What speakers know — or don't know — about history and about typology
In this paper I discuss the relationship, or lack thereof, between typology and historical linguistics. Much discussion about this relationship has focused on matters of methodology, e.g. regarding the connection between reconstruction and typology, but I argue here that such a focus is viewing things only from the perspective of the linguist. An alternative viewpoint is possible too, namely that of the speaker. I suggest that just as a key component of historical linguistics, namely history, is inaccessible to speakers — so that much of what historical linguistics is about is for linguists and not for speakers — so too is typology a pursuit more for linguists than for speakers. I develop these ideas with some examples from grammatical change in English, and explore as well several related notions, such as markedness and gradualness in language change. Ultimately, I attempt to answer where typological generalizations come from if speakers are unaware of typology and of history.
How do we get from individual behavior to typological patterns? The role of language change processes
Typologists frequently propose functional motivations for typological universals: a simple example is to motivate the crosslinguistically highly frequent SV word order (Greenberg’s Universal 1) in terms of a preference for topic to precede comment (Tomlin 1986:37). Such explanations link individual speaker behavior to crosslinguistic patterns which a speaker has no direct knowledge of. Hence the explanation must be due to an inherent property of human beings (cognition, social interaction, etc.). Yet how does individual speaker behavior lead to constrained typological variation? We model two hypotheses. The first is a widely-proposed one-step model, in which biases or preferences (which may be language-specific or due to general cognition) in either language acquisition or language use result in changes in community linguistic norms and ultimately typological diversity (e.g. Boyd and Richerson 1985; Smith 2009; Culbertson et al. 2012). The second is a two-step model, in which speaker behavior leads to linguistic innovations, and then community-level processes lead to the conventionalization of innovations in such a way that constrained typological variation emerges across languages (Croft 1995, 2000). We test the two hypotheses on their ability to model the grammaticalization of definite and indefinite articles and the resulting typological variation, using 51 language histories of articles and the typological distribution of the different stages of article grammaticalization in WALS. It turns out that a critical factor is whether likelihood of language change (article grammaticalization) is dependent on speaker population size. The one-step hypothesis is highly unlikely to adequately model the observed data, while the two-step hypothesis robustly models the observed data.
Towards a source oriented typology
While it is often assumed that the results of typological research can be used to test the validity of proposed historical reconstructions (Shields 2011, among others), the reverse does not usually hold. Despite the fundamentally diachronic orientation of Greenberg’s work, typological generalizations, such as typological universals and explanations thereof, usually refer to synchronic cross-linguistic patterns, not the actual diachronic processes that give rise to these patterns in individual languages.
Over the past decades, a number of typologists have raised the point that explanations of individual patterns should be based on these processes, rather than the patterns in themselves (Bybee 1988, 2006, and 2008, Aristar 1991). While more and more relevant data has been collected within grammaticalization studies and studies of language change in general, however, this data is not usually integrated into typological generalizations. Conversely, historical linguists do not usually address the implications of their findings for these generalizations.
This paper examines different types of cross-linguistic data on the diachronic development of various patterns described by some major typological universals pertaining to the encoding of number distinctions, alignment systems (for example, split ergativity and hierarchical alignment), alienable and inalienable possession, and word order. This data challenges existing generalizations about the relevant patterns in two ways:
Individual patterns are often explained in terms of general principles that assumedly lead speakers to create these patterns over time, for example economy or processing ease. Yet, these principles are postulated based on the synchronic properties of the relevant patterns, and they do not appear to play any obvious role in the actual diachronic processes that give rise to these patterns. These processes are rather motivated in terms of the properties of highly specific source constructions.
In many cases, individual patterns are assumed to reflect a single overarching principle, but they are actually a combined result of several particularized diachronic processes, which may have contrasting properties and may not be obviously amenable to a unified explanation. These facts call for a shift in perspective in typological research. Most typological generalizations are goal-oriented, in the sense that the development of particular cross-linguistic patterns is assumed to be motivated in terms of their synchronic properties. A thorough understanding of these patterns, however, requires a source oriented approach where qualitative and quantitative data are taken into account about what constructions actually give rise to what patterns, in what contexts, and through what mechanisms. This approach has a parallel, for example, in Evolutionary Phonology
(e.g. Blevins 2004, 2008).
Aristar, A. R. (1991). On diachronic sources and synchronic patterns: an investigation into the
origin of linguistic universals. Language 67, 1–33.
Blevins, J. (2004). Evolutionary phonology: the emergence of sound patterns. Cambridge: Cambridge
Blevins, J. (2008). Consonant Epenthesis: Natural and Unnatural Histories. In J. Good (Ed.),
Linguistic Universals and Language Change, pp. 79–107. Oxford: Oxford University Press.
Bybee, J. (1988). The diachronic dimension in explanation. In J. A. Hawkins (Ed.), Explaining
language universals, pp. 350–79. Oxford: Basil Blackwell.
Bybee, J. (2006). Language change and universals. In R. Mairal and J. Gil (Eds.), Linguistic
Universals, pp. 179–94. Cambridge: Cambridge University Press.
Bybee, J. (2008). Formal Universals as Emergent Phenomena: The Origins of Structure Preservation.
In J. Good (Ed.), Linguistic Universals and Language Change, pp. 108–21. Oxford:
Oxford University Press.
Shields, Kenneth, J. (2011). Linguistic Typology and Historical Linguistics. In J. J. Song (Ed.),
Handbook of Linguistic Typology. Oxford: Oxford University Press.
Affix Order, Multiple Exponence, and Morphological Reconstruction
Generally we assume that it is preferred for affixes to occur in the following order:
This paper examines two of the historical pathways that can lead to exceptions, with inflection closer to the root than derivation, or with contextual inflection closer to the root than inherent inflection (on these terms, see Booij 1996). Both pathways involve the creation of multiple exponence, the marking of a feature more than once within a word. For example, in (1), the morpheme glossed Aintr@ was originally an inflected auxiliary, which grammaticalized as a marker of intransitivity.
(1) k=nat …u-ren kalak-y v-e£-v-al-in (Dict 241b) Batsbi
boy(v/b).abs in-from city-dir cm-steal-cm-intr-aor
>The boy stole away from home to Tbilisi.=
Multiple exponence is seen in the two verbal markers v-, which indicate a masculine singular subject. I examine also remedies that languages can take regarding multiple exponents, where one exponent is a disorderly affix. Remarkably, there are a few examples of languages that have lost exponents in the preferred position, retaining instead a disorderly affix. Tentative explanations of this are proposed.
What does this mean for reconstruction? Some linguists have claimed that we can take morpheme order as an indication of the order of development of these morphemes, with those closer to the root developing before those at the periphery (e.g. Mithun 2000). Yet it is not reasonable to assume that languages develop all derivation before they develop any inflection, and inherent inflection before contextual inflection (Mithun acknowledges this). In some instances we have attested developments of the type root-infli-deriv-infli , with loss of the trapped infli , leaving root-deriv-infli (Harris & Faarlund 2006). It would be inaccurate to suppose that in such cases derivation necessarily developed before inflection. Similarly we find prefixes developing into suffixes and vice versa, both through multiple exponence. Some suggestions are made regarding the reconstructability of morpheme order and reconstructability of the order of grammaticalization of affixes.
Booij, Geert. 1996. Inherent versus contextual inflection and the split morphology hypothesis. Yearbook of Morphology 1995, ed. by Geert Booij and Jaap van Marle, 1-16. Dordrecht: Kluwer.
Harris, Alice C., and Jan Terje Faarlund. 2006 Trapped morphology. Journal of Linguistics 42: 289-315.
Mithun, Marianne. 2000. The reordering of morphemes. Reconstructing grammar: Comparative linguistics and grammaticalization, ed. by Spike Gildea. Amsterdam: John Benjamins.
Predictable grammatical constructions – a diachronic case from Greek
A number of recent papers combine the study of diachronic language change with a Construction Grammar (CxG) approach to language (e.g. Bergs & Diewald 2008, Fried 2009, 2013). Overall, these papers apply the CxG approach to diachronic data with the purpose of answering questions relating to language change, either specific empirical questions or theoretical questions. This paper, however, approaches the matter from the opposite perspective: it uses diachronic data to address a theoretical debate within CxG, namely the validity of nonpredictability in a gate-keeping function for constructions.
In traditional Berkeley Construction Grammar nonpredictability is a necessary requirement for a linguistic unit to qualify as a construction (e.g. Fillmore, Kay, O’Connor 1988; Goldberg 1995). The background for this requirement is that if the meaning of a linguistic unit can be predicted by combining other well-known patterns, then that linguistic unit is merely the result of straight forward semantic composition, and if formally that same linguistic unit is also an instance of another formal pattern, it does not need to be included in the grammatical inventory. Some recent descriptions within Construction Grammar (CxG) of grammatical constructions, however, take the view based on psycholinguistic research of how linguistic entities are stored in human cognition, that some fully predictable units are best understood as grammatical constructions (e.g. Goldberg 2006; Bybee 2013; Croft 2001).
My purpose is to provide evidence from diachronic linguistics for the view that some predictable units are entrenched in language in a way that makes them functionally and structurally equal to nonpredictable constructions, and therefore these predictable units should be considered grammatical constructions on par with the nonpredictable constructions.
As evidence for the predictability of some grammatical constructions I use the example of two Byzantine Greek future periphrases, [méllo (‘I am about to’) INFINITIVE] and [thélo (‘I want’) INFINITIVE], which differ with regard to predictability.
[thélo INFINITIVE] is formally nonpredictably. In order to master the construction, language users need to know a set of specific formal rules: 1) the variation in morphological forms of thélo in this construction is restricted and 2) the internal dependencies between thélo and the infinitive is strong. [thélo INFINITIVE] is also semantically nonpredictable, since the meaning of the construction cannot be deduced based on straight forward compositional semantics.
[méllo INFINITIVE] , on the other hand, is fully predictable: it does not display formal idiosyncrasies and its meaning can easily be deduced on the basis of compositional semantics.
Nevertheless, these two periphrases develop common syntactic patterns:
- retention of the INF [thélo/méllo INF] at a time when the INF was being replaced by a finite phrase in most contexts.
- embedding in subjunctive clauses [na (subjunctive marker) thélo/méllo INF].
- development of an impersonal phrase [théli/mélli (impersonal) + na (subjunctive marker) + FINITE NON-PAST].
These similarities show that despite the difference in predictability, [thélo INF] and [méllo INF] occupy parallel roles as grammatical constructions in Byzantine Greek, i.e. as units among the grammatical stock to be drawn from.
Bergs, A.; Diewald, G. (eds.). (2008). Constructions and Language Change. Berlin: Mouton de Gruyter.
Bybee, J. (2013). Usage-based Theory and Exemplar Representations of Constructions. In: The Oxford Handbook of Construction Grammar, T. Hoffmann and G. Trousdale (eds.). Pp. 49-69. Oxford: Oxford University Press.
Croft, William. 2001. Radical Construction Grammar – Syntactic Theory in Typological Perspective. New York: Oxford University Press.
Fillmore, C. J.; Kay, P.; O'Connor, M. C. (1988). Regularity and Idiomaticity in Grammatical Constructions: The Case of let alone. Language, vol. 64, 3. Published by Linguistic Society of America. Pp. 501–538.
Fried, M. (2009). Construction Grammar as a tool for diachronic analysis. In: Constructions and Frames. Vol. 2, 1. John Benjamins Publishing Company. Pp. 262-291.
Fried, M. (2013). Principles of Constructional Change. In: The Oxford Handbook of Construction Grammar, T. Hoffmann and G. Trousdale (eds.). Pp. 419-436. Oxford: Oxford University Press.
Goldberg, A. E. (1995). Constructions: A Construction Grammar Approach to Argument Structure. Chicago: The University of Chicago Press.
Goldberg, A. E. (2006). Constructions at work. The nature of generalization in language. Oxford: Oxford University Press.
Thomas F. Shannon (University of California, Berkeley)
Historical Linguistics and Typology: A Case from West Germanic – and Beyond
This study contributes to the discussion of the relationship between typology and historical linguistics by examining one neglected phenomenon: the order of pronominal objects and nominal subjects, specifically in (West) Germanic languages, and somewhat beyond. The typology based on the order of subject, verb, and object is well-known, but it only involves nominals, not pronominals (Dryer 2005, 2007). The question remains open regarding synchronic and diachronic crosslinguistic regularities for pronominals.
For instance, in modern German and Dutch, subjects normally preceed objects: in main clauses subjects usually preceed the finite verb, and even after the finite verb, they usually occur before objects. Postverbally, however, pronominal objects can be placed before nominal subjects (POP: Pronoun Object Preposing). In fact, German preposes pronominal objects more frequently than Dutch: 60% to 10% (Shannon 1997). Cf. (1–2): both orders are possible in German but POP is unacceptable in Dutch.
Da hatte a) der Lehrer es ihm/ b) es ihm der Lehrer doch sehr deutlich gezeigt.
there had the teacher it him.dat /it him.dat the teacher prt very clearly shown
Toch had a) de leraar het hem/ b) *het hem de leraar heel goed laten zien.
yet had the teacher it him/it him the teacher very good let see
‘Yet the teacher had shown it to him very clearly.’
Moreover, POP has been declining in Dutch (Shannon 2005): 16th century (85%) vs. 20th (14%). As we will show, German is following a similar trajectory, but more slowly, whereas Afrikaans has completely lost POP. However, 19th-century Yiddish preserved POP surprisingly well (97%).
Doherny (2012: 4) notes, “both synchronic and historical single-language studies commonly raise research questions and suggest hypotheses that could be addressed with a diachronic study or a broader comparative investigation…” Several such questions and hypotheses emanate from the above. Regarding synchronic typology, how frequent is POP crosslinguistically, what features correlate with it? While a major study is impossible here, we will see that at least one VSO language, Biblical Hebrew, frequently shows POP, while modern Irish does not (O’Siadhail 1989). Is early position of the verb necessary but not sufficient for POP or is it also possible in SOV languages?
Regarding diachronic typology (Hendery 2012), what factors could lead to the rise and fall of POP? Some potential favoring factors may include: clause-early verb; subject and object adjacency; Wackernagel’s Law; presence of inflectional cues to subject and object; null-subjects; lack of periphrastic verb constructions; little or no null instantiation of pronominal objects. Moreover, the influence of language contact cannot be ruled out. For instance, the strong preservation of POP in Yiddish may have been helped by contact with Biblical Hebrew and/or Slavic. As for the loss of POP, reduction of inflectional subject cues seems to be a relevant factor West Germanic (cf. Sapir 1921, Hawkins 1986), but other possible factors remain to be discovered.
Dryer, Matthew S. 2005. Order of subject, object, and verb. In Martin Haspelmath et al. (eds.), The world atlas of language structures, 330–333. Oxford: OUP.
Dryer, Matthew S. 2007. Word order. In Timothy Shopen (ed.), Language typology and syntactic description, 61–131. Cambridge: CUP.
Hawkins, John A. 1986. A comparative typology of English and German: Unifying the contrasts. Austin: University of Texas Press.
Hendery, Rachel. 2012. Relative clauses in time and space: A case study in the methods of diachronic typology. Amsterdam: Benjamins.
O’Siadhail, Micheal. 1989. Modern Irish: Grammatical structure and dialectal variation. Cambridge: CUP.
Sapir, Edward. 1921. Language: An introduction to the study of speech. New York: Harcourt, Brace & World.
Shannon, Thomas F. 2000. On the order of (pro)nominal arguments in modern Dutch and German. In Thomas F. Shannon & Johan P. Snapper (eds.), The Berkeley Conference on Dutch Linguistics 1997, 145–196. Lanham, MD: University Press of America.
Shannon, Thomas F. 2005. Drift in Dutch: Fleshing out the factors of syntactic change. In Arie Verhagen & Jeroen M. van de Weijer (eds.), Usage-based approaches to Dutch, 123–167. Utrecht: LOT.
Using Word Order Universals to Contextualize Noun Modifier Word Order in Old English Verse
Verse texts have a vexed status in the study of historical syntax: their poetic elements render them questionable sources of linguistic data. For instance, when Pintzuk and Kroch (1985) make observations regarding verb fronting in early Old English (OE) verse, Kemenade (1987: 64) responds, “It is not entirely clear to me what the status of this observation should be,” discussing the poetic features of the texts that complicate their syntactic study. Yet verse texts are some of the oldest extant texts representing OE. There has been considerable discussion of the relationship between OE verse and prose syntax, e.g., Anderson (2013), Blockley (2001), Pintzuk (2001), Minkova and Stockwell (1992), Mitchell (1985). However, use of OE verse data in syntactic research is still limited. Building upon Anderson (2013), which considers genitive and possessive adjective position in OE verse, this study investigates adjective position in the same OE verse texts. Thus this study provides a comparison of several noun modifiers in OE verse, considering the relationship between different prenominal and postnominal modifiers in OE as related to Gothic and other Germanic languages. Using word order universals and similarities between noun modifier positions in OE verse and Gothic, I argue for OE verse as a source of certain types of syntactic data from the earliest OE period.
This study employs TIGERSearch corpus search software to investigate the frequency of prenominal and postnominal adjectives, with and without other modifiers, in texts from the York-Helsinki Parsed Corpus of Old English Poetry. Frequencies of prenominal and postnominal adjectives are then compared against frequencies of prenominal and postnominal possessive adjectives, common noun genitives, and proper noun genitives for the same texts from Anderson (2013).
Frequencies of postnominal adjectives range from 7.3% (13/177) in Exodus to 24% (152/533) in Beowulf. These frequencies are consistently lower than those for postnominal proper noun genitives, which range from 9.1% to 49.3%, and postnominal possessive adjectives, which range from 26% to 74.4% (Anderson 2013: 852). The frequency of postnominal adjectives in the OE verse texts is generally higher than the frequency of postnominal common noun genitives, which ranges from 3.7% to 37.9% (852). With possessive adjectives showing the highest frequencies of postnominal position in the OE verse texts and adjectives and common noun genitives showing a stronger preference for prenominal position, this picture resembles Smith’s (1971) description of Gothic, with postnominal possessive adjectives at 73% and 86%, postnominal genitives at 43% and 23%, and postnominal adjectives at 27% and 17%. Hawkins (1983: 223) argues that Gothic, the oldest Germanic language with substantial extant records, is “consistent with all nonstatistical universal implications,” but “is one of the rare languages that are inconsistent” with the statistical implication that postnominal adjectives appear with postnominal possessive adjectives. Some OE verse texts demonstrate the same pattern, providing possible related evidence against this statistical implication. These connections suggest that OE verse demonstrates older patterns of noun phrase word order than typically found in OE prose, which shows less variation.
A Sociohistorical Study of Lexical Variation and Frequency of Siddur Tefillot: A Woman's Ladino Prayer Book
Liturgical language is a platform to examine the relationship between Historical Linguistics and Typology. Due to the consistent nature of prayers and texts from holy books (Spolsky 2003), determining lexical frequency in both fields may provide insight about the nature of language change. It has been cited in both diachronic (Diessel 2007) and synchronic (Frisch 2011) studies that frequent words undergo changes more quickly than ones that are not frequent (Bybee 2000). Genitive phonological models have provided a basis for order and patterns of language change diachronically; however experience of an individual and sociological factors affect language variation in diachronic studies (Romaine 1982). A new trend in linguistics is to quantify the rate at which language change occurs (Gries 2015). An exemplar model may point toward a synchronic connection between perception and production. (Pierrehumbert 2001) While a recent approach to examine language change and frequency has been through the application of probability (Bod, Hay, Jannedy 2003) (Pierrehumbert 2003), this theoretical model alone does not best serve variability in language contact as it incorporates the use of multiple regressions as a means to parse out internal linguistic effects of language (Bod, Jannedy, Kennedy 2003). Internal linguistic effects attributed to sociological variation may be present in historic texts under the guise of scribal errors with inconsistencies and corrections found only in the primary sources themselves or enhanced with technology viewed as a digital facsimile.
In the present study, we seek to forge a relationship between the aforementioned theoretical models by examining the language contact between Hebrew and Spanish, in an anonymous pre-sixteenth century text entitled Siddur Tefillot: A Woman's Ladino Prayer Book translated as Book of Prayers. While Moshe Lazar published a groundbreaking work of the transcription of this manuscript (Lazar 1995), no other work published seeks to challenge his transcription of the original facsimile. What little criticism does exist on this scholarship (Schwarzwald 2010) (Minervini 1998), only mentions Lazar’s transcription rather than the digital facsimile or the original copy found in the French National Library. Upon digitizing the text from a digital facsimile of the original language, we have noted inconsistencies in the representation of orthography which we are able to compare with other texts such as the Ladino Bible of Ferrara c.1553 (Lazar 1992) and concordances that accompany it (Lazar 1994).
Under the lens of frequency, in this study we specifically examine the rate at which lexical variation (Bybee 2000) occurs both in the text as a whole as well as among scribes by examining specific phonological environments (Hopper and Traugott 2003) in frequent and infrequent words. (Bybee 2002) By transliterating from the Hebrew character set into the Latin alphabet and digitizing and formatting the text, we create an electronic corpus through the use of metadata codes for automatic extraction of information. With this database, phonological, syntactic and semantic inconsistencies within the text may be quantified and provide for further evidence of how word frequency is key player in the linguistic phenomenon of grammaticalization (Hopper and Traugott 2003) (Bentz 2014).
Bentz, Christian and Buttery, Paula. Towards a computational model of grammaticalization and lexical diversity. Proc. of 5th Workshop on Cognitive Aspects of Computational Language Learning (2014)38–42.
Bod, Rens, Jennifer Hay, and Stephanie Jannedy, eds. Probabilistic Linguistics. Cambridge: MIT Press, 2003.
Bybee, Joan. Frequency of Use and the Organization of Language. Oxford: Oxford University Press, 2006.
Bybee, Joan. “Word Frequency and Context of Use in the Lexical Diffusion of Phonetically Conditioned Sound Change.” Language Variation and Change 14, no. 3 (2002): 261-290.
Bybee, Joan L. “The Phonology of the Lexicon: Evidence from Lexical Diffusion.” In Usage Based Models of Language, edited by Michael Broe and Janet Pierrehumbert. Cambridge: Cambridge University Press, 2000.
Diessel, Holger. Frequency effects in language acquisition, language use, and diachronic change New Ideas in Psychology 25 (2007) 108–127.
Gries, Stefan Th. Some Current Quantitative Problems in Corpus Linguistics and a Sketch of Some Solutions Language and Linguistics 16(1) (2015): 93–117.
Frisch, Stefan A. Frequency Effects in The Blackwell Companion to Phonology: Volume 2. Edited by van Oostendorp, Marc, Colin J. Ewen, Elizabeth Hume and Keren Rice (eds). The Blackwell Companion to Phonology. Blackwell Publishing, 2011.
Hopper, Paul, and Elizabeth Closs Traugott. “Phonogenesis.” In Perspectives on Grammaticalization. Edited by William Pagliuca, 29-45. Amsterdam: Benjamins, 2003.
Lazar, Moshe The Ladino Bible of Ferrara, 1553: a critical edition Labyrinthos1992
Lazar, Moshe, Concordance to the Ladino Bible of Ferrara (1553), Madison: Hispanic Seminary of Medieval Studies, 1994. Microfiches.
Lazar, Moshe. Siddur Tefillot: A Woman's Ladino Prayer Book (The Sephardic Classic Library ; Vol. 10. Paris B. N., Esp. 668; 15th C. Labyrinthos; Limited ed edition 1995.
Minervini, Laura, “Review of Moshe Lazar's edition of Siddur Tefillot: A Woman's Ladino Prayer Book,” Romance Philology 31 (1998) 404-419.
Pierrehumbert, Janet. “Exemplar dynamics: Word Frequency, Lenition and Contrast.” In Frequency and the Emergence of Linguistic Structure. Edited by Joan Bybee and Paul Hopper, 137-158. Amsterdam: Benjamins, 2001.
Pierrehumbert. “Probabilistic Phonology: Discrimination and Robustness.” In Probabilistic Linguistics, edited by Rens Bod, Jennifer Hay, and Stephanie Jannedy, 177-228. Cambridge: MIT Press, 2003.
Romaine, Suzanne. Socio-Historical Linguistics: Its Status and Methodology, Cambridge: Cambridge University Press,1982.
Spolsky, Bernard. “Religion as a Site of Language Contact.” Annual Review of Applied Linguistics 23 (2003): 81-94.
Schwarzwald, Ora Rodrigue. Two Sixteenth- Century Ladino Prayer Books for Women European Judaism Volume 43 No. 2 Autumn 2010.
Siddur Tefillot: A Woman's Ladino Prayer Book (The Sephardic Classic Library ; Vol. 10. Paris B. N., Esp. 668; http://gallica.bnf.fr/ark:/12148/bpt6k58873b
Armenian Prosody in Typology and DiachronyIn light of the current debate surrounding the usage of typology in historical linguistics, I would like to present a problem from the reconstruction of Armenian stress patterns. While it has long been known that Classical Armenian had persistent _nal primary stress, no account of secondary stress is found in the literature. My paper aims to _ll this gap by applying geohistorical, phylogenetic, and typological methods to the analysis of the two stress patterns found in modern Armenian dialects. The _rst pattern, found in select Eastern Armenian dialects, is a simple penult stress system with no associated secondary stress. The other, found in all Western Armenian dialects (those spoken by diaspora communities) and some Eastern Armenian dialects, is the typologically rare “hammock pattern,” in which primary stress is _xed on the _nal syllable and secondary stress on the initial syllable of the prosodic word. According to Gordon (2011:158–9), the only other language varieties documented with this stress pattern are Udihe (Kormushin 1998; Nikolaeva&Tolskaya 2001; Gordon 2002) and some dialects of Canadian French (Gendron 1966; Gordon 2002).
The penult dialects all cluster within a central area in the Eastern Armenian speaking region. Distributions of this type often happen as an innovation spreads outward from its focal point. Just such a situation likely occurred within the history of Armenian; a group of speakers gave up the hammock pattern in favor of the less typologically marked penult pattern, while the dialects at the periphery have maintained the more conservative accent pattern. This relationship is also indicated by the phylogenetic tree generated by PAUP* 4.0b10 (Swo_ord 1998), which was used to run a phylogenetic analysis of 21 modern and historical dialects of Armenian; all six of the penult dialects in our sample are nested together in the tree under a single node, suggesting common descent (Skelton & DeLisi forthcoming).
There is also evidence from manuscript and inscriptional variants that the hammock pattern predates the modern era. A reduction of medial vowels, similar to that seen in ModernWestern Armenian, can be found throughout the manuscript tradition, including sporadic attestation in both of the earliest manuscripts of the Classical Armenian Gospels (Künzle 1984:I 61, 65), as well as in the inscriptional record (Stone 1982:127). Synchronic phonology and diachronic dialectology support the conclusion to reconstruct the typologically rare hammock system against the typologically common penult system for the period of shared innovation, Proto-Armenian. I will further contend that only one possible diachronic pathway exists for the creation of the hammock system, and that this pathway bias drives its typological rarity. Hammock languages, with their inherent mismatch of foot structure (iambic primary stress and trochaic secondary stress) can only arise from an earlier iterative trochaic language which then loses its post-tonic _nal syllable, as was the case in both Proto-Armenian and Old French. Therefore, the typological rarity of hammock languages is due to pathway bias, and not an inherent property of the prosodic system.
Gendron, Jean-Denis. 1966. Tendances phonétiques du français parlé au Canada. Paris: C.
Gordon, Matthew. 2002. A Factorial Typology of Quantity Insensitive Stress. Natural
Language and Linguistic Theory 20. 491–552.
Gordon, Matthew. 2011. Stress Systems. In John Goldsmith, Jason Riggle & Alan C. L. Yu (eds.), The handbook of phonological theory, 141–163. Cambridge: Blackwell 2nd edn.
Kormushin, Igor Valentinovich. 1998. Udyxejskij (Udegejskij) jazyk [the Udihe language]. Moskow: Nauka.
Künzle, Beda O. 1984. Das altarmenische Evangelium / L’Evangile en arménien ancien, vol. I: Edition, II: Lexikon / Lexique. Bern: Peter Lang.
Nikolaeva, Irina & Maria Tolskaya. 2001. A grammar of Udihe. New York: Mouton.
Skelton, Christina & Jessica DeLisi. forthcoming. A Phylogenetic Analysis of Armenian
Stone, Michael E. (ed.). 1982. The Armenian inscriptions from Sinai. Cambridge: Harvard University Press.
Swo_ord, David L. 1998. PAUP*, Phylogenetic Analysis Using Parsimony (*and other
methods). Ersion 4, Sinauer Associates, Sunderland, MA.
Balancing Typology and Historical Linguistics: the Consonant Shifts
Typological evidence is clearly necessary when making plausible reconstructions in historical phonology, particularly when philological evidence from earlier stages of the language is lacking. This paper, however, examines several cases where the use of common typological patterns without access to philological data and comparative evidence might result in a fallacious reconstruction. The illustrative cases are all in some way associated -- broadly speaking -- with a consonant shift – which is itself a cross-linguistically rather rare type of change. We conclude that where typological patterns and historical data lead to differing conclusions, the historical evidence should trump typological considerations except where doing so would require positing an utterly unique development without known analogs.
The first example involves the operation of Grimm’s law in Germanic and assumes a starting point of the traditionally reconstructed PIE system including plain voiceless and plain voiced stops (Lehmann 1992: 152-3). Philological evidence is unavailable until at least two centuries after the shift and our reconstructions are based on comparative evidence and patterns of change seen in other languages. These factors suggest that a direct shift of voiceless stops to voiceless fricatives (PIE *p *t *k > f θ x) took place. By contrast, the later High German consonant shift (500-700 AD) produced a shift of Germanic p, t, k to the corresponding affricates pf, ts, kx which subsequently weakened to fricatives in some environments but not in others (Davis 2008:213-216). The attested facts of the HG shift raise the possibility that the Germanic shift might have originally yielded affricates that later weakened to fricatives prior to the first attestation of Germanic, even if the shift of stops directly to fricatives is cross-linguistically more common. Other cases where typological patterns could lead us astray include Hu, a language of mainland SE Asia, that like many others in this Sprachbund has undergone a change that is often dubbed a ‘Germanic-style’ shift. The common typological pattern would be for the initial voiced stops /b d g/ to lose the feature [+voice], but only after having depressed the pitch of the following vowel, while the unaspirated initial voiceless stops become aspirated, thus raising the pitch of the following vowel (Haudricourt, A.-G. 1965) and triggering tonogenesis. But the developments in Hu (Svantesson 1991:67) clearly differ from the prevailing areal pattern and have sporadic analogs elsewhere.
Finally, we return to the output of PIE *p *t *k in Germanic that produced [β ð ɣ] in part under the effects of vernerization. Philological and comparative evidence clearly suggest these voiced fricatives later hardened to stops to one extent or the other in all of the Germanic languages (Lass 1994:56) while Dutch continues to preserve the velar /ɣ/. The assumption that Dutch /ɣ/ reflects the cross-linguistically more common pattern of development –- whereby /g/ is replaced by /ɣ/ – is in this case trumped by substantial counter-evidence.
Davis, Garry W. 2008. Toward a progression theory of the Old High German consonant shift. Journal of Germanic Linguistics 20.3. 197-241.
Haudricourt, A.-G. 1965. Mutation consonantique en Mon-Khmer. Bulletin de la Société Linguistique de Paris 60. 160-172.
Lass, Roger. 1994. Old English: A Historical Linguistic Companion. Cambridge: Cambridge University Press.
Lehmann, Winfred. 1992. Historical Linguistics. 3rd ed. New York: Routledge.
Svantesson, Jan-Olaf. 1991. Hu – a language with unorthodox tonogenesis. Austroasiatic languages, Essays in honor of H.L. Shorto, ed. by J.H.C.S. Davidson, 67-80. London: School of Oriental and African Studies, University of London.
A discussion of the panchronic effects of rhythm on tonal systems
This paper examines the nature of rhythm as a constraint on tonogenesis while also considering the influence of prosodic structure on the properties and functions of tone systems. Specifically, it is argued that variation in the prosodic structure of languages undergoing tonogenesis plays a role in determining the synchronic typological variation we observe in the world’s tonal languages. Rhythm type can account for tonal phenomena such as the cross-linguistic differences in Tone-Bearing Units (TBUs), size and density of tonal systems as well as differences in the phonetic realization of tone. Furthermore, it is argued that when a language undergoes a shift in rhythm type, the functional and phonetic behavior of the tonal system also changes, driven by the new rhythmic structure. These issues are considered through a comparative analysis of Vietnamese, Sahaptin (Penutian) and Tibetic languages (e.g., tonal Lhasa Tibetan and non-tonal Amdo Tibetan).
Rhythm, defined here as the ordering of phonological elements into patterns of hierarchically organized prosodic units, plays an important role not only in the development of phonemic tone, but also the functional and phonological properties of the resulting tonal system. This paper considers the interaction between rhythm and tone from both a diachronic perspective, looking at how processes of tonogenesis are shaped by rhythm, as well as a synchronic perspective, as variation in the properties of different rhythm types is shown to correspond to variation in the functional and structural properties of tonal systems. It is argued that phonemic tone first develops within restricted prosodic environments. Tonogetic pathways have many proposed origins, including segmental features, as is the case for Asian tone languages such as Vietnamese, and prosodic features, as is the case for pitch-accent languages such as Sahaptin (Hargus and Beavert 2005). Regardless of the specific origin, tonogenesis is driven by phonological changes that occur in rhythmically prominent environments. This is to say that, while phonemic tone emerges through the transphonologization of predictable pitch variations associated with some other phonological phenomenon, be it stress or laryngeal features of consonants, these processes are initially confined to prominent, that is stressed, syllables. Consequently, the degree to which the phonology of a tonogenetic language is populated by prominent elements both determines the relative density of the tone system (i.e., the percentage of potential tone-bearing units, as defined phonologically, that require tone as well as impacts the domain of tonal contrast—whether contrast is templatic (contrasting the phonologically compositional melodies of groups of tone-bearing units) or marked by discrete “tonemes” linked to individual tone-bearing units.
Gussenhoven, C. (2004). The phonology of tone and intonation. Cambridge University Press.
Hargus, S. and V. Beavert (2005). A note on the phonetic correlates of stress in Yakima Sahaptin. University of Washington Working Papers in Linguistics 24: 64-95
Hyman, L. M. (2009). How (not) to do phonological typology: the case of pitch-accent. Language Sciences, 31(2), 213-238.
Different evidence, higher credence: towards a clearer methodological distinction between historical linguistics and typology
While both historical linguistics and typology claim to address diachronic change in languages, there is an asymmetry in the relationship between historical linguistics and typology as it manifests itself in practice. Perhaps because of the relative lack of theorization in historical linguistics, it is only a slight overstatement to say that the discipline provides data to typologists, but borrows methodological guidelines from them in the form of both tools for, and constraints on reconstruction. The circularity of typological arguments based on historical reconstructions along the lines of typological principles is one of the detriments of this relationship. Despite this discrepancy, there are no methodological debates aiming to solve this problem, or even to determine best practices. This paper draws on recent debates in the philosophy of science to argue for a less hierarchical, and for that reason epistemically more productive relationship between historical linguistics and typology. We suggest to formalize the products of both subfields as methodologically independent evidence, and argue that the overall credence in certain given findings can be increased by such mutual bolstering of evidence.
Current philosophical debates about the status of different kinds of evidence in evolutionary biology make this subfield more directly pertinent to linguists than most broader appeals to shared problems between both disciplines. In particular, there is a nascent discussion of the relation of evidence stemming from homologous versus homoplastic traits in two given evolutionary lineages, i.e. a trait shared by a common ancestor versus a trait that has evolved independently in two lineages without being present in the common ancestor. With this clear distinction in place, a different kind of evidential role is ascribed to traits resulting from the two different developmental processes, neither being subordinate to the other but mutually bolstering.
We wish to make a distinction along similar lines between the evidential role of the reconstructions made by historical linguistists, and that of the cross-linguistic data culled by typologists. In the former case genetic relatedness is the main epistemic warrant, in the latter it is genetic unrelatedness. While seemingly obvious, this distinction is currently frequently overridden in the transition of examples from historical linguistics to both diachronic-typological generalization, and as alleged instances of independent occurrences in synchronic typology.
Shifting the emphasis to the different evidential role of the different objects of investigation of typologists and historical linguists, we aim to emphasize and reinforce their methodological independence. The reasons for this are twofold. Firstly, the methodological anxieties within historical linguistics might be alleviated to an extent by drawing upon the extensive discussion of the evidential role of homologous development in biology which has received most of the attention so far, homoplastic development being the latecomer. Here, the roles of the dominating methodology are reversed with respect to linguistics, and spelling out why in terms of the respective types and scope of inferences, and credence-determining factors, goes some way towards filling the holes in terms of theorization, which, we argue, is still essential for historical linguistics to move forwards. Secondly, a certain degree of methodological independence between typology and historical linguistics is the prerequisite for avoiding circularity in their collaboration. Clarity on their respective types of evidence is crucial to identify overlaps in explanatory scope, and we seek to show such cases in which evidence of different kind will increase the credence of a claim.
Adrian Currie, ‘Convergence as Evidence.’ Brit. J. Phil. Sci. 64 (2013), 763-786;
Paul E. Griffiths, ‘The Phenomena of Homology’, Biology and Philosophy 22 (2007), 643-658;
Brian K. Hall, ‘Descent with modification: the unity underlying homology and homoplasy as seen through an analysis of development and evolution’, Biological Review 78 (2003), 409-433.
Modeling Spread Zones and Residual Zones in the Ancient Greek Dialects
In her seminal work Linguistic Diversity in Space and Time, Johanna Nichols develops a model of the relationship between linguistic diversity and physical geography (1991, 5). This model identifies two types of linguistic areas, spread zones and residual zones. Spread zones-- broad, unbroken expanses such as the Eurasian steppe-- are characterized by low genetic diversity and low stability, since one language family tends to succeed another. Residual zones-- peripheral or inaccessible areas such as the Caucasus Mountains-- show high levels of genetic diversity and high levels of stability, and tend to accumulate a greater number of language families over time (13-24). This model is used in the service of typology to study language at great time depths. Historical linguistics would also benefit from a geographic model of linguistic diversity and language development. For example, in studies of the Ancient Greek dialects, one finds statements that one dialect, Arcadian, retains many archaisms because of its geographic isolation (e.g. Woodard 2004, 650). No evidence supports this causal relationship, and its implications are never explored. Do dialects in well-connected areas experience more borrowing and innovation? More generally, what is the relationship between geographic connectedness, language contact, and language change?
Nichols' model makes it possible to turn such potentially unfounded statements into a testable hypothesis. Two of Nichols' criteria for spread zones and residual zones are applicable to any time depth or degree of relatedness. These criteria refer to the spread of innovation: spread zones have an innovating center and conservative periphery, while spread zones have no clear center of innovation (Nichols 1991, 16-17, 21). The amount and directionality of borrowed features can be evaluated using conditional probabilities. Given two dialects in a spread zone, where Dialect A is closer to the center of the spread zone than Dialect B, Dialect B should share a very high proportion of its non-inherited linguistic features with Dialect A, since Dialect A would be the major source of the innovations Dialect B adopts. Conversely, Dialect A would share a lower proportion of its non-inherited linguistic features with Dialect B, since Dialect A continues to adopt innovations from dialects closer to the center of the spread zone. Thus, the conditional probability of Dialect B possessing a non-inherited feature given that Dialect A also possesses it should be higher than the reverse. Applied to the Greek dialects, Nichols' model does hold true. The Aegean Sea, the most open and easily traversable part of Greece, contains a spread zone. Northern Greece and the Peloponnese (including Arcadia), which are very mountainous, appear to be residual zones. These results suggest that the model of linguistic diversity Nichols developed for typology can be successfully applied to historical linguistics to test hypotheses and uncover generalizations no-one would have previously thought to look for. These results also support Nichols' conclusions, since they imply that the patterns seen on very long timescales are simply the cumulative result of smaller-scale processes.
Nichols, Johanna. 1992. Linguistic Diversity in Space and Time. Chicago and London: The University of Chicago Press.
Woodard, R. D. 2004. Greek Dialects. in Woodard, R. ed., The Cambridge Encyclopedia of the World’s Ancient Languages. Cambridge: Cambridge University Press, 650–72.
Grammaticalization in Romance: the relationship between language change and cross-linguistic distribution of patterns
Determiners and auxiliaries where inexistent in Latin and developed in all Romance languages. This paper intends to show that diachrony and language comparison are complementary perspectives, since the diachronic analysis of the different evolutionary stages of these two grammatical categories within a single language helps to understand divergent degrees grammaticalization across languages of the same family and vice versa.
Our theoretical hypothesis is twofold. First, grammaticalization is a central mechanism in language change because it leads to the creation of abstract patterns (Noël 2007) which may become productive over time (Trousdale 2010). The fundamental principle at work is analogy (Fischer 2008, Itkonen 2005), in particular pattern finding, the ability to create syntactic types out of frequent tokens, and system mapping, the ability to apply the same pattern to another set (Tomasello 2003). The formal similarity, even at an abstract level, is ultimately motivated by a functional similarity.
Secondly, when frequent structures grammaticalize (Bybee 2003), their formal mold gets automatized and may become so rigid that competing forms get pushed out of the paradigm and disappear. However, with respect to Romance, as languages do not grammaticalize at the same pace (Lamiroy & De Mulder 2011), paradigmatization (Lehmann 2002) as well as progressive strengthening of syntagmatic constraints is far more radical in French than in the other Romance languages.
Methodologically, our analysis is based on translations of the same Latin source in different stages of French on the one hand and in contemporary Italian and Spanish on the other. By using translation corpora, we show the fruitfulness of this double perspective: comparing evolutionary stages of a single language and synchronic states of genetically related languages may lead to similar results.
Our main findings are:
- Historically, the NP and VP pattern emerged from Latin to Romance in a parallel way developing two analytical structures containing prenominal and preverbal markers viz. Det + N and Aux + V. This formal similarity is motivated by a similar function: both determiners and TAM-auxiliaries contribute to the anchorage in the situational and textual context.
- The evolution occurred earlier for NP than for VP. In all three languages, the definite article created a prenominal slot which later hosted the indefinite article and eventually other indefinites, possessives and demonstratives, thus creating the paradigm of determiners. Similarly and subsequently, a preverbal slot was created in the VP for hosting tense, aspect and modality markers, thus creating the paradigm of auxiliaries.
- Since the evolution occurred earlier for NP than for VP, it is more radical. Thus prenominal zero marking virtually disappeared in French, and determiners became obligatory, whereas verbal TAM marking is still expressed by verbal suffixes as well as by auxiliaries
- In comparative terms, French represents the most grammaticalized Romance language. Spanish is in many regards similar to Old French and Italian occupies the intermediate position.
- Due to the workings of analogy, implicational relations hold both historically and typologically within Romance: languages or language stages with a strict Det + N pattern will also have a more constrained Aux + V pattern.
Bybee, J. 2003. Mechanisms of change in grammaticization: the role of frequency. In Joseph B. & R. Janda (eds.). The Handbook of historical linguistics, 602-623.
Fischer, O. 2008. On analogy as the motivation for grammaticalization. Studies in Language, 32:2, 336-382.
Itkonen, E. 2005. Analogy as structure and process. Amsterdam: J. Benjamins.
Lamiroy, B. & De Mulder, W. 2011. Degrees of grammaticalization across languages. In Narrog, H. & B. Heine (eds.) The Oxford Handbook of Grammaticalization. Oxford: OUP, 302-318.
Lehmann, C. 2002. Thoughts on Grammaticalization. 2nd rev. edition. Erfurt: Seminar fur Sprachwissenschaft der Universität.
Noël, D. 2007. Diachronic construction grammar and grammaticalization theory. Functions of Language 14:2, 177–202.
Tomasello, M. 2003. Construction a language. A usage-based theory of language acquisition. Cambridge, MA: Harvard University Press.
Trousdale, G. 2010. Issues in constructional approaches to grammaticalization in English. K. Stathi, E. Gehweiler & E. König (eds.) Grammaticalization. Current views and issues. Amsterdam: John Benjamins, 51-72.
Diversification in Argument Structure Constructions of Indo-Iranian Cognate Verbs
Diversification in Argument Structure Constructions of Indo-Iranian Cognate Verbs This talk compares the distribution of cognate “do, make” verbs in argument structure constructions (henceforth ASC) in twelve Indo-Iranian languages. We argue that in order to analyze cross-linguistic variation in these ASCs, one needs to use methods from historical syntax and typological linguistics. The combination of these methods is required for any diachronic study of diversification patterns in the distribution of “do, make” and its correlation with any classification, especially genealogical and chronological, of Indo-Iranian languages. “Do, make” is common in “complex predicate” constructions (Mohanan 1997, Korn 2013) which express many semantic domains. Some such domains, like prototypical transitive events (Comrie 1989) or motion events, are associated with distinct ASCs. Sometimes, following re-analysis and extension (Harris&Campbell 1995:61-120, Gildea 1998:35-41), “do, make” distributes into ASCs associated with such fields. Arguing that “do, make” verbs were extended in a comparable manner into comparable constructions in different languages, requires a definition of these constructions in a cross-linguistically valid way. As clause alignment patterns vary in Indo-Iranian (Haig 2008, Deo&Sharma 2006), this definition cannot include structural factors, and should be based on typological comparative concepts (Croft 2001, Haspelmath 2011).
Data for this talk comes from documentation projects (e.g., Shukri et al. 2013) and philological editions (e.g., MacKenzie 1976). In the data, “do, make” commonly distributes in prototypical transitive ASCs. In (1a-b) hayā “egg” and panir “cheese” behave like the Gorani and Mazandarani P arguments respectively. Often, however, “do, make” ASCs differ when coding similar propositions. In (2a) from Sivandi, the helped entity is coded by a PP and komak “help” is coded as the P argument. In (2b) pišīyaka “a cat” is coded like a P argument in Gorani.
(1a) hāya bi-kar-im
“So I may lay eggs” (Gorani)
(1b) panir kārd-e
“He made cheese” (Mazandarani)
(2a) … ke ba hame komak kare.
… REL to 1PL help do.NMZ
“(Anyone) who helps / will help” (Sivandi)
(2b) … ka kumak pišī-yaka=šān kard-ē
… REL help cat-INDEF=3PL do.PST-3SG
“(The people) who helped the cat” (Gorani)
Motion events are also expressed using “do, make”. In (3a) one can identify an extension into motion ASC as the goal of motion is coded post-verbally. The direction of motion, xīz “up”, is coded like the P argument. (3b) exhibits similar semantics and word order, but is syntactically intransitive. Thus, the two constructions, while semantically comparable, are syntactically distinct.
(3a) se kākā xīz=šā kerd ū tā
three brother up=3PL do.PST DEM side
“The three brothers jumped to the other side” (Sivandi)
(3b) gala ma-kar-ī=ya ā das
flock IND-do.PRS-3SG=DIRC DEM side
“The flock went to the other bank” (Gorani)
Examples shown here are from three North-West Iranian languages. Once further languages are added, the variation in alignment patterns and available constructions targeted for extension increase, and the necessity for a delicate analysis based on comparable constructions increases with them.
Comrie, Bernard. 1989. Language universals and linguistic typology: Syntax and morphology. 2nd edition. Oxford: Blackwell.
Croft William. 2001. Radical Construction Grammar. Oxford: Oxford University Press.
Deo, Ashvini & Sharma, Devyani. 2006. “Typological variation in the ergative morphology of Indo-Aryan languages.” Linguistic Typology 10(3).
Gildea, Spike. 1998. On reconstructing grammar: comparative Cariban morphosyntax. Oxford: Oxford University Press.
Korn, Agnes. 2013. Looking for the middle way: Voice and transitivity in complex predicates in Iranian. Lingua 135.
Haig, Geoffrey. 2008. Alignment change in Iranian languages: a construction grammar approach. Berlin: De Gruyter.
Harris Alice & Campbell Lyle. 1995. Historical syntax in cross-linguistic perspective. Cambridge: Cambridge University Press.
Haspelmath, Matrin. 2011. “On S, A, P, T and R as comparative concepts for alignment typology.” Linguistic Typology 15(3).
MacKenzie, David N. 1976. The Buddhist Sogdian texts of the British library. Leiden: Brill.
Mohanan Tara. 1997. “Multidimensionality of Representation: NV Complex Predicates in Hindi.” In: Alsina, Alex, Bresnan Joan & Peter Sells (eds.), Complex Predicates. Stanford: CLSI
Mahmudweyssi Parvin, Bailey Denise, Paul Ludweig & Haig Geoffrey. 2012. The Gorani language of Gawarǰū. Weisbaden: Reichert Verlag.
Shokri Guiti, Jahani, Carnia & Barani Hossein. 2013. When Tradition Meets Modernity: Five Life Stories from the Galesh Community in Ziarat, Golestan, Iran. Uppsala: Acta Universitatis Upaliensis.
Sivandi:Lecoq, Pierre. 1979. Le dialecte de Sivand. Weisbaden: Reichert Verlag.