitics were replaced by personal suffixes in postverbal position (New Persian). In Kurdish the outcome is different in that the possessive clitics can still be hosted by clause-initial items as in Middle Persian (Kurdish (Sorani) Imperfect a=m kird vs. Farsi mi-kard+am). In the analytic Perfect the past participle is finitized by the innovative forms of the copula 'ast-am 'I am', 'ast-i: 'you are' by attaching the present tense suffixes +am, +i: to the 3rd Pers 'ast). This is a striking example for the change when the 3rd Pers Sg serves as the pivot for the rebuilding of the whole paradigm (as in Polish jest-em 'I am'): MP u=m kard he:m > kard 'ast+am > NP kard-e='am > kard+a:am. In Colloquial Farsi the final -e of the perfective participle kard-e contracts with the initial vowel of the suffix of the copula yielding distinctive new accented vowel forms of the Perfect kard+a:m 'I have done', kard+i: you have done' (unlike the forms of the perfective past which are accented on the root: kard+am 'I did', kard+i: 'you did'). Parallel Perfect constructions in West Iranian languages (Kurdish, Luri, Bakhtiari) will be provided and the major difference in their sequencing of the tense and person markers will be demonstrated: in Kurdish kird-u:m-a do+1Sg+be the reduced 3Sg form of the copula ='ast > +a follows the finitized form of the participle. The reduction of the person-number inflection and the generalization of the 3Sg form as an 'impersonal' subjectless predicate represents an interesting case of 'degrammation' (cf. Andersen 2010:145). The inferential mode in the Perfect Aspect will be discussed in the context of the maintenance of earlier categories and their later appearance due to language contact. The simple past is used when the speaker witnessed the event (esp. if its time is specified) but the resultative Present Perfect is appropriate for narratives based on hearsay, temporal remoteness and retrospectivity. Contrast s'ira:z raft-id 'Did you go to Shiraz ?' [last year] with s'ira:z raft='id ?' 'Have you gone to Shiraz ?' [sofar]. Diachronically, the inferential developed from the various usages of the Present Perfect whence it spread to other verbal categories: Imperfect (das'te mi-rafte and Pluperfect rafte bude-ast). The same process is observable in other West Iranian languages "probably under Turkish influence". (Windfuhr 1982): Kurdish: z'enaft-em-a Ali c'u-a bo s'ira:z 'I heard that Ali has gone to Shiraz' versus Ali c'u bo s'ira:z 'Ali went to Shiraz' [last week]. At the end of the continuum of Turish influence is Talishi which modeled its inferential mode on the pattern of Turkic inferential (cf. Osmanli i-mis') by featuring a single marker ban (adopted from the 3Pl form of the copula).

Andersen, Henning. 2010. From morphologization to demorphologization. In: Continuum Companion to Historical Linguistics, ed. by S. Luraghi and V. Bubenik, 117-146. London: Continuum. Khanlari, P. 1986. Ta:rikh e zaba:n e faarsi. Tehran: Nashr e now. Rokhzadi, A. 2000. The Phonetics and the Grammar of Kurdish. Tehran: Tarfand. Windfuhr, Gernot L. 1982 The verbal category of inference in Persian. Acta Iranica 22. ___________. 2009. The Iranian Languages. London: Routledge.

ABSTRACT. Developments in tense and aspect systems become most visible when new forms are grammaticalized and old forms become extinct. These dynamics can be shown best in oral varieties, i.e. dialects, as they lack the preservative force of literacy and standardization. This is the case in the language change process called Präteritumschwund (preterite loss) that primarily took place in the Upper German dialects (i.e. Bavarian, Alemannic, Swabian and East Franconian) starting in the Early New High German period. The underlying process which caused the loss of the preterite tense forms (as in 1) can be identified as the semantic and functional extension of the present perfect tense form (as in 2). (1)preterite tense form: Maya kam an. Paul aß Kekse. ‘Maya arrived’. ‘Paul ate cookies’. (2)present perfect tense form:Maya ist angekommen. Paul hat Kekse gegessen. ‘Maya has arrived’. ‘Paul has eaten cookies’. The German present perfect is a relatively young tense form that was grammaticalized in the Old High German period (from 800 on). First, it was used for expressing present anterior meaning only. Beginning in the late Middle High German period and starting in the Upper German dialects, the present perfect could also be used for situations that are temporally anchored in the past but nonetheless have a current relevance meaning. This usage serves as a bridging context between typical present perfect and past tense meanings. Gradually, the German present perfect developed into a past tense form that could then also be used for expressing past perfective and past imperfective meaning, both of which could only be expressed by the preterite tense form earlier. Fig. 1 illustrates this semantic extension path.

present anterior: Maya ist gerade angekommen. ‘Maya has just arrived.’ > past with current relevance: Maya ist heute morgen angekommen (und jetzt da). ‘Maya arrived this morning (and she is still here).’ > past perfective: Maya ist heute morgen angekommen. ‘Maya arrived this morning.’ > past imperfective: Früher ist Maya immer pünktlich angekommen. ‘In the past, Maya used to arrive on time.’ Fig.1 Semantic extension of the German present perfect tense form

In this development the present perfect form followed the universal grammaticalization path as described by Bybee/Dahl (1989, 58, 68): “have/be + past participle > perfect (anterior) > past or perfective”. The semantic enrichment of the present perfect entailed an increase in frequency of usage which culminated in the successive replacement of the preterite form. While the present perfect increased in meaning, the preterite gradually became restricted to certain contexts (e.g. narrative discourse mode, expression of imperfective past). As an additional factor the dialectal preterite paradigms became disadvantageous due to a series of sound changes that led to irregular and defective verb forms. The perfect extension and, consecutively, the loss of the preterite form started in the south-east region of the German speaking area and spread gradually to the west and the north. Today, the historical process is reflected in a specific areal distribution: dialect atlases show a diffusion fan with multiple tiers of verb specific isoglosses. The chronological process is mirrored in space. Both the expansion of the present perfect and the loss of the preterite in the German dialects have been repeatedly studied and contradictorily discussed by historical linguists and dialectologists (e.g. Lindgren 1957, Rowley 1983, Dentler 1997, Sapp 2009, Fischer 2015). The results were recently synthesized in Fischer (2016). In my talk I will begin with an overview of the areal distribution of preterite and present perfect tense forms in the German dialects and the underlying historical developments. Second, I will focus on the semantic shifts in the German tense and aspect system that allowed the preterite loss in the first place. Using modern temporal and aspectual concepts (e.g. as they are discussed in Binnick 2012) we can shed more light on the underlying semantic shifts in the German tense and aspect system. As the German language does not differentiate between perfective and imperfective meaning by form, the semantic analysis inevitably has to consider all the constituents that shape the temporal framing of situations such as temporal adverbials, lexical properties of the verb and pragmatic properties such as discourse functions. To offer some perspectives for further research, I will compare the German process to similar developments in Romanic and other Germanic languages.

Bybee, Joan L./Dahl, Östen (1989): The creation of tense and aspect systems in the languages of the world. In: Studies in Language 13, 51–103. Binnick, Robert I. (ed.) (2012): The Oxford Handbook of Tense and Aspect. Oxford, New York: Oxford University Press. Dentler, Sigrid (1997): Zur Perfekterneuerung im Mittelhochdeutschen. Göteborg: Acta Universitatis Gothoburgensis. Fischer, Hanna (2016): Präteritumschwund im Deutschen. Dissertation. Universität Marburg. Fischer, Hanna (2015): Präteritumschwund in den Dialekten Hessens. In: Elmentaler, Michael et al. (eds.): Deutsche Dialekte. Stuttgart: Steiner, 107–133. Lindgren, Kaj B. (1957): Über den oberdeutschen Präteritumschwund. Helsinki. Rowley, Anthony R. (1983): Das Präteritum in den heutigen deutschen Dialekten. In: Zeitschrift für Dialektologie und Linguistik 50/2, 161–182. Sapp, Christopher D. (2009): Syncope as the cause of Präteritumschwund. In: Journal of Germanic Linguistics 21/4, 419–450.

ABSTRACT. Previous Research: Broadening refers to changes where “the range of meanings of a word increases so that the word can be used in more contexts than were appropriate for it before the change.” (Campbell 2004) E.g.: Latin passer ‘sparrow’ > Spanish pájaro ‘bird’.

Problems: First, the concept of “meaning” is taken for granted, and has never been addressed in depth. Second, most researchers assume change of word meaning and where the word can be used are the two sides of a coin. They go together.

Proposal: First, I propose to use the distinction between intension and extension to study historical meaning change. Second, I ask whether the “two sides of a coin” in semantic change can be separate. Is it possible to use a word “in more contexts” without changing the “meaning” of the word? In temporal semantics, a word meaning is interpreted relative to a time variable. According to von Fintel and Heim (2002), the semantics of ‘teacher’ in the following sentence (1) is (2):

(1) John is a former teacher.

(2) For any t ∈ T: ⟦teacher⟧t = λx∈D. x is a teacher at t.

Formula in (2) shows the set of teachers at different times may differ. Suppose at t1, we have the following individuals who were teachers {Dylan, David, Daniel, Don}; at t2 {John, Jane, Jay, Jo}. On the surface, there is no change of meaning because it is change of specific members in a set. However, if at t1 all the teachers were male, while at t2 the teachers include both males and females, now this is a genuine semantic change. Although the intension of “teacher” did not change, the set of individuals has been extended to include another subset. I define such extensional broadening as:

(3) EXTENSIONAL BROADENING =df A at t1 > A ∪ B at t2, where t1≺ t2, A and B are well-defined subsets in the extension of the same word at t2, and A∩B = ∅.

A Case Study: I use the word *pljiət (Old Chinese) > bǐ (Modern Chinese) for “writing tool” in Chinese as an example. According to the definition of the word *pljiət in a dictionary from early 2nd century, the meaning of *pljiət is “that which is used to write”, i.e. “writing tool”. This is exactly the same as the modern word bǐ. Historical semantic change theory does not really recognize such changes. But if we look at the set of individuals that this word can be applied to, there is definitely a semantic change. In 2nd century China, it was mostly brushes that were used as writing tools. Nowadays, however, writing tools include many subsets, e.g. pens, pencils, brushes, etc. This is exactly a case of extensional broadening as defined in (3) above.

The Theoretical Implications: First, extensional broadening as defined here has generally been ignored by most researchers. However, attention to extensional broadening is very important to the interpretations of ancient texts. Second and more importantly, I have shown that if we consider both intension and extension, they can change independently. Now we have a more elaborate model of semantic change.

ABSTRACT. This paper studies the relation between two lexical-semantic diachronic processes of change of the prepositional phrase introduced by the Spanish preposition a + noun: delocativization and abstraction. The main finding of this research is that the locative basic meaning of this sort of phrases decreased, throughout the history of Spanish in favor of non locative ones. This led to an increase in the amount of abstract nouns, to the detriment of concrete nouns. The interest of this lexical and semantic change is that it was accomplished as a “domino effect”: a first delocativization cleared the way for a subsequent abstraction. The prepositional phrase headed by the preposition a is characterized by having significantly increased, throughout the history of Spanish, the number of functional contexts, required or not required by the argument structure —e.g. locative complement, indirect object, direct object, temporal complement, modal complement, discursive markers, jussive constructions, etc.— (Company and Flores 2014). The functional increase of the prepositional phrase with a is the cause and the effect of the progressive separation of basic etymological context of the preposition a, and of the subsequent acquisition of the innovative meanings and contexts. There are two basic meanings involved in the sequential change experienced by semantics of the noun phrase headed by the preposition a. Firstly, the etymological meaning ‘locative goal’ of the preposition a and, secondly, the concrete character inherent to this type of noun phrases. These two meanings have an intrinsic relation, since the locative goal of the etymological nature of the preposition a is preferably coded with concrete nouns. This correlation is based on the fact that a localization is linked better to specific entities rather than to abstract entities (cf. Luraghi 2003:11-48; Asbury, van Gehrke, Riemsdijk and Zwarts 2008:1-32). The separation from the locative meaning of the noun phrase occurred through a process of delocativization. Subsequently, the delocativization has impacted on the concrete character of the noun phrase and has resulted in a process of abstraction. The delocativization processes have taken place in different and subsequent stages: strict locative contexts (1a) > contexts with locative semantic extensions (1b) > contexts empty of locative nuances.

(1) a. Quitosse de la ciudat de Nazareth, e fue morar a Capharnaum [Mateo, 27] ‘He left the city of Nazareth, and moved to Capharnaum’ b. cuando llegaste junto a ella, ¿no sentiste un olor sabeo, una fragancia aromática y un no sé qué de bueno, que yo no acierto a dalle nombre? [Quijote I, 31] ‘When you got close to her, did you not smell a savor, an aromatic fragrance, and a hint of something good, that I cannot name?’ c. Me reí a solas [Bartleby, 12] ‘I laughed by myself’

The abstraction of the noun phrase was done by changing nouns of concrete character (2a), majority in the first centuries analyzed, to nouns of abstract character (2b), preferred in the latter centuries.

(2) a. Et yo dixe verdat et mostrarélo por prueba, et dixelo al rey [Calila,187] ‘And I told him the truth and showed him the proof, and I told it to the king’ Yo, viendo que era batalla nabal y que no se había de hacer a caballo, comencé a apearme [Buscón, 87] ‘I, seeing that it was naval battle and that it was not to be done on horse, began to leave’ b. Para dar mayores visos de legalidad a la trapisonda, Cárcamo lo registró en la prisión con su nombre cristiano [Ángeles, 479] ‘To give more glimpses of legality to the disturbance, Cárcamo registered him in the prison with his Christian name.’ Le ha bastado con enfrentarla a su fragilidad, a su pequeñez, a su insalvable dependencia del hombre que aún la ama [Vuelo, 271] ‘It has been enough to confront her to her fragility, her smallness, her insurmountable dependence on the man who still loves her’

I argue that both processes, delocativization and abstractions, occurred diachronically as a “domino effect”. This hypothesis is based on the quantitative analysis of 1390 occurrences distributed in three chronological sections: 442 in the 13th century, 427 in the 17th century and 521 in the 21st century. The quantitative evidence is supported by Z statistic proof.

References

Asbury, Anna, Berit Gehrke, Henk van Riemsdijk and Joost Zwarts. 2008. “Introduction: Syntax and Semantics of Spatial P”, in Anna Asbury, Jakub Dotlačil, Berit Gehrke and Rick Nouwen (eds), Syntax and Semantics of Spatial P, Amsterdam/Philadelphia, Benjamins, 1-32. Company Company, Concepción. 2003. “La gramaticalización en la historia del español”, en Gramaticalización y cambio sintáctico en la historia del español, in Concepción Company Company (ed), México, Universidad Nacional Autónoma de México, 1-62. Company Company, Concepción and Rodrigo Flores Dávila. 2014. “La preposición a”, in Concepción Company Company (dir), Sintaxis histórica de la lengua española. Tercera parte: Adverbios, preposiciones y conjunciones. Relaciones interoracionales, Mexico, FCE-UNAM, 1195-1340. Kroch, Anthony. 1989. “Reflexes of Grammar in Patterns of Language Change”, Language variation and change 1, 199-244. Luraghi, Silvia. 2003. On the Meaning of Prepositions and Cases, Amsterdam/Philadelphia, John Benjamins.

ABSTRACT. ABSTRACT

Using data from the Corpus del español, this study examines the historical evolution of certain Spanish structures formed with the verb caer ‘fall’ and nouns designating situations (e.g. peligro ‘danger’) or physical or psychological states (e.g. temor‘fear’, desesperación ‘despair’). In Modern Spanish these structures feature an experiencer (the participant that experiences the state or situation designated by the noun) that is realized as the grammatical subject of caer, whereas the noun predicate itself is introduced by the preposition en ‘in’, e.g. el caballero cayó en temores infundados (literally, ‘the knight fell into unfounded fears’). These structures represent a metaphorical extension of the basic notion of movement associated with caer, and they can be analyzed as collocations in Alonso-Ramos’s (2004) sense. Specifically, they can be analyzed as inchoative collocations in which an entity begins to experience the situation or state designated by the noun predicate (Alonso-Ramos 2004: 108).

Diachronically, inchoative collocations with caer are particularly interesting because in previous stages of Spanish the experiencer could be realized not only as a subject, but also as a dative (le cayeron temores infundados, literally ‘unfounded fears fell to him’) or as a metaphorical locative (cayeron temores infundados en él, literally ‘unfounded fears fell into him’). This is the same historical alternation found in inchoative collocations formed with two other movement verbs: entrar ‘enter’ (Alba-Salas 2016) and venir ‘come’ (Alba-Salas 2017, to appear); compare, for example, Modern Spanish le entró temor ‘he became frightened’ (literally ‘fear entered to him’) with earlier structures such as entró en temor (literally ‘he entered into fear’) and entró temor en él (literally, ‘fear entered into him’). In the case of venir, dative experiencers predominated over subject and locative experiencers –both of which gradually disappeared after the 1600s– already in the 1200s (Alba-Salas 2017, to appear). With entrar, on the other hand, locative experiencers, which predominated in the Middle Ages, were eclipsed by subject experiencers in the 1600s, but dative experiencers prevailed after the 1800s (Alba-Salas 2016). According to Alba-Salas (2016 and 2017, to appear), the victory of dative experiencers with entrar and venir followed from several grammatical changes, most notably the emergence of datives as the canonical realization of the experiencer in Spanish structures indicating involuntary physical or mental processes –a change that, following Elvira (2011) and Vázquez Rozas and Rivas (2007), is conceptualized as the generalization of a Construction (understood in Goldberg’s 1995 technical sense).

The corpus data to be presented suggest that, in contrast with entrar and venir, in the case of caer subject experiencers have predominated since the 1200s, whereas dative experiencers were never productive. The contrast would follow from a variety of factors, particularly from a key difference in the lexical semantics of these three movement verbs: whereas entrar and venir designate a trajectory that necessarily entails an endpoint, so they are inherently telic, caer does not encode the endpoint of the designated trajectory, so that endpoint (if it is to be expressed) must be realized as a prepositional complement, e.g. caer en el agua ‘to fall into the water’ (Morimoto 166-167 and 173). Because of its lexical semantics, then, caer favors figurative uses based on the metaphor of states and situations as containers, conceptualizing a change of state as a change of location. This is the conceptual basis for inchoative collocations with subject experiencers of the type caer en + state noun. Unlike entrar and venir, however, caer is ill-suited for metaphorical uses based on the schema of adlative movement (movement towards a specific endpoint), which constitutes the conceptual basis for collocations with dative experiencers such as entrarle/venirle + state noun (Cifuentes Honrubia 2015: 51-52; cf. Morimoto 2001: 204-207).

REFERENCES Alba-Salas, Josep (2016): “El triunfo del experimentador dativo: Las colocaciones con ‘entrar + nombre de estado’ en diacronía”, Revista de Filología Española, XCVI, pp. 9-38. Alba-Salas (2017, to appear): “Venir vergüenza: Cambios históricos en las colocaciones con venir”, Zeitschrift für romanische Philologie, CXXXIII/1. Alonso Ramos, Margarita (2004): Las construcciones con verbo de apoyo, Madrid, Visor Libros. Cifuentes Honrubia, José Luis (2015): Construcciones posesivas en español, Leiden, Brill. Corpus del Español: Davies, Mark (2002-): Corpus del Español (100 million words, 1200s-1900s). [consulted: 8-9/2015 & 5-8/2016]. Elvira, Javier (2011): “Constructions of Uncontrolled State or Event: The Increase in Productivity of a New Argument Structure in Old Spanish”, Constructions and Frames, III/2, pp. 184–207. Goldberg, Adele (1995): Constructions: A Construction Grammar Approach to Argument Structure, Chicago, Chicago University Press. Morimoto, Yuko (2001): Los verbos de movimiento, Madrid, Visor. Vázquez Rozas, Victoria, and Elena Rivas (2007): “Un análisis construccionista de la diacronía de gustar”, in Iraide Ibarretxe-Antuñano, Carlos Inchaurralde and Jesús Sánchez-García (eds.), Language, Mind, and the Lexicon, Frankfurt, Peter Lang, pp. 143-164.

ABSTRACT. This contribution assesses the proposal in Szabolcsi (2015) as regards its applicability in the analysis of indefinites in Old Hungarian. The main finding is that the main formal claim (quantificational particles signal the presence of a covert join or meet operator) is not accompanied by the analytic tools for a satisfactory analysis. In addition, alternative theories of indefinites (including diachronic theories) can be said to share some (very general) properties with Szabolcsi's main thesis, while they are better suited for detailed empirical analysis.

ABSTRACT. Disjunctive and conjunctive particles meet their negative concord relatives

The study of logical particles in recent literature has highlighted the fact that across many languages, the same morphemes occur in three different roles. (i) They form quantifier words with indeterminate pronouns, e.g. Japanese dare-ka `someone’, dare-mo `everyone, anyone’, (ii) they reiterate on all the members of disjunctions and conjunctions, e.g. ringo ka mikan (ka) `either apple or orange’, ringo mo mikan mo `both apple and orange’, and (iii) they occur as unary interrogative and additive particles, e.g. question markers and particles meaning `too, even’. It is well-known that the third member of the Boolean particle family, the negative particle has similar properties. To illustrate with English, (i) no one, (ii) neither apples nor oranges, and (iii) sentential negation markers, as in I did not and Nor did I. This paper will focus on the negative concord (NC) particles of Hungarian and situate them in the landscape of the logical particles of the language.

ABSTRACT. Synchrony and diachrony of a multifunctional particle: Latin nec

Latin nec and the non-apocoped form neque derive, quite transparently, from the combination

of the Indo-European negative morpheme *ne with the postpositive enclitic coordination

particle –que (= –c, cf. the pair atque / ac ‘and’). The particle knew three uses: (i) discoursestructuring

connective ‘and not’, at the beginning of new textual units (not shown); (ii)

correlative particle ‘neither’, cf. (1); (iii) stand-alone focus particle with an additive (‘also

not’) or a scalar (‘even not’) interpretation, cf. (2).

(1) Caput dolet neque audio nec oculis prospicio satis.

head:NOM hurt:3SG and.not hear:1SG and.not eyes:ABL see:1SG well

‘I have a headache, I can’t hear, and I can’t see well with my eyes’ (Pl. Amph. 1059, 3rd BCE)

(2) nemo mundus, nec infans

nobody:NOM pure:NOM and.not infant:NOM

‘No one is pure, not even an infant’ (Leo M. Serm. 21, 5th CE)

The coordinative functions (i) and (ii) are historically primary for neque / nec. The particles

are continued by the Romance languages in function (ii) (e.g. French ni, Italian né, Romanian

nicĭ). The use as stand-alone focus particle for nec appears later in time (I cent. CE) and

becomes particularly frequent in Late Latin. The Late Latin example in (2), where the particle

has a clear scalar import (even the most likely candidate for purity is not pure), represents the

context of grammaticalization for many Romance indefinite determiners, incorporating a

formally negative morpheme that derives from nec (e.g. Spanish ninguno, Italian nessuno,

Old Italian niuno). We see thus, that Latin nec manifests a multifunctional nature, reminding

us of better studied cases, like Japanese mo or Hungarian ki, and of a crosslinguistically

frequent scenario (Haspelmath 1997, Szabolcsi 2013, Mitrović & Sauerland 2013). The

analysis that I propose for nec contributes to this research field by examining the behavior of

a conjunction particle which intrinsically combines with a negative morpheme and by

highlighting the role of focus in its semantic development.

I define the different uses of nec on the basis of the contextual conditions determining its

various readings: I show that not only in function (iii), but also in function (ii) it qualifies a

focus-sensitive particle. I then propose a parsimonious syntactic implementation: capitalizing

on its bimorphemic nature (ne-c, ne-que), I propose that the particle has a complex internal

structure, and I discuss how it is integrated in the clause.

As a discourse-structuring particle (i), nec introduces a full clause belonging to a new

discourse unit, which may be connected in the discourse to a previous clause, independent of

the polarity of the latter. In its function as correlative negation (ii), nec relates two or more

negative constituents, which can be of various types, comprising CPs. However, unlike with

the discourse-structuring version, they belong to the same discourse unit. The discoursestructuring

use, where the preceding conjunct can be positive, clearly shows that nec is the

bearer of a semantic negation operator and can perform a switch in polarity, expressing

sentential negation by itself. In this, it conforms to the Double Negation system of Latin,

where each negatively marked element introduces a semantic negative operator, independently

of its position in the clause. Thus, in the discourse-structuring use nec has the meaning ∧ ¬,

where the negation is outscoped by the conjuction: this is consistent with the particle’s

etymology and ensures that the negation only takes scope over the conjunct directly

introduced by the particle. According to the analysis shown in (3), -que / -c is the head of a

Conjunction Phrase &P, which takes the CP it introduces as its complement. The reverse

surface order is due to prosodic factors, namely to the enclitic status of -que / -c, which forces

prosodic inversion. The negative particle ne- is itself proclitic: the two elements together form

a prosodically acceptable unit for Latin. Similarly, a meaning (¬ x ∧ ¬ y) can be attributed to

the correlation introduced by nec, which according to one of De Morgan's Laws, is logically

equivalent to a reading where the correlation is interpreted as a disjunction outscoped by

negation: ¬ (x ∨ y). However, in my analysis, the correlative particle itself does not contain a

Boolean conjunction operator in its lexical entry (cf. Szabolcsi 2013, Mitrović & Sauerland

2013). Rather, the correlative particle is a focus particle with an additive component: in the

case of correlative neque / nec, the morpheme -que / -c realizes an additive Focus operator,

not a conjunction, requiring the presence in the context of a salient alternative proposition

with the same truth value. The proposed structure is shown in (4): the reverse surface order is

due, as for (3), to prosodic restructuring. The conjuncts in correlative constructions like (1)

are asyndetically coordinated, as in Bianchi & Zamparelli (2004).

(3) discourse-structuring particle 

[Table included in PDF version of the abstract from the ICHL program index]

(4) correlative and focus particle

[Table included in PDF version of the abstract from the ICHL program index]

The structure in (4) has the advantage of positing the same scope relation between the

negative morpheme and -que / -c for both coordinative functions (i) and (ii). The difference

resides in the meaning contribution of -que / -c, which is a conjunction in the discoursestructuring

use and an additive focus particle in the correlative use. The proposal also allows

us to treat nec as unambiguously negative ([Neg]) across its uses. A further advantage is that

it can account for the later development of the stand-alone focus particle (iii). In the use in

(iii) nec can have an additive or a scalar interpretation, with the latter becoming more frequent

with time. The internal syntax for the stand-alone use remains unchanged and conforms to the

structure in (4). The precise meaning contribution of nec in (iii) depends on the way alternatives

are retrieved, which in turn influences the structure that the set of alternatives has. The use as

additive focus particle is possible only when suitable alternatives for the focus are explicitly

provided in the context, by means of correlation or by anaphoric linking to the previous

discourse. In the absence of these preconditions, additivity leads to presupposition failure; thus,

only a scalar interpretation is possible: in that case alternatives have to be accommodated by

evoking a scale, whose dimension is usually suggested by the element in focus. This process of

presupposition accommodation may have been responsible for the establishment of the scalar

meaning for nec: accommodation processes on the part of the hearer are costly and, if

systematic enough, may lead to a reanalysis of the conditions imposed by the lexical entry (cf.

Traugott & Dasher 2002, Eckardt 2006, and for presupposition accommodation especially

Schwenter & Waltereit 2010). In turn, the scalar contribution makes the particle a suitable

item to strengthen negation, according, again, to a crosslinguistically frequent pattern which

witnesses ‘even’ as a component of polarity-sensitive quantificational expressions

(Haspelmath 1997, Lahiri 1998, Chierchia 2013).

Selected references: den Dikken, M. 2006. ‘Either’-float and the syntax of co-‘or’-dination.

NLLT 24: 689-749. Haspelmath, M. 1997. Indefinite pronouns. OUP. Lahiri, U. 1998. Focus

and negative polarity in Hindi. NLS 6. 57–123. Mitrović, M. & U. Sauerland 2013.

Decomposing coordination. In Iyer, J. & L. Kusmer (eds.), Proceedings of NELS, 39-52.

Schwenter, S. & Waltereit, R. 2010. Presupposition accommodation and language change.

In Davidse, K. et al. (eds.), Subjectification, intersubjectification and grammaticalization, 75-

102. Berlin. Szabolcsi, A. 2013. Quantifier particles and compositionality. In Aloni, M. et al.

(eds.), Proceedings of the 19th Amsterdam Colloquium, 27-34.

ABSTRACT. It is rightly noted in the literature on analogy that paradigm leveling poses a number of interesting problems as to (i) its motivation — is it a matter of paradigm uniformity or adherence to an external pattern? (ii) its results — how does partial leveling fit in? and (iii) its directionality — which form leads and which form follows? To take the example of changes in English nouns with morphologically determined voicing in the plural, e.g. wife/wives, we see that leveling of the voiceless/voiced distinction has been proceeding for quite some time, in the largely expected direction away from the somewhat unmotivated and thus irregular voicing, leading to synchronic variation in the plurals between innovative hou[s]es vs. older hou[z]es, but also that some nouns have been drawn into this class even though historically they showed no voicing, as with older dwarfs vs. innovative dwarves. Partial leveling is hard to show with these English plurals since there is so little to change, but English verbs provide an example, as when the consonantism of Old English ceosan ‘to choose/INF’ versus coren ‘chosen/PPLE’ was leveled, giving a uniform consonantal shell, but the vocalism was not, cf. modern choose/chosen.

It is argued here that the solution to these kinds of problems can be found if one recognizes a basic truth about analogy, namely that it operates entirely on a local level, and not in the global (grammar-wide) manner that many views of analogy (traditional, early generative, OT) would have one believe it does. This is not a new idea, but it bears resuscitating and applying here. It was observed by Jespersen 1949, with his notion of “local efficiency” in change (discussed also by Dressler 2003), by Tiersma 1978, 1982, who talks of “local markedness”, by Joseph and Janda 1988 (and elsewhere), who spoke of “local generalizations”, by Vennemann 1990, who put forward a “hypothesis of local improvement on just one parameter”, by Myers 1999, working within an Optimality Theory (OT) framework, who said “analogy can be formalized within OT by combining three devices that are independently motivated … [including] parochial constraints (i.e. universal constraints parameterized to apply only to specific lexical items)”, and by others.

In this paper, I discuss this idea of analogy as a localistic process and phenomenon, and demonstrate the power of thinking (and acting) locally when dealing with analogical change. Beyond the English examples above and how to deal with them localistically, the key data to be adduced here comes from leveling tendencies in verb endings in later Greek, where, for instance, among other developments (Joseph 1980, 2009):

• -a- prevails as the theme vowel over e in most past tense person/number forms except the 2nd person singular, due to local analogy with the 3rd person singular

• the 1st person plural past nonactive ending –maste affects the 2nd person plural, giving —saste for earlier –este (< Ancient Greek –esthe)

• the 1st person and 2nd person plural past nonactive endings –mast-e/-sast-e affect the 3rd person, giving –ondust- in place of earlier –ondus-.

To some extent, the real question to be addressed in applying a localistic view of analogy is how much do speakers really know about their language? Clearly in some sense, they know everything, or at least everything they need to, but in another sense, they seem not to have access to all the information linguists do (or which linguists think they have access to), at least not all at the same time. A useful metaphor (Joseph 1992) is that speakers act as if they are in a fog – they can see clearly close at hand, i.e. localistically, but have a harder time seeing off in the distance. Thus, they can make connections among related forms and can regularize paradigms, but not on a global basis. They work from what is locally salient, e.g. connections between 2nd person and 3rd person forms (as adjacent cells in a paradigm), between present and past tense forms, singular and plural, and so forth.

References Jespersen, O. 1949. Efficiency in linguistic change. In O. Jespersen, Selected Writings. Tokyo: Senyo, 381-466.

Joseph, B. 1980. Watkins’ Law and the Modern Greek Preterite. Die Sprache 26.179-184.

Joseph, B. 1992. Diachronic explanation: putting speakers back into the picture. In G. Davis & G. Iverson (eds.) Explanation in historical linguistics, pp. 123-144. Amsterdam: John Benjamins.

Joseph, B. 2009. Greek dialectal evidence for the role of the paradigm in inflectional change. Morphology (Special Issue: Morphology Meets Dialectology, ed. by G. Booij, A. Ralli and S. Scalise) 19.1.45-57.

Joseph, B. & R. Janda. 1988. The How and Why of Diachronic Morphologization and Demorphologization. In M. Hammond & M. Noonan (eds.) Theoretical morphology: approaches in modern linguistics, pp. 193-210. San Diego: Academic Press.

Myers, J. 1999. Lexical Phonology and the Lexicon. Rutgers Optimality Archive (#330-0699).

Tiersma, P. 1978. Bidirectional leveling as evidence for relational rules. Lingua 45.65-77.

Tiersma, P. 1982. Local and General Markedness. Language 58.832-849.

Vennemann, T. 1990. Language change as language improvement. In V. Orioles (ed.), Modelli esplicativi della diacronia linguistica. Pisa: Gardini, 11-35.

ABSTRACT. In inflectional morphology, the comparative-historical linguistics fundamentally relies upon two concepts which are known as »regular sound change« and »morphological analogy«. These concepts restrict the range of devel-opments conceived of as possible in a language and therefore show the ways towards an explanatory analysis of the data. The interplay between the »regular sound change« and the »morphological analogy« was the theoreti-cal background of all the most impressive achievements in the field, such as the coherent explanation of the in-tricate distribution of allomorphs in the inflection of Slavonic nouns and verbs (Slavonic »palatalisations of ve-lars I through III«), in Germanic conjugation (»Verner’s Law«) or in the Sanskrit perfect tense (»Brugmann’s Law«). However, as far as our present day theoretical understanding of these operational principles of inflectional change is concerned, one cannot help noticing a clear difference. We possess a rich and at the same time restric-tive theory of phonological change (cf. most recently Ringe & Eska 2013: 78–104, N. Hill 2014). We know that sound change operates regularly in that a particular change affects all relevant phonological structures existing in the relevant language. Deviations from this principle are only possible if (a) the structure in question is subject to a different sound change with a complementary conditioning or (b) the relevant structure is part of a loan from a different language system or, finally, (c) the regular development is superseded by the »morphological analogy«. We still do not agree on causes of phonological change but the mentioned insights—deduced from a vast amount of empirical data—tell us how the phonological change operates in a natural language and, accord-ingly, what is possible and what is not possible in each particular case. In contrast to this, at present we do not yet possess a deep understanding of »morphological analogy« and its operating principles. Despite the central importance of this concept for research on change in inflection, it has never been established how the »morphological analogy« actually works. Here two different opinions are, most often only implicitly, held in contemporary historical linguistics. The first position assumes that the »mor-phological analogy« modifies inherited forms making them more similar to other inflectional forms (»mor-pheme-based approach«, s. most explicitly Andersen 2008, 2009, 2015, for more references cf. Fertig 2013: 71–76). According to the second position, in the course of the »morphological analogy« inherited inflectional forms are not merely modified but rather replaced by forms created entirely anew on a model already present in the grammar (»word-based approach«, cf. Hill 2007, 2010, Garrett 2008, Fertig 2013, 2015). It is obvious that »paradigmatic levelling« may be seen as a prototypical subtype of »analogy« according to former position but does not even exist as such according to the latter. In my talk I will try to establish which of these two approaches to the »morphological analogy« is closer to reality. The central point will be the question as to what kind of data may help to reach this goal. It is clear that such data may only be obtained in a situation in which the approaches in question generate results which are observably different. By comparing these results with the recorded history of the relevant language(s) we may decide which of the two theoretical positions on the »morphological analogy« is in a better agreement with it. I hope to have identified at least one situation in which the theoretical positions described above generate results which might be observably different. A decision between »analogy« as modification (by replacing mor-phemes) and »analogy« as replacement (of inflectional forms as a whole) is hardly possible on an instance of change in which the output differs from the input in one morpheme only (such as Old Russian 1sg. pres. ind. pek-u ~ 2sg. imp. pec-i ‘to bake’ → Modern Russian pek-u ~ pek-i). In such a case, »analogy« may be equally assumed ei-ther to have replaced this particular morpheme by one of its other allomorphs or the inflectional form as a whole (then following the model of such verbs as nes-u ~ nes-i ‘to carry’ etc.). The situation is different when the output of change differs from its input in more than one morpheme (cf. Old Norse inf. ró-a ~ 3sg. pret. re-ri ‘to row’ → Norwegian 3sg. pret. ro-dde). In such a case, the approaches under discussion make different predictions. »Analogy« as modification (by replacing morphemes or parts of mor-phemes in the relevant inflectional form) predicts that a whole set of different combinations of inherited and replaced or modified morphemes should be equally possible and, ideally, also attested. In particular, in the case of two morphemes being subject to replacement in an inherited inflectional form, one may expect intermediate output forms containing the first morpheme modified and the other unchanged (Norwegian †ro-re) or vice versa (Norwegian †re-dde). By contrast, »analogy« as whole word replacement predicts that in an inherited inflectional form with the given properties, either both morphemes remain unchanged or they are both replaced at one and the same time. This approach does not allow for intermediate output forms with only one morpheme modified and the other retained unchanged. I will argue that the evidence of languages with sufficiently documented history clearly points to the latter possibility, i.e. in a case of the output of change being different from the input in two morphemes, both are al-ways replaced at one and the same time and no intermediate forms occur. To demonstrate this, I will use two particularly well-documented instances of morphological change taken from verb inflection in two different branches of Indo-European. The first case study will deal with the past tense formation of the so-called »verba pura« in North Germanic, the second with the present tense inflection of the formerly reduplicated verbs in dia-lects of Lithuanian which belongs to the Baltic branch of Indo-European. In both cases, the relevant inherited inflectional forms are changed in both morphemes in the bulk of North Germanic languages resp. contemporary Lithuanian dialects. In both cases only forms with both morphemes preserved or both morphemes changed but no intermediate variants are attested. In my opinion, these findings constitute a strong evidence for the second theoretical position on »morphological analogy« which views this process as analogical extension of pre-existing inflectional patterns to new lexical material. Finally, I will briefly discuss further phenomena potentially providing evidence relevant to the discussion on »analogy« (the so-called »externalisation of inflection« as well as »group inflection« and the ways it may emerge) and briefly address the potential relevance of the conclusions reached in the talk to research into the synchrony of inflection (on which cf. Blevins 2003, 2006, 2015).

REFERENCES

Andersen, Henning. 2008. Grammaticalization in a speaker-oriented theory of change. In: Eythórsson, Thórhallur (ed.). Grammatical Change and Linguistic Theory. Amsterdam: Benjamnis. 11–44. Andersen, Henning. 2009. Noget om analogi. In: Therkelsen, Rita & Eva S. Jensen (eds.). Dramatikken i Grammatikken. Festskift for Lars Heltoft. Roskilde: Roskilde Universitet. 15–25. Andersen, Henning. 2015. Review of Analogy and Morphological Change by David Fertig. Diachronica 32. 115–120. Blevins, James P. 2003. Stems and paradigms. Language 79. 737–767. Blevins, James P. 2006. Word-based morphology. Journal of Linguistics 42. 531–573. Blevins, James P. 2015. Inflectional paradigms. In: Baerman, Matthew (ed.). The Oxford Handbook of Inflection. Oxford: OUP. 87–111. Fertig, David. 2013. Analogy and Morphological Change. Edinburgh: EUP. Fertig, David. 2015. Analogy and morphophonological change. In: Honeybone, Patrick & Joseph Salmons (eds.). The Oxford Handbook of Historical Phonology. Oxford: OUP. 205–218. Garrett, Andrew. 2008. Paradigmatic uniformity and markedness: Historical convergence and universal grammar. In: Good, Jeff (ed.). Linguistic Universals and Language Change. Oxford: OUP. 125–143. Hill, Eugen. 2007. Proportionale Analogie, paradigmatischer Ausgleich und Formerweiterung: ein Beitrag zur Typologie des morphologi-schen Wandels. Diachronica 24. 81–118. Hill, Eugen. 2010. A case study in grammaticalized inflectional morphology: origin and development of the Germanic weak preterite. Diachronica 27. 411–458. Hill, Nathan. 2014. Grammatically conditioned sound change. Language and Linguistics Compass 8. 211-229. Ringe, Don & Joseph F. Eska. 2013. Historical Linguistics. Toward a twenty-first century reintegration. Cambridge: CUP.

ABSTRACT. The western Himalayas are an ancient linguistic area in which Indo-European (IE) and Tibeto- Burman (TB), the two largest language families of Eurasia, have been in close and longstanding contact for more than three millennia (see Masica 1991; van Driem 2001). The region thus constitutes an ideal laboratory for studying the long-term effects of language contact between two typologically very divergent language families. However, detailed investigations of such contact phenomena presuppose detailed knowledge about the history of the languages that are spoken in the region. Unfortunately, our understanding of the linguistic prehistory of the western Himalayas is still very fragmentary. This talk aims at filling this gap and shedding light on the linguistic past of this area based on a historical study of selected languages from the region. The talk will draw on field data from poorly described and endangered TB minority languages that were collected by the author between 2010 and 2016. In the western Himalayas, the border between the IE and TB languages nowadays runs along the Himalayan mountain range. This border area is occupied by the so-called “West Himalayish” (WH) languages, a group of fifteen endangered minority languages that belong to the TB family and separate the Tibetic linguistic area, which spans over the northern flank of the Himalayan range and the Himalayan Plateau, from the Indo-Aryan linguistic area, which extends over the Himalayan foothills and the Gangetic plain.

There is no evidence for major migration movements and / or language shifts in the western Himalayas during the past few centuries. This suggests that the current linguistic situation has persisted for a long time. At the same time, there is reason to believe that the WH languages, which are nowadays only spoken by small-sized ethno-linguistic communities that are scattered over remote and inaccessible regions of the North Indian Himalayas, must have had a wider geographical distribution in the past. This study will adduce evidence for this hypothesis based on a historical study of the WH subgroup. In a first part, the talk will address the internal classification of the WH subgroup. Based on lexical and grammatical evidence, it will be demonstrated that WH consists of two main branches, a western branch (marked with symbol ○ in Figure 1) and an eastern branch (marked with symbol △ in Figure 1) (see Widmer forthcoming). This classification will then be compared with the contemporary distribution of WH languages. It will be shown that closely related WH languages are often separated from each other by territories occupied by Indo- Aryan or Tibetic communities, which corroborates the hypothesis that the WH subgroup was once spread over a wider territory. In a second part, the talk will reconstruct the former distribution of WH languages based on both linguistic and non-linguistic evidence. It will be demonstrated that (i) the eastern WH languages bear affinities to the language of Zhangzhung, a polity that once controlled major parts of the Himalayan plateau before it was conquered by the Tibetan empire in the 7th century CE (see Denwood 2008), and that (ii) numerous western Tibetic varieties contain both lexical and morphological borrowings of WH origin. This suggests that WH languages were once spoken in major parts of the Himalayan Plateau. It will further be shown that several WH communities were highly mobile societies until the recent past and used to migrate to the Himalayan foothills together with their livestock in winter (see Zoller 1983; Willis 2007). Speakers of the WH language Jangrami (no. 7 in Figure 1), for example, used to spend the winter months in the valley of Kullu, where the closely related WH language Kanashi (no. 4 in Figure 1) is spoken to the present day. This suggests that WH communities may once have permanently settled the Himalayan foothills, but were then gradually pushed into the Himalayan range by Indo-Aryan communities, who may have arrived in the area as early as 1,500 BCE (see Masica 1991). Finally, the talk will point to directions for future research and identify further languages that may have been in contact with WH in ancient times, e.g. the languages of the Karakoram (e.g. Burushaski, Dardic languages) and the languages of the Central Himalayas (e.g. Tamangic). The talk will thus contribute to a better understanding of the linguistic prehistory of the western Himalayas and demonstrate the value of endangered, underdescribed minority languages for the study of ancient linguistic areas. 

References 

Denwood, Philip. The Tibetans in the West. Journal of Inner Asian Art & Archaeology 3, 7–21. 

Masica, Colin P. 1991. The Indo-Aryan languages (Cambridge Language Surveys). Cambridge: Cambridge University Press. 

van Driem, George. 2001. Languages of the Himalayas: an ethnolinguistic handbook of the greater Himalayan region, 2 vols. Leiden: Brill. 

Widmer, Manuel. Forthcoming. A grammar of Bunan (Mouton Grammar Library 71). Berlin: de Gruyter Mouton. 

Willis, Christina. 2007. A descriptive grammar of Darma: an endangered Tibeto-Burman language. Austin: University of Texas at Austin dissertation.

Zoller, Claus Peter. 1983. Die Sprache der Rang pas von Garhwal (Raṅ Pɔ Bhāsa): Grammatik, Texte, Wörterbuch. Wiesbaden: Harrassowitz.

ABSTRACT. This talk presents a case study of Chuvash, an endangered language that plays a key role in historical Turkic linguistics. Chuvash is a Turkic language spoken by approximately 1 million speakers in the Middle Volga region. According to UNESCO's Atlas of the World's Languages in Danger, Chuvash is a vulnerable language because of the dominant position of Russian in the region (Salminen 2010: 42). Recent sociolinguistic studies (Alòs i Font 2016) indicate very rapid and extensive language shift in the Republic of Chuvashia. From 2002 to 2010, Chuvash lost about a quarter of its speakers. The Chuvash people are an ethnic majority in the cities and towns of Chuvashia, but only around 1 percent of the republic's urban children speak Chuvash with their parents (Alòs i Font 2015a). Language shift has also spread to the rural Chuvashia (Alòs i Font 2015b), not to mention Chuvashpopulated areas outside the republic (Dolgova et al. 2004: 78-81). The endangered Chuvash language is the only living representative of the Bulgar branch, the earliest offshoot of Proto-Turkic (PT), which is in many respects opposed to the Common Turkic (CT) languages. Evidence from Chuvash is of vital importance in reconstructing Proto-Turkic, particularly its phonology. Chuvash represents characteristic features of the Bulgar branch, such as two types of rhotacism (PT *ŕ > CT /z/, Bulg. /r/; PT *δ > Bulg. /r/ with /j/, /d/, /t/ and /z/ in different subgroups of CT), lambdacism (PT *λ > CT /š/, Bulg. /l/), the “Bulgar palatalization” (PT *s- > Bulg. /š-/ and PT *t- > Bulg. /č-/ in certain contexts) etc. (Dybo 2010; Róna-Tas & Berta 2011). These correspondences provide a more complete reconstruction of the Proto-Turkic phonological system.

As a near-native speaker of Chuvash, I will discuss the role of Chuvash dialect data in studying Turkic historical phonology. For a century, Chuvash-Common Turkic comparative studies have predominantly used data from Standard Chuvash, which is phonologically simpler and much more innovative than Chuvash dialects. I will argue that, because of some crucial methodological shortcomings, dialect records from the 20th century are hardly suitable for comparative purposes. Therefore, from a historical perspective, documentation of Chuvash dialects is still a highly relevant research task.

From 2011 to 2015, I was involved in fieldwork on the endangered dialects of Chuvash conducted by the Institute of Linguistics of the Russian Academy of Sciences (Moscow) and the Chuvash State Institute for the Humanities (Cheboksary). I had the opportunity to analyze fieldwork data from the very specific Northwestern dialect area (Savelyev 2015) and to carry out my own fieldwork on other Chuvash dialects in Chuvashia and a Chuvash enclave in Tatarstan (Savelyev 2013, 2014).

The Northwestern dialect (NW) represented most clearly by the speech of the village of Maloye Karachkino in Yadrinsky District, Chuvashia, has long attracted attention from scholars due to its striking peculiarities (Ašmarin 1898). The recent field studies confirmed and advanced Mudrak's hypothesis (1993) that the speech of Maloye Karachkino represents a highly distinctive Chuvash dialect, which may have separated from the Bulgar stock already in the Middle Ages. For instance, the NW dialect retains PT *ẹ and *ö that developed into /i/ and /ü/ respectively in all the other varieties of Chuvash, cf. PT *gẹr- ‘necessary’ > Chuv. kirlə, NW kẹrlə id.; PT *t.rü ‘law, custom’ > Chuv. türə, NW törə ‘judge’. In certain contexts, this dialect gives specific reflexes of PT *-γ-, which cannot be deduced from the common Chuvash system, cf. PT *eγ- ‘to bind’ > Chuv. av-, NW aj- id.

My fieldwork on the Anatri dialect of Chuvash revealed some archaic phonological features that cannot be found in Standard Chuvash but are consistent with the reconstruction of Proto- Turkic. To give an example, in a variety of Anatri, I detected labialized reflexes of the PT rounded vowels giving (in general) unrounded variants in Chuvash, cf. PT *tut- 'to grasp' > Chuv. tït-, Anatri Chuvash langauge; Middle Volga region tʉt- id. Using these and other examples, I will demonstrate how less-studied dialect data can support the reconstruction of a proto-language. Given the binary structure of the Turkic family, Chuvash data turn out to be crucial in studying its linguistic history. In a broader context, an accurate reconstruction of Proto-Turkic allows a more reliable inference regarding the place of Turkic among such families as Mongolic, Tungusic, Japonic and Koreanic (Robbeets 2005).

References:

Alòs i Font, H. 2015a. Otnošenie gorodskogo naselenija Čuvašskoj Respubliki k ispol'zovaniju gosudarstvennyh jazykov. In: I.I. Bojko, A.V. Kuznecov (eds.). Issledovanie jazykovoj situacii v Čuvašskoj Respublike. 48–89.

Alòs i Font, H. 2015b. Etnojazykovaja situacija v sel'skih rajonah Čuvašskoj Respubliki (po dannym sociolingvističeskogo oprosa škol'nikov Morgaušskogo i Alikovskogo rajonov). In: In: I.I. Bojko, A.V. Kuznecov (eds.). Issledovanie jazykovoj situacii v Čuvašskoj Respublike. 136–163.

Alòs i Font, H. 2016. The Chuvash Language in the Chuvash Republic: An Example of the Rapid Decline of One of Russia’s Major Languages. In: M. Sloboda, P. Laihonen, A. Zabrodskaja (eds.). Sociolinguistic Transition in Former Eastern Bloc Countries: Two Decades after the Regime Change. Prague Papers on Language, Society and Interaction. Frankfurt am Main: Peter Lang, 2016. 51–73.

Ašmarin, N.I. 1898. Materialy dlja issledovanija čuvašskogo jazyka 1-2. Kazan': Tipo-litografija Imperatorskogo universiteta.

Dolgova, A.P. et al. 2004. Simbirsko-saratovskie čuvaši. Čeboksary: Čuvašskij gosudarstvennyj institut gumanitarnyh nauk.

Dybo, A. 2010. Bulgars and Slavs: Phonetic Features in Early Loanwords. In: E. Mańczak-Wohlfeld and B. Podolak (eds.). Studies on the Turkic World. Kraków. 21-40.

Mudrak, O.A. 1993. Istoričeskie sootvetstvija čuvašskih i tjurkskih glasnyh: Opyt rekonstrukcii i interpretacii. M.: DAJMOND.

Robbeets, M. 2005. Is Japanese related to Korean, Tungusic, Mongolic and Turkic? Wiesbaden: Harrassowitz.

Róna-Tas, A. & Berta, A. 2011. West Old Turkic. Turkic Loanwords in Hungarian 1-2. Turcologica, 84. Wiesbaden: Harrassowitz.

Salminen, T. 2010. Europe and the Caucasus. In: C. Moseley (ed.). Atlas of the World's Languages in Danger, 3rd edn. Paris, UNESCO Publishing. 32-42.

Savelyev, A.V. 2013. Fonetičeskie osobennosti nizovogo dialekta čuvašskogo jazyka (po dannym govorov Nurlatskogo rajona Tatarstana), In: Materialy XIII meždunarodnoj konferencii “Aktual'nyje problemy dialektologii jazykov narodov Rossii”. Ufa. 105–111.

Savelyev, A.V. 2014. Ogublennyje /ɨ/ i /i/ v nizovyh govorah čuvašskogo jazyka. In: Vestnik KIGI RAN. № 1. 36–41.

Savelyev, A.V. 2015. Slovar' malokaračinskogo dialekta čuvašskogo jazyka (available online at http://lingvodoc.ispras.ru/dictionary/296/1/perspective/296/1/view).

Sergejev, L.P. 1992. Čăvaš dialektologijĕ: Verenű posobijĕ. Šupaškar: Čăvaš universitečĕn

ABSTRACT. Background and Introduction: The New Guinea region is home to around 1200 languages, around twenty percent of the world’s total (Foley 2000). As the large majority of these languages are poorly documented, little is understood about the history and classification of many languages in this region. Based on lexical resemblances in core vocabulary, McElhanon and Voorhoeve (1970) proposed the Trans New Guinea (TNG) phylum, emcompassing a large portion of the languages of Papua. Later work on the TNG hypothesis (Wurm et al. 1975) included a group of around 80 languages in Madang province. If this hypothesis is correct, this would make the Madang group one of the largest branches of the putative TNG phylum (Foley 2000). 

The current study concerns a group of around twenty languages that Ross (2006) terms the Croiselles linkage, a branch of the proposed Madang group. Using primary data from my own fieldwork on several endangered languages of Madang province, as well as previously published wordlists and dictionaries, I apply the comparative method to demonstrate that Croiselles is a valid genetic grouping. I propose an internal structure for the language family, based off of shared phonological innovations, and reconstruct lexical items for Proto-Croiselles. I then discuss the prospects for demonstrating a genetic link between Croiselles and the other languages of the proposed Madang group. 

Previous work on the Croiselles languages: A survey of the Croiselles languages was carried out by Johannes Z’graggen, who published comparative wordlists of many of the languages in the Madang region (Z’graggen 1980a-d). For many of the Croiselles languages, these wordlists are the only published materials available. Z’graggen (1971) applies lexicostatistical methods to classify the languages of Madang region into two main branches, which he terms the Madang superstock and the Adelbert Range superstock, together comprising the “Madang subphylum” of TNG. Ross (2005) makes a preliminary hypothesis about the classification of Madang languages through a comparison of pronoun forms alone, arriving at a group he calls the Croiselles linkage, comprised of languages from both branches of Z’graggen’s Madang subphylum. It is primarily the languages of Ross’ Croiselles linkage that I am concerned with in this study. One limitation of both Z’graggen’s and Ross’ classifications is that they do not rely on regular sound correspondences to establish cognacy, but assume cognacy based on subjective judgements of similarity. This is due in part to a lack of adequate descriptive data on many of the Croiselles languages. 

The current study: I use regular sound correspondences to reconstruct the Proto-Croiselles sounds system and vocabulary, and propose an internal subgrouping based on shared phonological innovations. The findings of this study support Ross' hypothesis that the Croiselles languages are related, but indicate a significantly different internal structure for the group than that proposed by Ross. 

This study differs from previous works in two important ways: (1) I employ the comparative method, the accepted standard method of establishing genetic relationship, to identify recurrent form-meaning pairs and establish regular sound correspondences. (2) This study benefits from a larger pool of data, including dictionaries and other materials made available in recent years, as well as primary data from my fieldwork. In Summer 2016, I gathered data on seven Croiselles languages, including two not previously recorded in the literature: Qkuan Kambuar and Yamben. All seven of these languages are endangered– Qkuan Kambuar especially so, with only a handful of elderly speakers. 

Significance of this study for historical linguistics: As home to a significant portion of the world’s languages, the linguistic history of Papua is of interest to historical linguistics working in all areas, as any theory of language change cannot be said to be complete unless it can account account for the patterns of change found in this region. The linguistic ecology of the Papua New Guinea, characterized by an extremely high density of languages in long-standing close contact, with almost no written record, is quite different from some regions where the comparative method has been applied more extensively, such as Western Europe and Polynesia, with more closely related languages spread over larger areas. This close contact of great time depth, along with prevalent multilingualism, has resulted in a situation where languages freely borrow from each other features that have been said to be resistant to borrowing, such as basic vocabulary and pronouns (Foley 2000). This situation makes Papuan languages an interesting test case for the application of the comparative method, as distinguishing between inherited and borrowed material can be especially challenging. This study adds to a growing body of work, such as Daniels (2015) on the Sogeram languages and Holton et al. (2012) on Alor-Pantar, which have successfully applied the comparative method to smaller families of Papuan languages– a necessary prerequisite to evaluating the validity and scope of proposed larger groupings, such as Madang and Trans New Guinea. 

References: 

Daniels, D. R. (2015). A Reconstruction of Proto-Sogeram Phonology, Lexicon, and Morphosyntax (Doctoral dissertation). University of California, Santa Barbara 

Foley, W. A. (2000). The Languages of New Guinea. Annual Review of Anthropology, 29, 357-404. 

Holton, G., Klamer, M., Kratochvíl, F., Robinson, L. C., & Schapper, A. (2012). The historical relations of the Papuan languages of Alor and Pantar. Oceanic Linguistics, 86-122. 

McElhanon, K. A., & Voorhoeve, C. L. (1970). The Trans-New Guinea phylum: Explorations in deep-level genetic relationships. Pacific Linguistics B-16. 

Ross, M. (2005). Pronouns as a preliminary diagnostic for grouping Papuan languages. Papuan pasts: Cultural, linguistic and biological histories of Papuan-speaking peoples, 15-65.

 Wurm, S. A., Voorhoeve, C. L., & McElhanon, K. A. (1975). The trans-New Guinea phylum in general. New Guinea Area Languages. Vol. 1. Papuan languages and the New Guinea Linguistic Scene, Pacific Linguistics C-38. 

Z'graggen, J. A. (1971). Classificatory and Typological Studies in Languages of the Madang District, Papua New Guinea. Pacific Linguistics C-19. 

Z'graggen, J. A. (1980a). A comparative word list of the Rai Coast languages, Madang Province, Papua New Guinea. Pacific Linguistics C-30. 

Z'graggen, J. A. (1980b). A comparative word list of the Northern Adelbert Range languages, Madang Province, Papua New Guinea. Pacific Linguistics C-31. 

Z'graggen, J. A. (1980c). A comparative word list of the Mabuso languages, Madang Province, Papua New Guinea. Pacific Linguistics C-32. 

Z'graggen, J. A. (1980d). A comparative word list of the Southern Adelbert

ABSTRACT. In reconstructing grammatical systems, it is not always clear whether differences found between sister languages represent shared innovations in some rather than loss and generalization in others. Furthermore, in the context where sister languages have been in intense contact with one another subsequent to their split, teasing apart changes due to contact versus those due to inheritance can be difficult. The interplay between these two sources of language change is explored here with regards to the pronominal prefix systems in a subset of Northern Iroquoian languages.  
 Among the various parts of the polysynthetic Northern Iroquoian verb, the set of pronominal prefixes stand to be the most elaborate and highly developed components of the morphology. These prefixes encode the core arguments of the clause and are categorized into agent, patient and transitive paradigms (see Mithun 2005). In general, pronominal prefixes are marked for person (1, 2 or 3), number (singular, dual, plural), gender 
(masculine, feminine zoic, neuter) in the third person, and clusivity in the first person. Additionally, there is a third person indefinite pronoun which refers to an underspecified, defocused third person; in some languages, this pronoun was extended to also refer to females. As Chafe (1999; 2002) points out, the same level of complexity is not shared by the entire Iroquoian family. Lake Iroquoian languages, consisting of two branches (Wendat as a separate branch and a Five Nations branch which includes Seneca, Cayuga, Onondaga, Oneida, and Mohawk), have more complex pronominal systems than those found in the other members of the family 
(Cherokee and Tuscarora).  
 The richness of these Lake Iroquoian systems raises the question of how this complexity developed over time. Chafe (1999) points to the development of gender in the third person as a factor in the elaboration of these pronouns in Northern Iroquoian. Fortunately, there are still other clues to explain how Lake Iroquoian developed this intricate system, namely data that originate from the historical corpus of Wendat. Wendat, also known as Huron, has been dormant since the mid-19th century. Despite the lack of modern speakers of the language, there is a large historical corpus of the language stemming from the strong missionary presence among the Wendat in the 17th and 18th centuries. Since Wendat constitutes its own branch in the family and has had close contact with Five Nations populations, it is no surprise that these historical data have allowed linguists to elucidate contact-induced changes in the other Lake Iroquoian languages (Lukaniec and Chafe 2016; Mithun 2013).  
 Within this documentation, specifically Potier’s 1745 grammar, we find that Wendat had an extremely elaborate set of pronominal prefixes. According to this grammatical description, there were a total of 87 different pronominal categories in Wendat, in comparison to the approximately 58 categories found in the Five Nations languages (Chafe 2015; Michelson 2015; Mithun Ms; Woodbury 2003). Thus, the fundamental question is whether or not the large disparity in the number of pronominal categories represents innovations in Wendat after its split from the Five Nations, the collapse of categories in the Five Nations languages, or a combination of these possibilities. Furthermore, how are these differences manifested across the pronominal system?  
 From examining the specific distribution of these pronominal categories across the Lake Iroquoian languages, it is clear not only that Wendat makes many more functional distinctions than the others, but also that there are patterns to the generalizations made in Five Nations pronouns. For third person non-singular agents acting upon first or second persons, for the most part, the Five Nations languages use the reflexes of the pronouns with third person indefinite agents, as evidenced by the forms in table (1). In other words, unlike Wendat, these transitive pronouns are underdifferentiated in the Five Nations languages, using what was originally (and still is in certain contexts) combinations with a defocusing, non-specific third person pronoun as agent. Yet, the presence of these distinctions in some languages and the lack thereof in others does not automatically point to either innovation or loss. However, when examining these patterns in connection with  discourse tendencies, it becomes clear that topicality plays an important role in the elaboration of pronominal categories. Mohawk and Oneida show the least amount of differentiation in third persons acting upon first or second persons. Since speech act participants, i.e. first and second persons, are more topical than third persons, distinct pronominal categories were never developed in these two languages. Instead, the original indefinite pronoun was generalized to refer to all third person non-singular agents. Similar to Wendat, Onondaga and Seneca developed more distinctions, specifically for third persons acting upon first or second persons, yet only when acting upon a singular patient. Again, these particular developments show that innovations are in line with topicality, in that singular participants are treated as more important, and obviously, more individuated than groups of participants.  

With respect to non-singular third persons acting upon third persons, as found in Table (2), different 
generalization patterns can be found. In particular, judging from Wendat cognates, we can tell that the Five 
Nations languages use third person non-singular forms that are not reflexes of the indefinite pronoun, but forms 
that take into account both number and gender. Although these transitive combinations do not show the same generalization with the indefinite pronoun, these patterns are still motivated by discourse topicality. Thus, when 
third persons act upon other third persons, these same third person non-singular agents receive greater 
differentiation because they are more topical referents than their third person patients. 
  Judging from the systems found in these Lake Iroquoian languages, aside from combinations with third person singular participants, the Proto-Lake Iroquoian pronominal system contained only the underspecified 
third person indefinite pronoun as a general non-singular pronoun acting upon first or second persons. Over 
time, certain Lake Iroquoian languages developed more distinctions in the third person non-singular transitive combinations, particularly with singular patients, as in Seneca and Onondaga. The development of more 
distinctions in Seneca and Onondaga does not, however, necessarily point to a different reconstruction of the 
family, but rather the effects of close contact between Wendat people and these groups in the 17th and 18th 
centuries, as found in other aspects of the grammar and lexicon (Lukaniec and Chafe 2016; Mithun 2013). 
Furthermore, with regards to 3NSG agents acting upon other third persons, the generalized Five Nations forms 
are cognate with the gendered non-singular pronouns in Wendat. These cognates suggest that these gendered 
non-singular pronominal categories developed first in Proto-Lake Iroquoian, with a later collapse of the category with the indefinite pronoun in the Five Nations languages. 
 In summary, generalizations in the Five Nations languages in comparison to the more developed paradigms 
in Wendat elucidate certain discourse preferences that have shaped the pronominal systems. Within the Lake 
Iroquoian languages, pronominal distinctions slowly developed with regard to topicality, with preferences for 
speech act participants over third persons, singular over non-singular referents, and agents over patients. Due to 
these preferences, certain pronominal distinctions never manifested themselves in the Five Nations languages. Wendat contains the most pronominal categories and these seem to have been innovated after its split from Five 
Nations. Again, topicality drove the development of these distinctions in Lake Iroquoian, and from there, these 
discourse-based motivations become generalized in Wendat, leading to further pronominal differentiation with 
non-topical referents. 
References 
Chafe, Wallace. 1977. The Evolution of Third Person Verb Agreement in the Iroquoian Languages. In Charles N. Li (ed.), Mechanisms of Syntactic Change, 493-
524. Austin: University of Texas Press.
_______. 1999. Florescence as a Force in Grammaticalization. In Spike Gildea (ed.), Reconstructing Grammar: Comparative Linguistics and Grammaticalization, 
39-64. Amsterdam/Philadelphia: John Benjamins.
_______. 2002. Masculine and Feminine in the Northern Iroquoian Languages. In N. J. Enfield (ed.), Ethnosyntax: Explorations in Grammar and Culture, 99-
109. Oxford: Oxford University Press.
_______. 2015. A Grammar of the Seneca Language. University of California Publications in Linguistics 149. Oakland, CA: University of California Press.  
Chafe, Wallace; and Michael K. Foster. 1981. Prehistoric Divergences and Recontacts between Cayuga, Seneca, and the Other Northern Iroquoian Languages. 
International Journal of American Linguistics 47: 121-142.  
Michelson, Karin. 2015. Gender in Oneida, in Gender Across Languages, Volume 4, ed. by Marlis Hellinger and Heiko Motschenbacher, 277-301 John Benjamins 
Press. 
Michelson, Karin; and Mercy Doxtator. 2002. Oneida–English/English–Oneida Dictionary. Toronto: University of Toronto Press.  
Mithun, Marianne. 2005. Beyond the core: Typological variation in the identification of participants. International Journal of American Linguistics. 71:445-472.  _______. Ms. Mohawk Pronominal Prefixes. 79 pp. 
Potier, Pierre. 1745. Elementa Grammaticae Huronicae. Huron Manuscripts from Rev. Pierre Potier’s Collection, Fifteenth Report of the Bureau of Archives for the 
Province of Ontario, ed. Alexander Fraser [1920], pp. 1–157. Toronto: Clarkson W. James.  
Rudes, Blair. 1999. Tuscarora–English/English–Tuscarora Dictionary. Toronto: University of Toronto Press.  
Woodbury, Hanni. 2003. Onondaga–English/English–Onondaga Dictionary. Toronto: University of Toronto Press.

ABSTRACT. Since the early 1990s there has been an increased concern for language endangerment in the field of linguistics (Hale et al. 1992). Especially in the last fifteen years, unprecedented financial support, technological advancements and more developed standards of practice have made it possible for linguists to document natural discourse and interaction in many endangered languages. In this presentation, I will highlight the importance of naturalistic data in enriching our understanding of language use and, consequently, the functional motivations underlying language change. Specifically, I will show that the development of the complex system of grammatical relations in Shiwiar, an endangered language of eastern Ecuador, can be unravelled by examining patterns of language use in interactional data collected during fieldwork between 2011 and 2016. The interaction of syntax, semantics and pragmatics is known to play an important role in the marking of grammatical relations in many languages (Croft 2003; Dixon 1979; Silverstein 1976), resulting in varying degrees of syntactic complexity in the marking of core arguments. Shiwiar is a Chicham language spoken by 1,200 people in the north-western Amazon. It is a highly synthetic language with both head and dependent marking. Grammatical relations are manifested through a complex system involving morphological coding (case marking on nouns and cross-referencing of core arguments on verbs) as well as syntactic behaviour (switch-reference and nominalisations). Interestingly, arguments behave differently depending on their person and number. In terms of pronominal indexing on the verb, Shiwiar distinguishes subject (S/A) and object (P/T/R). In finite clauses, subjects are always indexed, but only objects which are speech act participant (1st and 2nd person) are indexed. In ditransitive clauses, only one object is indexed: if there is more than one speech act participant object, the 1st person object is indexed preferentially. This suggests that there is a person hierarchy (1 > 2 > 3) at play in Shiwiar argument marking.

(1) wii nuatkuŋɡa, amiŋɡa waanak ajuɾatɲuithjamɨ̃.

[wii  [1SG  nua-t-ku-n=ka] 
                 marry-FUT-SIM-1SG=COND] 

[ami-n-ka               waa-na-k                          a-ju-ra-t-ɲu-it-hamɨ̃-i] 
[2SG-ACC-TOP     tinamou-ACC-RESTR     CAUS-eat-PFV-FUT-NMLZ=COP-1SG>2SG-DECL] 

‘If I marry you, I’ll only feed you tinamou (a bird).’ 

On the other hand, in terms of morphological flagging on nouns and pronouns, Shiwiar manifests a pragmatically conditioned alignment split: clauses with 1SG/3 subjects have nominative-accusative alignment with accusative-marked objects whereas in clauses with 1PL/2 subjects, speech act participant objects are accusative-marked and 3rd person objects are not marked for case (examples 4-5).  This split is much more difficult to account for hierarchically. 

(2) numin atʃikhjaj.

numi-n           atʃi-k-ha-i 

stick-ACC      grab-PFV-1SG-DECL 

‘I grabbed a stick.’ 

(3) numi atʃikmɨ̃.

numi         atʃi-k-mɨ̃-i 

stick         grab-PFV-2SG-DECL 

‘You grabbed a stick.’ 

The data from Shiwiar raises questions about the diachronic origin of this complex system of argument marking.  While differential object marking is common in the world’s languages, a split between 1SG/3 person and 1PL/2 person is rare.  Unlike the case of object indexing on verbs in Shiwiar, it is difficult to account for this pattern with a straightforward hierarchical explanation.  However, the motivation for this split becomes apparent when examining natural discourse in Shiwiar. 
A 21-hour corpus of Shiwiar collected over the past five years, including 11 hours of conversational and interactional data, reveals that the vast majority of clauses in discourse have first person singular and third person subjects.  Second person and first person plural subjects occur much more rarely, and have a similarly low frequency of occurrence.  Furthermore, in the cases when second person and first person plural subjects are used, the object of the clause is overwhelmingly a speech act participant.  Crucially, the frequency of occurrence of specific person/number combinations precisely mirrors the distribution of differential accusative marking in Shiwiar, suggesting that the alignment split may have been motivated by patterns of frequency in usage. 
Without natural discourse data, a usage-based motivation of morphological marking of grammatical relations in Shiwiar would not have been uncovered.  Examining discourse practices, stylistic behaviour and language contact (as shown by Mithun 2012) can shed much light on the historical development of these systems.  This highlights the central importance of collecting natural discourse and interaction data when documenting endangered languages. 
References 
Croft, William. 2003. Typology and Universals. 2nd edition. Cambridge: Cambridge University Press. 
Dixon, R. M. W. 1979. Ergativity. Language 55:59-138. 
Hale, Ken, Michael Krauss, Lucille J. Watahomigie, Akira Y. Yamamoto, Colette Craig, LaVerne Masayesva 
Jeanne & Nora C. England. 1992. Endangered languages. Language 68(1). 1–42. 
Mithun, Marianne. 2012. Core argument patterns and deep genetic relations: Hierarchical systems in Northern 
California. In Bernard Comrie, ed., Typology of Argument Structure and Grammatical Relations. Amsterdam: John Benjamins, 257-294.  
Silverstein, Michael. 1976. Hierarchy of features and ergativity. In Grammatical categories in Australian languages, ed. R. M. W. Dixon, 112-71. Canberra: Australian Institute of Aboriginal Studies. 

ABSTRACT. In 1983, academics in the creole world were introduced to the existence of a remarkable secret speech in Jamaica which has come to be known as the Maroon Spirit Language (MSL) (Bilby, 1983, 1992),  distinct from even the ‘deepest’ (i.e. most basilectal) Jamaican Creole. Called Diip Patwa, Diip Konchri Taak or Kramanti, it is the ritual speech used in the Koromanti Play of the Windward Maroons in eastern Jamaica, necessary for communication with the spirits of the ancestors since it is understood that this was how they themselves spoke. Since it matches in some of its features the ELACs (English-lexifier Atlantic Creoles) spoken in Suriname on South America’s north-eastern coast, it was assumed that Suriname was indeed its country of origin.  This likelihood was supported by the fact that following the Dutch takeover of Suriname from the British in 1667, numbers of the latter left for Antigua and Jamaica with their slaves, presumably therefore bringing it to the Antilles.  Smith (1984: 13) for example believes that “the resemblances are so striking that a relationship must be regarded as proved.” However, MSL is almost entirely only distinctive lexically and phonologically; its structure is that of Jamaican rather than that of Suriname’s Sranan or Ndjuka. It in fact shares some of its features with a number of other cryptolectal registers, British, American and Irish as well as Creole, suggesting a different—or perhaps modified—account of its origins. Whether MSL may alternatively represent a conservative retention of an earlier ELAC (English-lexifier Atlantic creole), or of Jamaican Creole itself, or have been introduced by Krio-speaking Maroons returning to Jamaica in the mid-19th century has yet to be determined and I shall address these briefly. This paper, however, will focus on a wide variety of unrelated and understudied cryptolects, themselves of considerable sociolinguistic interest, in order to argue that MSL is, in part, just such a register rather than the retention of an ELAC introduced into Jamaica from Suriname in the 17th century. These are Cant, Shelta, Boontling, Polari, Caló, Texas Pachuco, Texas Pogadi, Jive Talk and Nordlinn. Each is characterized by a number of techniques designed to create an opaque in-group lexicon, all of which are also found in MSL and with which they are compared; these include phonological word-disguise by epenthesis, insertion, reversal, &c., the introduction of foreign words, euphemism, rhyming and paraphrasis. Cant dates from 16th century England and was the disguised speech intended for criminal activity. Shelta is the ethnolect of the population of Irish origin known as the Minceirs, or Irish Travellers, and is widely spoken in Britain and the United States. Boontling emerged among the population of Boonville, California, and was extensively used between the 1880s and 1920s as the town’s private speech in its rivalry with neighboring communities in the Anderson Valley. Polari reached its heyday in the mid 20th century mainly in Britain but was also heard in other English-speaking countries. It was the in-group register used by gay males, and likely had its origins in the mediaeval Mediterranean Lingua Franca. Caló is the ethnolect of the Spanish Romani population; it consists of Romani items in the Spanish grammatical and phonological framework. It has contributed many items to American Spanish as a result of Spain’s solución americana, the expulsion of Roma to its American colonies during the 17th and 18th centuries. Pachuco was the in-group speech of the Pachucos or Tiriloneños, and consisted of disguised Spanish used among other reasons as a mark of group membership. Pogadi is the ethnolect of the Romanichals, the Romani people who came to North America from Britain mainly during the mid-19th century. Well represented in Texas, it matches Caló in consisting of Romani words in a non-Romani (English) framework. Jive Talk was popular amongst jazz musicians during the 1940s and produced a number of explanatory publications. Nordlinn was a conlang created in the late 1950s by the author and a friend while in highschool, and which is still spoken by them today. While (unlike the other cryptolects discussed here) it has a distinct grammatical structure, its processes of word-formation are the same.

Bilby, Ken, 1983. “How the ‘older heads’ talk: a Jamaican Maroon spirit possession language and its relationship to to the creoles of Suriname and Sierra Leone,” Nieuwe West- Indische Gids, 57(1/2): 37-88. Bilby, Ken, 1992. “Further observations on the Jamaican Maroon spirit language,” Paper presented at the 1992 annual meeting of the Society for Pidgin and Creole Linguistics, Philadelphia. Smith, Norval, 1984. “The epithetic vowel in the Jamaican Maroon spirit possession language compared with that of the Surinam Creoles,” Amsterdam Creole Studies 7: 13-19.

ABSTRACT. Of all the Germanic languages, only Gothic preserved the old mediopassive – and only in the present tense. In all the other languages, periphrastic constructions are used. The same periphrastic constructions exist in Gothic as well, but are deemed to perform a different function (Ferraresi 2005: 121-124). The talk will focus on the specific role of the periphrastic past tense as a possible 'entry point' for the ultimate eclipse of the mediopassive and the emergence of a new passive tense system that consists entirely of periphrases. Of course, the Gothic text is for the most part a translation, and in the talk, it is analyzed as a translation – not a slavish one, but one that, Greek influence notwithstanding, uses all the available instruments to convey the original meaning. A lot has been said on how the periphrastic form encompasses essentially the same range of passive meanings as the mediopassive one, but is superior because it is free from any ambiguity (Abraham 1992); however, the main point of interest for the purposes of this talk is the fact that the mediopassive paradigm was in fact suppletive when the Gothic writer needed to speak in the past tense (like in Jn 18:10 Seimon Paitrus habands hairu, uslauk ina jah sloh þis auhumistins gudjins skalk jah afmaimait imma auso taihswo; sah þan haitans was namin Malkus " Then Simon Peter having a sword drew it, and smote the high priest's servant, and cut off his right ear. The servant's name was Malchus", Lk 4:1 Iþ Iesus, ahmins weihis fulls, gawandida sik fram Iaurdanau jah tauhans was in ahmin in auþidai " And Jesus being full of the Holy Ghost returned from Jordan, and was led by the Spirit into the wilderness", and many others), and therefore had to be used in contexts where the present tense was impossible. This, in turn, was a powerful incentive to use the present periphrastic forms more amply. The hypothesis is corroborated by the fact that there are only 6 instances of the usage of the wairþan-periphrasis in the present tense, against 74 in the past tense: also, the vast majority of the wairþan-periphrases in the past tense are used to translate Greek passive aorists and often perform a similar function to that of the mediopassives, but, again, cannot be used in the present tense (like Mt 9:25 þanuh þan usdribana warþ so managei, atgaggands inn habaida handu izos, jah urrais so mawi." But when the people were put forth, he went in, and took her by the hand, and the maid arose", etc.). The situation in Gothic seems to mark the final stage of the demise of the mediopassive and the early stage of the emergence of a full-fledged periphrastic passive paradigm with two different verbs that marked whether the action was ongoing or resultant.

References Abraham W. (1992) 'The emergence of the periphrastic passive in Gothic', Leuvense Bijdragen 81, 1–15. Ferraresi, G. (2005). Word order and phrase structure in Gothic. Peeters.

ABSTRACT. When a language distinguishes the perfective/imperfective opposition, the distinction is not realized in all available tenses (Comrie 1976:71). A typical example of such phenomena can be found in Russian, where perfective is marked in past and future tenses, but not in present tense. That is, the perfective/imperfective distinction is neutralized in present tense (Labeau 2007). This sort of asymmetry is explained from a logical standpoint; anything perfective (i.e., any event already completed) cannot be a present situation. Furthermore, if a language does have a present perfective form (e.g. South Slavonic languages), its reference is not limited to present tense (Comrie 1976: 67-68).

If we look at the asymmetry from a different angle, it entails that only imperfective aspect is semantically available in present tense. Therefore, some languages have a simple present tense form, which indicates a variety of aspectual meanings such as imperfective, habitual, progressive, and iterative (Russian, Italian, etc), although other languages have the simple present tense indicates imperfective, while other aspectual senses are expressed by other constructions (English, Spanish to some extent, etc). In either case “imperfective in present tense” is considered to be “indistinguishable from a present tense”, as Bybee et al. (1994:141) state.

In the latter type of languages, the neutralization of perfective/imperfective opposition is usually manifested as incompatibility of so-called progressive markers with stative verbs. For example, English ‘be + ~ing’ cannot co-occur with stative verbs (*I am knowing French). However, there is an exception to this rule. When expressing a non-permanent state, one can say “Oh, he is being silly” or “She is being a teenager”. Both of the examples are perfectly acceptable, although a stative verb ‘be’ is in the so-called progressive construction. The difference between “he is silly” and “he is being silly” is that the former indicates a permanent/habitual situation unlimited to the present, while the latter indicates a temporary situation limited to the present moment. In other words, if the English grammar requires the present progressive forms, instead of plain present tense forms, in order for expressing a temporary situation, one can claim that the simple present form ‘is silly’ is in the process of losing the present imperfective function.

Even more systematic distinction between the present imperfective and simple present can be seen in two East Asian languages, Belhare, a Sino-Burmese language, (Balthasar 1996 and personal communication) and Japanese. Belhare distinguishes simple forms and imperfective forms in all tenses including present tense, where both dynamic and stative verbs can take the imperfective marking. The a variety of Western dialects in Japanese (such as Osaka and Fukuoka dialects) as well as Old Japanese show the present imperfective and simple present distinction.

For instance, in Fukuoka dialect, the so-called progressive suffix –yo:, co-occurring with an existential verb aru, indicates that a given situation is temporally. On the other hand, the simple present tense form of a stative verb indicates a permanent state (Watanabe 2008a, b).

(1) Ima yakyu:-ga ari-yo: Now baseball-NOM exist-yo: "Now, there is a basball game." (temporally state)

(2) Kawa-ga aru River-NOM exist “There is a river.” (permanent state)

In Osaka dialect, where the function of ‘progressive’ suffix has merged with simple present, –teru is obligatory for indicating a present state. That is, the suffix has become the imperfective marker. The original simple present forms, in turn, have become a mere habitual marking. In other words, stative verbs in simple present forms and in –teru forms are synonymous.

(3) santa kuro:su-wa i-teru. Santa Clause-TOPIC exist-teru "There is Santa Clause!"

In Old Japanese, a suffix –(ye)ri, which is derived from an existential verb ari, seems to have undergone the same change as Osaka dialect. Watanabe (2008c) has found that the so-called ‘kari conjugation’ of the adjectives in OJ is the present imperfective form of the adjectives, which initially indicated a temporally state (just as ‘being silly’ in English), while the plain finite form was for a permanent state (i.e., ‘is silly’ in English).

(4) kimi-ni yori-te-pa koto-no yupe-mo naku ari you-DAT depart-CONJ-TOPIC word-GEN accident-PART non-existing ari

"….from you, accident of words (misunderstanding) be non-existing…." (MYS 13.3288) (Watanabe 2008c:86)

However, the formal distinction between simple present and present imperfect can create an unstable synchronic system, since the semantic distinction of the two forms is subtle. This tension lead to the subsequent changes in Old Japanese (loss of –(ye)ri) and Osaka dialect (present imperfective taking over simple present).

To conclude, the formal distinction between simple present and present imperfect is, indeed, possible, as seen in two languages in East Asia, Japanese and Balhare. However, the rareness of such opposition is due to the subtle semantic distinction between the two forms. Such unstable synchronic systems can lead to reduction in the present tense paradigm in order to keep the equilibrium in the temporal system of a language.

References

Bickel, Balthasar. (1996) Aspect, Mood, and Time in Belhare. Studies in the Semantics-Pragmatics Interface of a Himalayan Language. Zürich: Universität Zürich

Bybee, J., Parkins, R., and Pagliuca, W. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. The University of Chicago Press: Chicago

Comrie, Bernard. (1976) Aspect. Cambridge: Cambridge University Press. Labeau, Emmanuelle (2007) ‘Et un, ou deux, ou trois ? Les temps-champions du compte rendu sportif depuis 1950’, in Labeau, E., Vetters, C. & Caudal, P. (éds) Sémantique et Diachronie du système verbal français. Cahiers Chronos 16, 203-233. Amsterdam / Philadelphia: Rodopi.

Watanabe, Kazuha (2008a) Tense and Aspect Systems of Western and Eastern Dialects in Japan. Paper presented at Dialect as testing ground at Methods IX; Leeds, UK

Watanabe, Kazuha (2008b) Quantitative Analysis of a Current Change in Japanese Aspect. Paper presented at New Reflection or Grammaticalization 4; Leuven, Belgium.

Watanabe, Kazuha (2008c) Tense and Aspect in Old Japanese: Synchronic, Diachronic, and typological perspectives. Doctoral dissertation: C

ABSTRACT.  Adding meaning to Indo-Aryan aspectual adverbials then and again

Introduction: We examine a subset of aspectual adverbials in Hindi & Nepali and provide an account of their meaning

which captures the inter-relations signalled by their compositional morphology. Specifically we examine Hindi phir

(ambiguously “then, after that” or “again”) and phir bhī (“still”, both temporal and concessive), as well as their Nepali

counterparts pheri and pheri pani. Of particular interest are the “still” adverbials which involve the combination of the

temporal Hindi phir/Nepali pheri with an additive/scalar particle Hindi bhī/Nepali pani, the latter having a distribution

partially overlapping with other additive/conjunctive/universal particles like Japanese mo, Malayalam um, Sinhala t (cp.

Slade 2011, Mitrović 2014, Szabolcsi 2015).

Though in languages like English still, then, again have no obvious connections, in Hindi & Nepali they do, and thus

ideally the morphological compositionality of these items can be related to a (at least partially) compositional semantics.

Likewise, items like English then have a variety of interpretations which are not always available to their crosslinguistic

counterparts. Such considerations drive this investigation of possible decompositions of aspectual adverbs into more basic

pieces, taking a cue from earlier investigations such as those of Ippolito (2004, 2007).

(1) John took a shower. Then he went to bed. [ordering then]

(2) John was sleeping at 10. He was still sleeping at midnight. [temporal still]

(3) John was sleeping all day. He was still sleepy the next day. [concessive still]

Hindi bhī and Nepali pani are additive/scalar particles often meaning something like “also, even”, e.g. Hindi Rām bhī so

gayā “Ram slept too” or “Even Ram slept.” These particles are also used with indefinites to form NPIs, e.g. Hindi koī

bhī, Nepali kohi pani “anyone” and free choice items in combination with relative pronouns, e.g. Hindi jab bhī “whenever”,

Nepali je bhae pani “whatever happens”.

The basic idea is to provide an explanation of the relationships between these aspectual elements via an compositional

account of their semantics. The account should capture the morphological relationship between phir/pheri “then, after

that; again” and phir bhī/pheri pani “still”. The relation, which is not indicated in languages like English or German, seems

surprising at first blush.

 Basic data: Hindi Rām phir yahā āyā can mean either “Ram came here again” OR “Then Ram came here”; Rām phir bhī

yahā̃ nahī̃ āyā is “Ram still isn’t here” (in the temporal sense of still); Rām guṇḍā hai, phir bhī mera dost hai is “Ram is a

villain, (but) still he’s my friend” (with the concessive sense of still).

Unlike English then, Hindi phir/Nepali pheri can only denote “temporal ordering” then, i.e. “after that”, not the “anaphoric”

then, i.e. “at that time” (as in I dream of the age of kings; people were nobler then). This sense of then would be expressed

with the separate lexical items tab (Hindi); uhile, tahile, taba (Nepali).

 Defining then , again , still : We begin with definitions for then, again, and still for English (adapting Ippolito 2007 on

again and still; note that t is part of the runtime of the event):

 Adding meaning to Indo-Aryan aspectual adverbials then and again

[Examples 4, 5, & 6 included in PDF version of the abstract from the ICHL program index]

All three items take as arguments a time t, an eventuality e, and a predicate P, and assert that it is true that at time t e is an

eventuality of type P. Then John snored, Again John snored, Still John snored all assert that there is an event of John snoring

which is true at time t. They differ in what they presuppose. All presuppose the existence of a temporally prior time t′.

Then additionally presupposes the existence of an eventuality e′ s.t. there is some predicate R s.t. it is true that there is an

eventuality e′ of type R at time t′. Again additionally presupposes the existence of an event e′ s.t. it is true that there is

an eventuality e′ which occurs at time t′ which is of type P (i.e. the same type of eventuality as the asserted eventuality).

Still like again posits a separate prior eventuality of type P and additionally posits a “supereventuality” e′′ with a run-time

of  s.t. e and e′ are subevents of e′′.

 Hindi & Nepali “then”, “again”, “still”: We assume that the meanings of Hindi phir, Nepali pheri (“then”, “again”) match

the English translations given in (4), (5). Either of these entries could be the starting point for phir bhī/pheri pani. For the

scalar meaning of bhī/pani we assume the basic definition in (7).

[Example 7 included in PDF version of the abstract from the ICHL program index]

For sake of ease, only considering the temporal “still” for the moment, and thus taking bhī, pani to be functioning as additive

particles [APs], the addition of an AP to “again” as per (4) adds a presupposition that there is an additional eventuality

e′′ which is of also of type P. This fails to be fully compositional, since it still lacks the requirement that eventualities e,

e′ be sub-events of e′′, and there is likely a role for language change here: roughly, the AP adds a presupposition of the

additional type P eventuality e′′, and in context e′′ should somehow relate to e, e′. That e′′ contains both e and e′ is one

such possible relation, and apparently the one which became conventionalised over time.

In fact, the same explanation can apply if we take (ordering) “then” as the base for phir bhī/pheri pani; the diachronic

fixing of the relation of the additional presupposed event to that of containment will end up subsuming R as also being

of type P (nothing in (4) precludes R from being identical to P). Further, given the ambiguity in both Hindi & Nepali,

we might even take (4) to be the underlying meaning of phir/pheri, with certain contexts imposing a felicity condition that

R=P resulting in contextual “disambiguation” of “again” from “then”. This in fact is an ideal outcome, showing a relation

between the two items that does not appear from English, German, &c.

For the concessive sense of “still”, we assume that the APs bhī, pani contribute their scalar reading:

[Example 8 included in PDF version of the abstract from the ICHL program index]

Here assuming that the AP combines with an item with essentially the meaning of (4), this adds a presupposition that

there exists not only an eventuality e′′ which (through diachronic change) has the containment relation to e and e′, but

also triggers the generation of a set of alternatives, each of which an alternative to eventuality e′′. Each of these alternative

“supereventualities” contain both the asserted eventuality e and an alternative to the presupposed eventuality e′. And, these

alternative eventualities are ranked on a scale of likelihood s.t. e′′ is less likely than any of the alternatives to e′′.

[Example 9 included in PDF version of the abstract from the ICHL program index]

 Remaining issues: In addition to ambiguous phir, Hindi also has the bimorphemic form phir se which is unambiguously

“again”. There is no immediately obvious explanation for why adding the instrumental/ablative postposition se to phir

should pick out the “again” sense. We can only gesture generally towards a potential answer. Though present-day Hindi

se lacks a comitative function, but this appears represents a later development, as se derives from Skt. sahita “accompanying”/

sahitam “together with” (Turner 1962:13310), and so the ablative and instrumental senses both seem to be later

developments, with a loss, at some point, of the comitative sense. In that comitatives would seem to have some sort of

additive function, it is possible that this is what resulted in the disambiguation in some fashion or other.

Also of interest is Sanskrit punar for similar reasons as Hindi phir and Nepali pheri:— its inherent ambiguity. Punar can

mean “back”, “again”, “still” and so might well be analysed as even more underspecified then Hindi phir/Nepali pheri, with

contextual constraints aiding in precise interpretation.

[References included in PDF version of the abstract from the ICHL program index]

ABSTRACT. Hittite, along with other Anatolian languages, possessed a conjunctive marker -(y)a (μ) that expressed not only conjunction but also universal quantification, free-choice inferences, and focal additivity.

We provide empirical evidence for the the particle -(y)a in Hittite as featuring the range of conjunction-related meanings mentioned above.

[Due to the complex nature of Hittite characters, the bulk of this abstract is only legible in its PDF version located in the ICHL program index]

References:

Bobaljik, J. D. (2012). Universals in Comparative Morphology: Suppletion, superlatives, and the structure of Words. Cambridge, MA: MIT Press.

Chierchia, G. (2013). Logic in Grammar: Polarity, Free Choice and Intervention. Oxford studies in semantics and pragmatics 2. Oxford: Oxford University Press.

Embick, D. (2010). Localism versus Globalism in Morphology and Phonology. Cambridge, MA: MIT Press.

Hahn, E. A. (1933). Light from Hittite on Latin indefinites. Transactions and Proceedings of the American Philological Association, 63:28–40.

Hoffner, H. A. and Melchert, H. C. (2008). A Grammar of the Hittite Language. Part 1: Reference Grammar. Languages of the Ancient Near East. Winona Lake: Eisenbrauns.

Marantz, A. (2011). Locality domains for contextual allosemy. Paper presented at the Columbia Lingusitic Society.

Mitrovic, M. (2014). Morphosyntactic atoms of propositional logic: a philological programme. PhD thesis, University of Cambridge.

Toosarvandani, M. (2014). Contrast and the structure of discourse. Semantics & Pragmatics, 7(4):1–57.

ABSTRACT.  

Why are there disjunctive particles in Sinhala & Dravidian relative-correlatives?

            -existential particles in non-existential environments

Brief: Relative-correlative constructions in Dravidian and literary varieties of Sinhala use a quantifier particle as a “closing particle” in the relative clause; such constructions often involve free choice interpretations. Unexpectedly, the quantifier particle involved is part of the disjunctive/existential group (used in forming disjunctions, indefinites, questions) rather than the additive/universal group, as opposed to the case in Hindi and other languages. I provide an analysis which treats these particles as variables over choice-functions carrying an antisingleton

presupposition, which accounts for their use in the formation of epistemic indefinites in Sinhala & Malayalam. In the case where such particles are internal to a relative clause, the quantificational force contributed by the relative pronoun produces a universally quantified environment out-scoping the existentially-bound choice function variables. This forms part of a larger effort to understand the nature of particles expressing what seem to be elementary logical operations (Szabolcsi 2010, 2015, Slade 2011, Mitrović 2014 &c.)

Overview: In many Dravidian languages and in certain varieties of Sinhala we find relative-correlative constructions [henceforth RCCs; see Lipták 2009 for general discussion of RCCs] which include a particle which elsewhere appears in the formation of disjunctions (and, in many of these languages, as in the formation of questions). In many cases, the relatives have the force of free relatives like English -ever forms, as in the Sinhala & Malayalam examples below, using the particles da and o, respectively:

[Table included in PDF version of the abstract from the ICHL program index]

Sinhala da/də and Malayalam ō belong to the class of “quantifier particles” [henceforth Q-particles] which appear in a wide range of syntactically-heterogeneous contexts which can be semantically unified in that they all involve alternatives, and these particles serve to quantify over these alternatives (cf. Slade 2011, amongst others). What is striking about Sinhala and Dravidian RCCs is not even so much that they use a Q-particle (though the use of particle in such syntactic constructions is indeed unusual) but that they apparently use the “wrong” Q-particle.

Cross-linguistically many language use morphemes from two series in a wide range of “quantificational” contexts, split roughly into additive/conjunctive/universal morphemes like Japanese mo and interrogative/ disjunctive/existential morphemes like Japanese ka (cf. Kratzer & Shimoyama 2002, Szabolcsi 2010, Slade 2011, Mitrović 2014 &c.):— following the terminology of Mitrović (2014), the μ-series and κ-series of Q-particles, respectively. Sinhala da, Malayalam ō belong to the disjunctive/existential κ-series, with the μ-series additive/universal counterparts being t and um, respectively.

Examples of these particles da and ō appearing in interrogative and disjunctive contexts include Classical Sinhala

[Table included in PDF version of the abstract from the ICHL program index]

Given that the Sinhala & Dravidian RCCs frequently have a free choice -ever flavour, we might expect them to employ a Q-particle from the μ-series which orients them with universal-quantification as do free choice items [FCIs] in other languages like English what-ever (cp. universal ever-y) or Hindi relative jo bhī [relative “who”+“also,even” = “whatever”. Even more striking are Nepali relative pronoun+pani constructions, in which the additive/scalar particle pani occupies the position immediately following the right-edge of the relative clause — very like the Sinhala and Malayalam RCCs and unlike the Hindi or English constructions where the μ-series

element attaches directly to the wh-/relative-pronoun:

[Table included in PDF version of the abstract from the ICHL program index]

Epistemic indefinites[EIs]: In modern Sinhala and Malayalam, these particles also serve in the formation of epistemic indefinites (signaling some sort of ignorance regarding identity of entity in question):

[Table included in PDF version of the abstract from the ICHL program index]

However, Nepali pani like English -ever and Hindi bhī is also a member of the additive/universal μ-series. Sinhala da and Malayalam ō are both members of the disjunctive/existential κ-series and thus are quite unexpected in a FC construction.

Analysis of epistemic indefinites[EIs]: I suggest that da/ō can be treated as variables over choice-functions (cp. Hagstrom 1998, Cable 2007, Slade 2011 &c.); this, combined with existential closure, accounts for their functions in disjunctions and questions. The WH+də/WH+ō indefinites of Sinhala & Malayalam carry an additional pragmatic load beyond that of a plain indefinite, overtly signalling a lack of some sort of knowledge regarding the entity in question. These EIs are compatible with environments in which there is a individual which is not uniquely identifiable by either name or definite description. I posit that this correlates with reference to intensional entities (individual concepts in the simplest cases) and that “the lack of knowledge” represents an additional anti-singleton presupposition (later cashed out as a conversational implicature, cp. Alonso-Ovalle & Menéndez-Benito 2010) which requires that the cardinality of the set of choice functions which select an individual satisfying the requirements of its context (e.g. for which the proposition is true) not equal 1.

[Table included in PDF version of the abstract from the ICHL program index]

Relative-Correlatives appear in Classical and modern literary Sinhala, where the clause containing the relative pronoun must be followed by either the Q-particle də or the conditional particle nam. E.g. [ yam kumariyaki ohuj duṭuvā ] də/nam [ ōi ohuj kerehi piḷin̆da sit ætikara gattāya ] “Whichever princessi saw himj fell in love with himj”, which has two readings, a quantificational reading (“every princess who saw him fell in love with him”) and a referential reading (“the specific princess who saw him—whoever she may be—fell in love with him”) (for general discussion, see Bittner 2001, Brasoveanu 2012). The analysis of də/ō given above for EIs carries over here, except that the yam relative pronoun creates an environment with universal quantification over worlds or situations/perspectives out-scoping the existentially-bound choice function variables. (7) gives the denotation for the sentence under discussion for the referential “epistemic” reading; the quantificational reading would be similar, substituting situations/perspectives for worlds.

[Table included in PDF version of the abstract from the ICHL program index]

[References included in PDF version of the abstract from the ICHL program index]

ABSTRACT. This talk explores a number of ways in which perceptual mechanisms play a primary or secondary role in paradigm leveling and in other types of analogical change. I first consider several cases of ‘non-proportional’ paradigm leveling, changes that cannot be accounted for in terms of the application of an existing, non-alternating surface pattern because the leveling results in an inflectional pattern not found elsewhere in the language (cf. Paul 1886: 161; 1920: 116n.1; Hill 2007; Garrett 2008). Examples include a number of levelings of ablaut alternations in stem-forming suffixes in various Germanic noun classes, e.g. (unleveled) Gothic brōþar (NOM/ACC SG)– brōþrs (GEN SG) ‘brother’ vs. Old High German bruoder–bruoder; Gothic guman (ACC SG)–gumins (GEN SG) ‘man’ vs. Old English guman–guman. The perceptual mechanism that I propose to account for such levelings is akin, on the one hand, to what Paul (1886: 183) suggests for (unintentional) folk etymology, and on the other hand to the types of phonological reanalysis that Ohala (1981) labels “hypercorrection” and “hypocorrection”, the key difference being that here the innovative analyses are attributable to the biasing effects of related wordforms. I argue further that even where there is an existing ‘proportional’ model for a new, partially or completely leveled pattern, the same kinds of phonological reanalysis often play a secondary role, either by making some items more susceptible to leveling than others or by favoring one available model over another. The levelings and non-levelings of the alternations that resulted from open-syllable lengthening of vowels in German verbs – discussed at some length by Paul (1886: 166–167) – exemplify the first type. The related partial leveling in verbs such as lesen /leːzən/ (INF)–lĭst (2/3SG) → liest /liːst/ ‘read’ is a good example of the second. Finally, I examine other types of analogical change, including certain extensions of stem alternations (e.g. German wissen (INF)–wiste (PST SBJV) ‘know’ → wissen–wüste) and shifts from one inflectional class to another, such as that involving the originally strong English verbs creep, leap, sleep, and weep, and argue that hyper- and hypocorrective phonological reanalysis plays exactly the same kind of role in these non-leveling developments as it does in the examined cases of paradigm leveling. My two main conclusions are: (1) The familiar conception of analogical innovation as a reflection of the normal operation of speakers’ productive grammars, whereby innovators are making their best guesses at forms for which they cannot access a stored representation, is only part of the story; guesswork at the level of phonological analysis of frequently encountered forms also plays a crucial role in many cases. (2) At neither level is the guesswork guided by any universal preference for ‘paradigm uniformity’, and paradigm leveling is thus not distinct in any fundamental way from other types of analogical change (cf. Garrett 2008). Rather, the biases that lead language users to come up with certain innovative analyses and innovative forms are very much a ‘system dependent’ matter, in Wurzel’s (1989) sense.

References Garrett, Andrew. 2008. Paradigmatic uniformity and markedness. In Jeff Good (ed.), Linguistic universals and language change, 125–143. Oxford: Oxford University Press.


Hill, Eugen. 2007. Proportionale Analogie, paradigmatischer Ausgleich und Formerweiterung: ein Beitrag zur Typologie des morphologischen Wandels. Diachronica 24. 81–118.

Ohala, John J. 1981. The listener as a source of sound change. In Carrie S. Masek, Roberta A. Hendrick & Mary Frances Miller (eds.), Papers from the Parasession on Language and Behavior, 178–203. Chicago: Chicago Linguistic Society.


Paul, Hermann. 1886. Principien der Sprachgeschichte, 2nd edn. Halle: Niemeyer.

Paul, Hermann. 1920. Prinzipien der Sprachgeschichte, 5th edn. Halle: Niemeyer.

Wurzel, Wolfgang Ullrich. 1989. In ectional Morphology and Naturalness. Boston: Kluwer.

ABSTRACT. The shift from ablauting, i.e. strong verbs to the weak class in the history of Germanic languages is often explained by the universal preference for uniform stem exponence and additive marking over stem changing processes as well as language-specific systemic pressures, i.e. the Class-preference Principle: the most type-frequent class of weak verbs wins out at the expense of the strong class in the long-run (e.g., MAYERTHALER 1981, WURZEL 1984, BITTNER 1996), compare (1) for the German verbs backen ‘to bake’ and weben ‘to weave’:

(1) strong > weak PRET, PP PRET, PP (a)‘to bake’ backen: buk, gebacken > back+te, geback+t (b)‘to weave’ weben: wob, gewoben > web+te, geweb+t

However, these accounts miss to explain the fact that many strong verbs have only partially regularized, i.e. they have undergone partial leveling, and what is even more interesting, this partial leveling does not consist in the extension of the weak pattern but in the extension of a strong but less complex ablaut pattern, exhibiting the vowel o in the preterite and past participle forms, compare (2) for German weben ‘to weave’ and heben ‘to heave’:

(2) strong > strong (uniform exponence) PRET, PP PRET, PP (a)‘to weave’ weben: wab, geweben > wob, gewoben (b)‘to heave‘ heben: hub, gehaben > hob, gehoben

This [oPRET = oPP]-pattern originally stems from the Germanic 2nd ablaut class (fliegen – flog – geflogen ‚to fly‘) and has been extented to more than 20 strong verbs in the history of German (cf. NOWAK 2010a,b , 2015). The direction of pattern extension involves two fixed steps: first, acquiring the o in the PP by interparadigmatic analogy with the most type-frequent past participle pattern of strong verbs, i.e. [PP = o], compare geholfen ‘helped’, genommen ‘taken’, gekommen ‘come’. And second, adapting the o in the preterite forms, resulting from intraparadigmatic leveling (extension of the new PP-vowel o, e.g., gehaben > gehoben  hub > hob) as well as from interparadigmatic analogy with the o-preterites of the 2nd ablaut class (e.g., flog ‘flew’, schloss ‘closed’).

The present talk aims to show that ablaut pattern extensions can be interpreted as frequency-driven partial regularization strategies in the sense of Werner’s MORPHOLOGICAL ECONOMY THEORY (cf. WERNER 1987, 1989, see also FENZ-OCZLON 1990, 1991), thus guarantying less frequent strong verbs to remain in the strong class in the case of German. In some cases, acquiring the [oPRET = oPP]-pattern is an intermediate step towards the weak class (e.g., ‘to bark’ bellen: ball – gebollen > boll, gebollen > bellte, gebellt). Data from other Germanic languages, especially Luxembourgish and Dutch which also document similarly motivated ablaut pattern extensions (cf. WERNER 1990, DURRELL 2001, NÜBLING/DAMMEL 2006, NOWAK 2015, NOWAK forthc.), will also be considered to round off the picture of partial leveling processes as well as the direction of leveling within strong verbs within a usage-based account of language in the sense of BYBEE’s network model (e.g., BYBEE 1988, 1995, 1996, BYBEE/MODER 1983).   References BITTNER, Andreas (1996): Starke ‚schwache’ Verben und schwache ‚starke’ Verben. Deutsche Verbflexion und Natürlichkeit. Tubingen. BYBEE, Joan L. (1988): Morphology as lexical organization. In: HAMMOND, Michael/NOONAN, Michael (eds.): Theoretical morphology. Approaches in modern linguistics. San Diego et al., 119-141. ― (1995): Regular morphology and the lexicon. In: Language and cognitive processes 10:5, 425-455. ― (1996): Productivity, Regularity and Fusion: How languages use affects in the lexicon. In: SINGH, Rajendra (ed.): Trubetzkoy’s Orphan. Proceedings of the Montreal Roundtable Morphonology: Contemporary responses (Montreal, September 30 – October 2, 1994), 247-294. ―/MODER, Carol L. (1983): Morphological classes as natural categories. In: Language 59:1, 251-270. DURRELL, Martin (2001): Strong verb Ablaut in the West Germanic languages. In: WATTS, Sheila et al. (eds.): Zur Verbmorphologie germanischer Sprachen. Tübingen, 5–18. FENK-OCZLON, Gertraud (1990): Ökonomieprinzipien in Kognition und Kommunikation. In: BORETZKY, Norbert/ENNINGER, Werner/STOLZ, Thomas (eds.): Spielarten der Natürlichkeit – Spielarten der Ökonomie. Beitrage zum 5. Essener Kolloquium über „Grammatikalisierung: Natürlichkeit und Systemökonomie“, 2. Band, 1. Halbband. Bochum, 37-51. ― (1991): Frequenz und Kognition – Frequenz und Markiertheit. In: Folia Linguistica XXV, 3-4, 361-394. MAYERTHALER, Willi (1981):Morphologische Natürlichkeit. Wiesbaden. NOWAK, Jessica (2010a): Im Spannungsfeld zwischen starken und schwachen Verben. Zur Entstehung einer "8. Ablautreihe" im Deutschen, Niederländischen und Luxemburgischen. In: Dammel, Antje/Kürschner, Sebastian/Nübling, Damaris (eds.): Kontrastive Germanistische Linguistik 206-209, Teilband 2. ― (2010b): On the emergence of an "8th ablaut class" in German and Dutch. In: Hüning, Mathias et al. (eds.): Journal of Germanic linguistics 22.4 - Special issue on Comparative linguistics: Dutch between English and German, 361-380. ― (2015): Zur Legitimation einer 8. Ablautreihe. Eine kontrastive Analyse zu ihrer Entstehung im Deutschen, Niederländischen und Luxemburgischen. Hildesheim: Olms. ― (forthc.): Ablaut refunctionalisation in German. In: Antje Dammel/Matthias Eitelmann/Mirjam Schmuck (eds.): Reorganising grammatical variation. Diachronic Studies in Retention, Redistribution and Refunctionalisation of Linguistic Variants. Amsterdam/Philadelphia: John Benjamins (Studies in Language Companion Series). NÜBLING, Damaris/DAMMEL, Antje (2006): The Superstable Marker as an Indicator of Categorial Weakness? In: Folia Linguistica XL/1–2, 97–113. WERNER, Otmar (1987): The aim of morphological change is a good mixture – not a uniform language type. In: RAMAT, Giacalone A. et al. (eds.): Papers from the 7th international Conference of Historical Linguistics. Amsterdam, 591–616. ― (1989): Sprachökonomie und Natürlichkeit im Bereich der Morphologie. In: ZPSK 42, 34-47. ― (1990): Die starken Präterita im Luxemburgischen: Ideale Analogie oder vergeblicher Rettungsversuch? In: German Life and Letters 43:2, 182-190.

ABSTRACT. In studying Ancient Greek and other ancient languages, one inevitably comes across the fact that many forms of lexemes are not attested in the existing record of the language. This poses a unique problem for Ancient Greek scholars as the vast majority of lexical reference books list words by a single form. In verbs, this form is the first person singular indicative in the root voice and aspect. Thus, a significant portion of the words listed in such reference books are unattested. In order to make these words easier to find, the compilers of such collections generally put the simplest possibility. For instance, Cunliffe lists aphikneomai, 'to arrive at an indicated place,' as the first person singular indicative middle despite the fact that such a word does not appear in any extant writings in the Homeric dialect. He bases this conclusion on forms of the verb that are present in existing writing.

The purpose of this proposed study is to determine likely possibilities for the actual form of such lexemes. To accomplish this goal, new linguistic methods will be applied to this elder discipline. Stump and Finkel liken verbal paradigms to Sudoku puzzles. Principle parts, or pieces of each paradigm necessary to determine the full paradigm, have long been a key component of language acquisition pedagogy, especially pertaining to languages such as Ancient Greek and Latin which are paradigm-rich. Stump and Finkel go on to establish the nature of their collaborative work on network morphology and computational linguistics.

Many research studies have used such methods involving writing DATR and KATR (computational languages designed specifically for lexical knowledge representation) scripts. Corbett and Fraser, for instance, were able to use this method to derive a more elegant description of Russian nominal inflection than was previously available. Thus, these computer inflectional models can be used to predict the inflection of unknown words from just a few principle parts.

This study will involve the creation of a DATR or KATR network morphology model for verbs in Homeric Greek. This model will then be used to input all known forms of words for which the desired first person singular indicative is unknown. Several unattested forms will be derived from the model and compared to the current consensus forms present in Greek scholarship.

ABSTRACT. This is in part a programmatic paper, outlining a methodology which will allow us to study historical word order change at the interface of syntax and pragmatics in a way that approximates real life historical change. The case study is the development of V2 in English, and its loss, which took place roughly between 1400 and 1650. Word order is deeply rooted in a native speaker’s command of his/her native language, and is among the more stable characteristics of language, even though there may be variation, e.g. for syntactic or pragmatic reasons. Word order change is often caused by language or dialect contact resulting from societal upheaval such as foreign invasions, or mass migration. Since language contact situations typically induce spontaneous second language (L2) acquisition, they potentially lead to a degree of disruption in the transmission of grammar and use that forms the input for the next generation of learners/speakers. Studying language change in historical texts allows description of and informed theorizing on the forces that first initiated and then propelled the changes, but our understanding of the processes of change is hampered by gaps and evidential restrictions in the historical record. Word order change takes place in the shifting dynamics of speech communities over time, in the case of English bringing various layers of language contact resulting in large-scale spontaneous second language (L2) learning, followed by L1 acquisition by new generations of speakers. Our method aims to resurrect these interactive processes, taking a multidisciplinary perspective to study the impact of language contact on processes of change over time, and placing the actuation of word order change in the interaction between speakers and listeners with different language backgrounds. The syntactic and pragmatic properties of the historical data form the basis for eliciting online data on the L2 acquisition by individuals. We simulate the small-scale, individual short-term changes in language arising under these experimental conditions, and aggregate them to societal interaction, and to scenarios for change. Our approach thus moves beyond the evidential restrictions of textual data, while at the same time staying very close to the detailed evidence that they do offer. V2 in early English is well-known to be of two types (e.g. van Kemenade 1987; 2012; Pintzuk 1999), with demonstrably different distributional properties. They are both subject to illocutionary and pragmatic distinctions. Type 1 is categorical (cf. German, Dutch) and involves V to C movement in questions, negative-initial clauses, and clauses introduced by correlative adverbs þa and þonne ‘then’. It involves inversion of all types of subject, and V-movement is restricted to main clauses.The second type of V2 involves other types of first constituent. The verb is in a lower position in the prefield, where pronominal subjects precede it, and nominal subjects typically follow it. V-movement in type 2 is not fully asymmetric. The following template summarizes:

[CP XP C [ FP Subject position 1 F [TP Subject position 2 T... [VP]]]] Vf (1) Vf (2) 1 Why would he deny him such a small thing 2 By that we may quite clearly perceive that…

The two types of V2 share the same initial position for non-subjects (van Kemenade 1997), and a prefield position for the verb, in Old English quite dominantly in main clauses. The trigger for V2 is taken to be in clause typing/illocution marking, and distinguishing main clauses from subclauses; the latter distinction was still somewhat blurred in Old English, which is developing from a paratactic to a hypotactic clausal organization. The two types of V2 are an artefact of information structural distinctions. The development in Middle English (following language contact with Scandinavian) shows up both convergence and divergence between the two types, suggesting changes in the trigger. The facts support the hypothesis that the trigger for type 1 V2 was first extended in Middle English to include a wider range of focussed initial constituents. Both types of V2 were lost over the 15th-16th centuries (following large-scale dialect contact and multilingualism). The loss of type 1 V2 (ca. 16th century) consists in the loss of V2 with initial constituents other than wh- and focussed initial negatives. Type 2 V2 converged in part with type 1, but was always more pragmatically ‘vulnerable’ in the sense that it overlapped with subject-initial clauses: pronominal subjects had always preceded the finite verb, as had many anaphoric nominal subjects. Type 2 V2 was lost (15th century) because its word order pattern converged with SVO word order. Our methodology makes the following assumptions: 1) interactions between speakers and listeners with different linguistic backgrounds lead to changes in grammar and pragmatics, depending on the language characteristics of both interlocutors 2) such changes in language systems can be aggregated to community level by considering communities as composed of individuals with some variation in their language systems 3) historical texts reflect such structural changes as affected by societal developments (e.g., changes in size and composition of different language communities due to foreign invasions or waves of migration). 4) Word order change and L2 acquisition tap into the same mechanisms that allow L2 learners to make and refine hypotheses on the basis of linguistic input. This applies in varying degrees to all contact situations that result in spontaneous L2 acquisition.

Steps Historical study: code the historical data for five stages (9th century, 10th century, 1100-1250, 1350-1500, 1500-1640): specialized software (Komen 2014a-b) takes parsed historical corpora, and codes the mapping of word order patterns with the types of first constituent, type of verb and type of subject. Regression analyses calculate the relative impact of syntactic and pragmatic factors at each stage. L2 study: focus on the learning mechanisms by which learners with a V2 L1 acquire: a) another V2 language (Dutch L1 speakers learning German L2). b) a language with strict SVO word order (Dutch L1 learning French L2). c) a mixed V2 language (Dutch L1 learning an artificial language, modelled on Old English grammar with a common Germanic lexicon). Experiments (with 14 year old Dutch students) include a) tracking L1 pragmatic transfer on German and French; b) interactive on-line production experiment by means of the confederate scripting paradigm, with stimuli based on historical material; c) off-line production experiment of narrative language use; d) On-line comprehension experiments elicit grammaticality judgments; e) classroom acquisition of artificial mixed V2 language. Correlate experimental results with quantitative data from historical study, to closely imitate the historically attested contact situations. Simulation study: Simulate the psycho-linguistic processes driving grammatical change in individual L2 learners, and their aggregation to whole communities over generations, by means of agent-based models (Bloem et. Al. 2015, Beeksma et. al. 2016). The simulation takes the data of the first two studies as input, in order to clarify the key mechanisms underlying speaker-listener interactions and societal language change. These simulations are then linked to specific historical scenarios. The ‘core assumptions’ for speaker and group interactions above are developed into an explanation for real-world change scenarios. I will present results for the historical study and pilot results for the simulation study.

References: Beeksma, M; de Vos, H; Claassen, T; Dijkstra, T; van Kemenade, A. A probabilistic agent-based simulation for community level change in different geographical locations. Ms. RU Nijmegen. J., A.P. Versloot & F.P. Weerman (2015). An agent-based model of a historical word order change. In Proc.of the 6th Workshop on Cognitive Aspects of Computational Language Learning (CogACLL). Association for Computational Linguistics. Kemenade, Ans van. 1987. Syntactic case and morphological case in the history of English. Dordrecht: Foris. Kemenade, Ans van. 1997. V2 and embedded topicalisation in Old and Middle English. In Ans van Kemenade & Nigel Vincent (eds.) Parameters of Morphosyntactic Change, 326−352. Cambridge: Cambridge University Press. Kemenade, A. van. 2012. Rethinking the loss of V2. Traugott, E. and Nevalainen, T. (eds.) The Oxford Handbook of the History of English. OUP, 1182-199. Komen, Erwin. (2014a). Cesax Homepage. http://erwinkomen.ruhosting.nl/software/Cesax/ Komen, Erwin. (2014b). CorpusStudio Homepage. http://erwinkomen.ruhosting.nl/software/CorpusStudio/. Pintzuk, Susan. 1999. Phrase Structures in Competition: Variation and Change in Old English Word Order. New York: Garland publishing.

ABSTRACT. Language change has long been observed to follow an S-curve, which can be mathematically modelled by the logistic function (Weinreich et al. 1968: 113; Kroch 1989; Denison 2003; Pintzuk 2003; Blythe & Croft 2012). This insight has led to the application of binomial logistic regression analysis to linguistic changes in various languages. The resulting regression models describe ongoing changes, and give an idea of the role of one or more predictors on a linguistic outcome variable (see Wolk et al. 2012 for an illustration), but they can, in principle, also be used for prediction. In this talk, the predictive power of binomial regression analysis for historical linguistics is put to the test. Three case studies are discussed in which mathematical modelling of the past leads to falsifiable predictions of a future synchronic state. The method involves fitting a binomial regression model to historical corpus data with time as the predictor, then calculating the estimated values for a future some time beyond the span of the original corpus, and then checking these estimates against observed values in another, comparable corpus with data from a later period. The case studies deal with changes in Late Modern Dutch in the hortative, in predeterminers and in complex prepositions, building on earlier quantitative studies by Van de Velde (2014, forthc.) and Vranjes (2012). It is shown that the future can be predicted with remarkable accuracy in the cases at hand. While the predictability might only work in cases where a number of language-external conditions are met, the results nevertheless refute excessively pessimistic views that have been expressed about the predictability of grammatical change. On a methodological level, the application of binomial regression will be extended from working with a binary outcome variable, as is customary in linguistics (Baayen 2008; Gries 2013; Speelman 2014), to frequency counts, where binomial regression is less often applied. This overcomes the problem of non-linearity, often precluding the use of linear regression.

Baayen, Rolf Harald. 2008. Analyzing linguistic data: a practical introduction to statistics using R. Cambridge: Cambridge University Press. Blythe, Richard & William Croft. 2012. ‘S-curves and the mechanisms of propagation in language change’. Language 88: 269-304. Denison, David. 2003. ‘Log(ist)ic and simplistic S-curves’. In: Raymond Hickey (ed.), Motives for language change. Cambridge: Cambridge University Press. 54-70. Gries, Stefan Th. 2013. Statistics for linguistics with R. A practical introduction. 2nd edn. Berlin: De Gruyter. Kroch, Anthony S. 1989. ‘Reflexes of grammar in patterns of language change’. Language Variation and Change 1: 199-244. Pintzuk, Susan. 2003. ‘Variationist approaches to syntactic change’. In: Brian D. Joseph & Richard D. Janda (eds.), The handbook of historical linguistics. Oxford: Blackwell. 509-528. Speelman, Dirk. 2014. ‘Logistic regression: a confirmatory technique for comparisons in corpus linguistics’. In: Dylan Glynn & Justyna A. Robinson (eds.), Corpus methods for semantics: quantitative studies in polysemy and synonymy. Amsterdam: John Benjamins, 487-533. Van de Velde, Freek. 2014. ‘Nederlandse predeterminatoren als levend fossiel’. Nederlandse Taalkunde 19: 87-103. Van de Velde, Freek. Forthcoming. ‘Understanding grammar at the community level requires a diachronic perspective. Evidence from four case studies’. Nederlandse Taalkunde. Vranjes, Jelena. 2012. Voorzetseluitdrukkingen in het Nederlands sinds de 16de eeuw: een diachroon corpusonderzoek. Ma-thesis, University of Leuven. Weinreich, Uriel, William Labov & Marvin Herzog. 1968. ‘Empirical foundations for a theory of language change’. In: Winfred P. Lehmann & Yakov Malkiel (eds.), Directions for historical linguistics. Austin: University of Texas Press. 95-188. Wolk, Christoph, Joan Bresnan, Anette Rosenbach & Benedikt Szmrecsanyi. 2013. ‘Dative and genitive variability in Late Modern English: Exploring cross-constructional variation and change’. Diachronica 30: 382-419.

ABSTRACT. Do tones change faster than segments?: perspectives from recent documentation of the Chatino languages (Zapotecan, Mexico) 

Linguists have noted that tones are the “first” (Josserand 1983:243) or “most salient” (Morey 2005:146) features to vary across related languages, and they “tend to change rapidly and in unexpected ways” (Ratliff 2015:249). Does this mean that tones change faster than segments? If so, why? This paper explores these questions by comparing four recently documented endangered or understudied Chatino varieties: Zenzontepec (Campbell 2014), Tataltepec (Sullivant 2015), Zacatepec (Villard 2015), and Quiahije (Cruz 2011). 

Monosyllabification has precipitated recent consonantal changes in Quiahije and Tataltepec, but otherwise a relative paucity of segmental changes across Chatino varieties suggests shallow diversification (Campbell 2013). Segmental correspondences are phonetically close (Table 1, shaded), but tonal correspondences less so. 

Table 1. Chatino cognates and reconstructions gloss 

                                        Zenzontepec     Tataltepec     Zacatepec     Quiahije     proto-Chatino 

‘salt’                                  teheʔ Ø              theʔ Ø          tiheʔ Ø          theʔ Ø          *teheʔ Ø 

‘copal incense’                 jánā HM              janá H          jānã́ MH         jna H            *janá *H 

‘jaguar’                             kʷītʃí MH             kʷtʃí H          kʷitʃǐ L͡H         ktʃi M͡H         *kʷìtsí *LH 

‘ant’                                  kʷitʲeeʔ Ø            kʷtʲeèʔ L      kʷitʲēēʔ MM     kʷtʲeʔ L͡H       *kʷi-tèèʔ *LL 

‘arm of’                            ʃikȭ M                  skõ̀ L           sikȭ MH         skõ MH          *sikõ̀ *L 

‘six’                                 súkʷa HØ             skʷa Ø         sukʷa Ø           skʷa M͡L         *súkʷa *HL 

‘night’                               telā M                 H̋talʲà H̋L     tilàLH̋ LLH̋        tla H͡LH̋         (*Htelà) *HL? 

‘raw’                                jéʔē HM               jeʔe H͡L       jaʔa Ø             jʔa Ø             (*jaʔa) ? 

 Proto-Chatino (based on Campbell & Woodbury 2010) contrasted *H vs. *L vs. Ø. Most words were bimoraic, and tones linked to morae one-to-one from the end of the word. *H was culminative (max. one H per word). Thus there were six tonal melodies on simplex bimoraic words (rightmost column of Table 1) that display regular correspondences in their reflexes. However, there are many irregular correspondences as well (last two rows of the table), for which tone cannot yet be confidently reconstructed (pCh words in parentheses). 

While the modern varieties maintain privativity (Ø) and H culminativity, their tone systems have diverged significantly in terms of inventory, distribution, and processes, as shown in Table 2 (H̋ = extra-high). 

 Table 2. Some features of some Chatino tone systems

                                                            Zenzontepec        Tataltepec             Zacatepec                    Quiahije 

Inventory of tonal

  contrasts on the mora                       H, M, Ø                 H, L, Ø, H̋, H͡L       H, M, L, Ø, L͡H, L͡H̋       H, M, L, H̋, Ø, M͡H̋, M͡H, L͡M, L͡H, H͡L, M͡L 

# of tonal melodies on the word           7                            8                           15                                 15 

Floating tones                                       ―                         H̋                            H, L, L͡H̋                       H, H̋, M͡H 

Tones that spread                                 H                           H, L                        H, H̋                              ― 

 Some mechanisms that have propelled such diversification are illustrated in Table 3. The perfective form of ‘cry’ (a) displays regular reflexes of pCh *LH. The potential mood prefix (b) carried a H tone that perturbed the stem tone and yielded various types of floating tones. Such derived melodies then associated with words of other classes in Zacatepec (c) and Quiahije (d), resulting in new lexical tone melodies and three additional tone correspondence sets from pCh *LH. No two varieties group the four forms in a fully corresponding way. 

Table 3. Tone change and new tonal correspondence sets 

gloss                     Zenzontepec              Tataltepec             Zacatepec             Quiahije              proto-Chatino 

a. PFV-cry            nkaj-ūná MH               ntj-uná H              nkaj-unǎ L͡H            j-na M͡H              *nkaj-ùná *LH 

b. POT-cry            k-unā M                      H̋kunà H̋L             kunàL LL                kna H͡LH̋              (*kH-una) 

c. ‘twenty’             kālá MH                      kalá H                   kalà LL                   kla M͡LM͡H            *kàlá *LH 

d. ‘tomato’            nkʷīʃí MH                    nkʷʃi̋ H̋                 nkʷi̋ʃì` L͡H̋LL             wʃi H͡LH̋               (*nkʷisi) 

 While tone change has been attributed to loss of laryngeal contrasts (Haudricourt 1954) or reanalysis of secondary parts of a tone’s complex composition as its primary one (Ratliff 2015:254), this paper argues that 

tone’s autosegmental nature, tonal perturbations via morphology (Donohue 1997:379), and analogy may contribute to tone’s predilection to change. Experimental work on Chinese languages (Cutler & Chen 1997; Sereno & Lee 2015; Weiner & Turnbull 2016) has shown that tones are less crucial for lexical processing than segments, which could open the door for tone to change more rapidly than segments. “Tone can do everything [and more] that segmental and metrical phonology can do” (Hyman 2011: 236), and since tone is not well represented among well-studied and major world languages—at least outside of Asia—tone is a good example of how the study of endangered (and lesser-studied) languages might inform our broader understanding of language change. 

References 

Campbell, Eric. 2013. The internal diversification and subgrouping of Chatino. International Journal of American Linguistics 79(3): 395–420. 

Campbell, Eric William. 2014. Aspects of the phonology and morphology of Zenzontepec Chatino, a Zapotecan language of Oaxaca, Mexico. Ph.D. dissertation, University of Texas at Austin. 

Campbell, Eric, & Anthony C. Woodbury. 2010. The comparative tonology of Chatino: A prolegomenon. Paper presented at the Society for the Study of the Indigenous Languages of the Americas, Annual Meeting, Baltimore, MD. 

Cruz, Emiliana. 2011. Phonology, tone and the functions of tone in San Juan Quiahije Chatino. Ph.D. dissertation, University of Texas at Austin. 

Cutler, Anne & Hsuan-Chih Chen. 1997. Lexical tone in Cantonese spoken-word processing. Perception & Psychophysics 59(2): 165–179. 

Donohue, Mark. 1997. Tone systems in New Guinea. Linguistic Typology 1: 347–386. 

Haudricourt, André-Georges. 1954. De l’origine des tons du vietnamien. Journal asiatique 242: 69–82. 

Hyman, Larry M. 2011. Tone: is it different? In: John Goldsmith; Jason Riggle & Alan C. Yu (eds.), The Handbook of Phonological Theory, 197–239. Chichester: Wiley-Blackwell. 

Josserand, Judy Kathryn. 1983. Mixtec dialect history. PhD dissertation, Tulane University. 

Morey, Stephen. 2005. Tonal change in the Tai languages of northeast India. Linguistics of the Tibeto-Burman Area 28(2): 145–212. 

Ratliff, Martha. 2015. Tonoexodus, tonogenesis, and tone change. In Patrick Honeybone & Joseph Salmons (eds.), The Oxford Handbook of Historical Phonology, 245–261. Oxford: Oxford University Press. 

Sereno, Joan A. & Hyunjung Lee. 2015. The contribution of segmental and tonal information in Mandarin spoken word processing. Language and Speech 58(2): 131–151. 

Sullivant, John Ryan. 2015. The phonology and inflectional morphology of Cháʔknyá, Tataltepec de Valdés Chatino, a Zapotecan language. Ph.D. dissertation, University of Texas at Austin. 

Villard, Stéphanie. 2015. The phonology and morphology of Zacatepec Eastern Chatino. Ph.D. dissertation, University of Texas at Austin. 

Wiener, Seth & Rory Turnbull. 2016. Constraints of tones, vowels and consonants on lexical selection in Mandarin Chinese. Language and Speech 59(1): 59–82.

ABSTRACT.  

A diachronic typology of the universal superparticle: an inter-genetic view

Background: We investigate the synchronic and diachronic status of the conjunctive

quantifier particle (μ) in three genetically independent language families: Japonic

(Old Japanese), Indo-European (Anatolian, Celtic, etc.), and Sino-Tibetan (Mandarin).

All three language families possessed (some still possess) a multifunctional conjunction

particle which can express not only conjunction (as English and) but also

universal quantification, existentially polar quantification, free-choice inferences,

and focal additivity.

Japanic: In Old Japanese (OJ; c. 8th ce), the [wh+μ] quantificational expressions were

confined to inherently scalar (σ) complements, i.e. either numeral nominals or inherently

scalar wh-terms (e.g. how-many/when), as Whitman (2009) first noticed.

The only two kinds of wh-terms which can serve as μ-hosts we find in OJ are intervaland

quantity-denoting wh-terms, i.e. those wh-abstracts with only a σ-domain of

alternatives. Non-scalar domain (δ) complements to OJ μ cannot be found – we claim

this was disallowed. In Classical and Early Modern Japanese (Cl/MdJ), existential

constructions arise. This semantic shift is given by the pair below.

[Examples 1 & 2 included in PDF version of the abstract from the ICHL program index]

Indo-European: It has been well-investigated (Mitrovic, 2014) that all old IE languages

had a multi-functional μ. In non-coordinate contexts, we discuss the diachronic nature

of the semantic split between those old IE language where μwh express a universal

quantifier and those languages where μwh denotes a polar existential. Under

the assumption that the lives of μ operators are diachronically constrained by universal

factors, OJ facts shed light onto Indo-European (IE) development of polarity and

quantification [wh+μ]. While all old IE languages show generalised wh-quantification

(cf. ??b), the vast majority of IE languages exhibit gramamticised polarity in [wh+μ]

forms (cf. Skt. kaˇsca, Goth. hwazuh, etc. ‘any(one)’). The more archaic Anatolian

and Celtic branches, however, show plain universals (Hit. kuiˇsˇsa, OIr. cach ‘each(one)’),

which is consistent with the diachronic trends in Japonic.

[Examples 3 & 4 included in PDF version of the abstract from the ICHL program index]

Archaic Chinese: Archaic Chinese (AC) particle ji¯e (皆) is obligatorily distributive and

functions not only as a universal quantifier but also as a polar-existential quantifier

and conjunction marker. (Mitrovic and Hu, 2016) While we omit data here, we conjecture

a diachronic development of μ from AC to modern Chinese which has not underwent

as many semantic changes as μ in Japonic and IE did.

Interestingly, the μ particle in all three families eventually ended up with a quantificational

type of its hosts, i.e. ⟨⟨e; t⟩ ; t⟩.

A Diachronic-Typological Outlook: The analysis will provide uniform treatment of

μ particles in all three languages, including an explanation for the rise of polarity

sensitivity in EMJ that obtained by virtue of loss of restriction to scalar complementation.

The locus of historical change in the inferential procedure from a SI (‘not

every’) to a polar-sensitive expression (NPI/‘(not) any’) can thus be found in the featural

makeup of the μ particle and its featural change (grammaticalisation). In this

regard, the analysis we provide is based on Chierchia (2013). The properties we look

into are given in the table below

[Table included in PDF version of the abstract from the ICHL program index]

References:

Chierchia, G. (2013). Logic in Grammar: Polarity, Free Choice and Intervention. Oxford studies in semantics and pragmatics 2. Oxford: Oxford University Press.

Mitrovic, M. (2014). Morphosyntactic atoms of propositional logic: a philo-logical programme. PhD thesis, University of Cambridge. 

Mitrovic, M. and Hu, X. (2016). The grammar of distributivity in Ancient Chienese. Paper presented at the 9th International Symposium on Ancient Chinese Grammar. Berlin, 30/7.

Vovin, A. (2003). A Reference Grammar of Classical Japanese Prose. London: Routledge. 

Whitman, J. (2009). Hitei kobo to rekishi teki henka. Ms. NINJAL.

ABSTRACT. Inflectional morphology of the Finnish language presents a number of interesting instances of stem levellings. In particular, we investigate the extension of so-called vocalic stems to cells of paradigms occupied by consonantal stems. In our paper we attempt to illuminate the current, rapidly evolving state of the change and observe the morphological contexts conducive to the innovative forms. At present many innovative forms are emerging in previously unattested environments in a larger scale.

We will be looking at the following categories: - nouns and adjectives: stem alternation between nominative singular, oblique cases in singular, and partitive singular - verbs: infinitives in -a/-da/-ta and -e-/-de-/-te-, passives, imperative and potential mood, and participles

The diachronic change where vocalic stem variants have been introduced in the paradigm has long been observed in nouns, cf. (1)-(2):

(1) suksi 'ski': obsolete partitive sus-ta (attested dialectally), replaced with sukse-a, cf. e.g. genitive sukse-n (2) hapsi 'hair': obsolete partitive has-ta (attested dialectally), replaced with hapse-a, cf. e.g. genitive hapse-n

Currently a widespread variation exists with (3):

(3) veitsi 'knife' with both partitives attested: older veis-tä and innovative veitse-ä, cf. e.g. genitive veitse-n

Similarly with a small number of verbs, cf. e.g. in (4):

(4) kait-a 'to look after, take care of (a child)', which has been replaced with productive kaitse-a

Individual derivational suffixes behave differently while one would expect them to behave in a similar manner given that they share the same allomorphy, especially alternation in nominative singular -nen : oblique stem -se- : partitive stem -s-, cf. (5):

(5) NSG kärpä-nen, oblique stem kärpä-se-, partitive singular kärpä-s-tä 'fly'

We observe that the probability of an innovative form often relates to its morphological transparency in a sense that lexemes appearing to be non-derived have a higher probability of forming an innovative partitive, cf. (6)-(7).

(6) NSG nai-nen, oblique stem nai-se-, partitive singular nai-se-a 'woman' – current normative form nai-s-ta (7) NSG ihmi-nen, oblique stem ihmi-se-, partitive singular ihmi-se-ä 'human' – current normative form ihmi-s-tä

In contrast, lexemes that are more transparently derived rarely form innovative vocalic stems, cf. (8)-(9):

(8) ?puna-ise-a 'red', cf. puna 'red' (noun) – current normative form puna-is-ta (9) ?syy-llise-ä 'guilty', cf. syy 'reason, fault' – current normative form syy-llis-tä

Similarly we observe that transparently derived verbs only occasionally display innovative vocalic forms, cf. examples (10)-(11) with momentary -aise-:

(10) ?luk-aise-a 'read (cursorily, quickly)', cf. luke-a 'read' – current normative form luk-ais-ta (11) ?kisk-aise-a 'pull (rapidly), cf. kisko-a 'pull' – current normative form kisk-ais-ta

Unproductive suffixes seem, in contrast, to be able to form innovative vocalic stems, cf. (12)-(13). They are examples similar to (6)-(7) in the sense that their derived status is not as clear as, say, with (8) and (9).

(12) infinitive tarv-itse-a 'need': present passive tarv-itse-taan : past passive participle tarv-itse-ttu – current normative forms tarv-it-a, tarv-it-aan, tarv-it-tu (13) infinitive hark-itse-a 'ponder, consider' : passive hark-itse-taan : past passive participle hark-itse-ttu : also imperative 2PL hark-itse-kaa – current normative forms hark-it-a, hark-it-aan, hark-it-tu, hark-it-kaa

Nevertheless, we also observe that certain suffixes almost systematically exhibit innovative vocalic stems despite morphological transparency, cf. (14)-(15):

(14) superlative: suur-impa-a 'biggest', vanh-impa-a 'oldest', kov-impa-a 'hardest' – current normative forms suur-in-ta, vanh-in-ta, kov-in-ta (15) caritive -ton : -ttoma-: ole-ma-ttoma-a 'non-existent', pää-ttömä-ä 'headless', huole-ttoma-a 'careless' – current normative forms ole-ma-ton-ta, pää-tön-tä, huole-ton-ta

What can be observed is that there is a clear tendency towards extending the oblique stem to the partitive singular. In current forms, the stem allomorph in partitive singular is in some cases similar to the form of the nominative singular (mies : mies-tä), but in many cases there is no direct connection between the form of the nominative singular and partitive singular (e.g. NSG kuusi vs. partitive kuus-ta vs. oblique stem kuuse- 'spruce'). In terms of frequency, the partitive is very frequent, but in relative terms the occurrence of the oblique stem is higher due to the fact that other 12 Finnish cases in the singular and the entire plural paradigm display the oblique stem. Nominative singulars resist the analogy, which is not unexpected given the privileged status being the unmarked case, hence no reliable instances of *miehe 'man', *naise 'woman', *lapse 'child' (cf. mies, nainen, lapsi), etc.

Verbs evince a similar situation: the stem that serves as the model for the extension is the active present indicative stem. Given the markedness relationships between present : non-present, indicative : non-indicative, active : non-active, the directionality of the change is expected (for markedness, see e.g. Garrett 2008, leveling in verbs in general, see Albright 2010).

The change in nouns and adjectives can potentially be seen as instances of partial levelling as a particular kind of allomorphy is reduced, but it can be viewed also as extension of cases like kivi : oblique kive- : partitive kive-ä 'stone', where nominative singular contrasts with partitive and oblique stems. Hill (2007: 88) argues that proportional analogy is to be preferred to levelling unless clear cases are found where no morphological model can be established; in this case a model paradigm exists.

References:

Albright, Adam. 2010. Base-driven leveling in Yiddish verb paradigms. Natural Language & Linguistic Theory 28: 475–537.

Garrett, Andrew J. 2008. Paradigmatic uniformity and markedness: Historical convergence and universal grammar. Pp. 125–143 in: Explaining linguistic universals. Jeff Good (ed.) Oxford: Oxford University Press

Hill, Eugen. 2007. Proportionale Analogie, paradigmatischer Ausgleich und Formerweiterung: Ein Beitrag zur Typologie des morphologischen Wandels. Diachronica 24: 81–118.