720 Proseminar on Semantic Theory (3 units)

Fall 2008


TuTh 2:30-3:45 - SOM 120


Barbara H. Partee []

Seth Cable []


A bridge between the introductory courses 610 and 620 and more advanced seminars. Content variable from year to year.

This seminar will explore recent, theoretically-informed research into the semantics and syntax of the languages of the American Pacific Northwest.

The Northern Pacific Coast of North America is home to some of the most typologically distinct languages on the continent. Amongst their notable features, these highly endangered and understudied languages have played a pivotal role in the development of linguistic science as the earliest and most sustained impetus to the work of Franz Boas and his students.

More importantly, since even this earliest work, scholars have recognized that these languages are of great value to the perennial debate concerning the scope and limits of linguistic variation, as they appear to display many rather marked differences from the grammars of 'standard European' languages. Indeed, when Ken Hale was once asked in a popular interview which languages he considered the 'most unlike' English, he named a language of the Pacific Northwest (Straits Salish).

Interestingly, these alleged differences are almost always 'absences'. For decades, grammatical study of these languages concluded that they lack a variety of properties common to more familiar languages: nominal arguments, lexical categories, adnominal quantifiers, tense, etc. In recent years, however, theoretically-informed research into these languages has argued that many of the earlier claims regarding them were mistaken, and that those features initially thought to be absent from these languages are simply less surface-apparent than in more familiar languages.

Either answer, of course, holds long-ranging consequences for our understanding of the nature of linguistic variation and thus of UG. Furthermore, the argumentation on either side of this debate has greatly advanced both our knowledge of these particular languages, as well as the empirical/experimental techniques that can be used with untrained native speakers to diagnose various semantic and syntactic properties of their languages. Thus, theoretically- informed research into these languages can often be exported directly to novel field situations, to further the analysis and documentation of other endangered and understudied languages.

This seminar will explore a variety of areas - both semantic and syntactic - where the languages of the Pacific Northwest have advanced or otherwise impacted our theory of UG. The topics discussed will be among the following: 1) (Non)-Configurationality and the Pronominal Argument Hypothesis 2) Lexical Categories (or Lack Thereof) 3) Syntax and Semantics of Transitivity 4) Presuppositions 5) Quantification 6) Principle C (and Lack Thereof) 7) Tense 8) Modals and Evidentials 9) Wh-Questions 10) Polysynthesis (in the Wakashan language family) 11) Information Structure and Intonation 12) Topic-Tracking, Argument Hierarchies and the Passive/Inverse [13) The Left-Periphery] [14) Ergativity] [15) Possessor Raising] [16) Control and Backwards Control] [17) Matrix Subordinates (main, independent clauses that have the surface form of subordinate clauses).]

Following this description is a brief summary of the issues and debates surrounding these topics, as covered in the seminar. It is likely that time will permit only subjects (1) - (12) above to be covered in class. However, materials covering subjects (13) - (17) will also be made available to anyone that is interested.

The only course requirement will be a final paper (or project). Given that the seminar focuses on languages whose speakers are rather inaccessible to us here in Massachusetts, final papers(/projects) need only touch upon the issues and readings discussed in class. For example, a paper simply working towards a compositional semantics for wh-questions in Nuu-chah-nulth (in which wh-words are obligatorily incorporated) would certainly be acceptable. However, participants are especially encouraged to develop original elicitation tasks that could be used 'in the field' to test either existing or novel hypotheses. While a well-designed elicitation task could alone be sufficient as a final project, it might in some cases also be possible to obtain original data from linguists currently working on the relevant language(s).

For those who are interested, what follows is a brief description of each of the seventeen topics listed above.

1) (Non)-Configurationality and the Pronominal Argument Hypothesis Virtually all the languages of the Pacific Northwest have at some time been argued to be either 'Non-Configurational' or 'Pronominal Argument Languages'. We critically assess these claims for languages in each of the major language families of the area - Salishan, Wakashan, Tsimshianic, Na-Dene - especially in light of more recent research supporting configurational analyses. In so doing, we also critically assess the diagnostics typically employed to establish (non-) configurationality, as well as the more general viability of 'macroparameters' in typological research.

2) Lexical Categories (or Lack Thereof) The languages of the Salishan, Wakashan and Chimakuan families have long been noted for appearing to lack any morpho-syntactic distinction between 'nouns' and 'verbs'. In all these languages, any open class lexical item can freely inflect as either a predicate or an argument. This apparent lack of any morpho- syntactic diagnostics for 'noun/verb-hood' has classically been taken to show that these languages simply lack the distinction been nouns and verbs altogether. We critically assess these claims in light of more recent research, which suggests that there in fact are some (extremely subtle) structural differences between nouns and verbs in these languages. Along the way we also assess the consequences either result holds for our theory of UG and the nature of lexical categories.

3) Syntax and Semantics of Transitivity Languages of the Salishan family (or 'Salish languages') exhibit a remarkably extensive and productive 'transitivizing' system. By and large, it appears that all simple roots in these languages are intransitive (unaccusative), and any transitive stem is created only through the suffixation of a 'transitivizer' to one of these roots. Intriguingly, this holds even for such thoroughly 'transitive' concepts such as 'hit' and 'kick' -- such concepts are expressed via complex stems formed from basic roots with the meaning 'get hit' and 'get kicked'. We will read and critically assess theoretically-informed analyses of this system. Of particular interest is the claim made by Davis & Demirdache (2000) that while simple roots of the 'get kicked' variety are *syntactically* unaccusative, they are nevertheless *semantically* transitive.

4) Presuppositions Semantic research into Salish languages has uncovered the following startling fact - these languages seem not to linguistically encode presuppositions. For example, pronouns, demonstratives and clefts can all be freely used in 'out-of- the-blue' contexts where their correlates in (e.g.) English are strongly disallowed. We will read and critically assess the literature surrounding this issue, one that plays a crucial role in the subjects that follow.

5) Quantification We examine the nature of quantification across several language families of the area. As the Salish languages have received the most extensive study by far, our discussion will center largely on them. We will examine the claim that (some) languages of this family lack so-called 'D-quantification', as well as the notion that such an absence could be connected to their putative lack of an N/V-distinction. We will also examine research by Lisa Matthewson into the semantics of specific indefinites and distributive numerals in (one of) these languages.

6) Principle C (and Lack Thereof) Along with their apparent lack of presuppositions, Salish (and certain neighboring) languages also appear to permit violations of Principle C. That is, it is (sometimes) possible for a pronoun in these languages to co-refer with a name that it c-commands. We will critically assess the literature surrounding this issue, as well as the consequences it might have for our understanding of the nature of Principle C. Of special importance will be the work of Davis (2008), who seeks to derive these puzzling facts from the more general absence in these languages of linguistically encoded presuppositions.

7) Tense We will examine a variety of issues related to tense in the languages of this area. As regards the Salish languages, we will critically examine the claim that (some) languages of this family express tense distinctions via *nominal* (rather than verbal) morphology. We will also examine the competing claim that, while sentences of these languages appear not to have any verbal tense morphology, they nevertheless possess a phonologically empty and semantically underspecified Tense projection. We will then examine recent research that makes a parallel claim for certain Tsimshianic languages. Finally, we will examine a curious tense/modal in Tlingit (a Na-Dene language) which may provide evidence for a similarly 'unpronounced and underspecified' tense projection in that language.

8) Modals and Evidentials We will first review recent work by Henry Davis, Lisa Matthewson and Hotze Rullman regarding the semantics of modals in St'at'imcets (a Salish language). This work argues that - while modals in more familiar European languages linguistically encode quantificational force and leave the nature of the modal base underspecified - modals in St'at'imcets linguistically encode features of the *modal base* and leave *quantificational force* underspecified. We will then examine the ways in which Davis, Matthewson and Rullman extend this work to an analysis of evidentials in St'at'imcets. Finally, we will examine recent research by Tyler Peterson that applies the work of Davis and colleagues to the treatment of evidentials in Tsimshianic languages.

9) Wh-Questions We will examine a variety of issues related to the syntax and semantics of wh- questions in the languages of this area. Regarding the Salish languages, we will examine recent work by Henry Davis, who argues that contrary to surface appearance, Salish languages employ a wh-in-situ strategy directly parallel to that in Japanese. Regarding the Wakashan languages, we will review work by Henry Davis and Naomi Sawai, who seek to analyze the puzzling surface structure of wh-questions in Nuu-chah-nulth, a language that requires wh-words to simultaneously be incorporated *and* to be left-peripheral in the clause. Finally, we will critically examine work probing the structure of wh-questions in the Na-Dene languages Tlingit and Haida, as well as the claim that wh- fronting in these (and perhaps all) languages is ultimately an operation targeting Q(uestion)-particles, and not the wh-words themselves.

10) Polysynthesis (in the Wakashan language family) The languages of the Wakashan family exhibit polysynthesis (incorporation) to a degree that is otherwise uncharacteristic of the Pacific Northwest, and more akin to languages of the arctic (e.g. Inuit). We will critically examine a variety of competing theories regarding the nature of complex word formation in the Wakashan languages. First, we examine work by Emmon Bach and Stephen Anderson arguing that polysynthesis in the Northern Wakashan languages Haisla and Kwakwala is a lexical (rather than syntactic) phenomenon. Following this, we examine research by John Stonham and others arguing that polysynthesis in the Southern Wakashan language Nuu-chah-nulth is a *syntactic* (rather than *lexical*) phenomenon. Finally, we examine two other competing analyses of Nuu- chah-nulth polysynthesis: the 'PF' analysis of Rachel Wojdak, and the HPSG analysis of Ryan Waldie.

11) Information Structure and Intonation While linguists have only just begun to study the relationship between information structure, intonation and syntax in the languages of this region, some intriguing results have already been obtained. We will first examine research by Karsten Koch, which argues that properties such as 'givenness' or 'focus' have no effect upon the prosody of sentences in Nlhe7kepmcin (a Salish language). Following this, we will examine research exploring the prosodic structure of several Salish languages, and critically assess claims made by David Beck and Allison Benner that they pose difficult challenges to current models of the syntax/prosody interface.

12) Topic-Tracking, Argument Hierarchies and the Passive/Inverse One feature of Pacific Northwest languages that has garnered much attention from linguists is the way in which the 'voice' of a transitive clause is affected by properties of the semantic agent, generally either its 'topicality' or its position along a 'person hierarchy'. As in many languages across the world, languages of this area tend to require that a clause appear in the 'passive/ inverse' voice if the topicality or person of the patient is 'higher' than that of the agent. We will critically review studies of this phenomenon across the languages of this region. Special attention will be paid to the question of whether the effects of the person hierarchy on voice can - as is often claimed - be reduced to the (more productive) effects of topicality. In addition, we will explore the parallel controversies concerning the very nature of 'voice' morphology in the Pacific Northwest: whether such morphology is best analyzed as encoding an active/passive contrast (as in European languages), an inverse/ direct contrast (as in other North American languages), or something else entirely.

13) The Left-Periphery There are a variety of interesting studies exploring the 'fine structure' of the left-periphery in Salish, Wakashan and Na-Dene languages. We will critically examine this work, with an eye towards cross-linguistic comparison, as well as towards comparison with theoretical models such as that of Rizzi (1997).

14) Ergativity Many languages of the Pacific Northwest exhibit ergativity to some degree. We will examine the nature of ergativity in three language families of this area: Tsimshianic, Salishan and Na-Dene. First, we will examine the controversy regarding the 'depth' of ergativity in the Tshimshanic family. Many early studies argue that the Tsimshianic languages qualify as one of the few instances of 'deeply' (or 'syntactically') ergative languages. We will review this original research, as well as more recent research that reevaluates these languages as possessing only the more common 'surface' (or 'morphological') ergativity. Next, we will examine work by Martina Wiltschko that seeks to analyze the puzzling person-based split ergativity seen in Halkomelem, a Salish language. Finally, we will explore the related system of 'active agreement' in the Na-Dene languages Tlingit and Haida, a system where the appearance of subject agreement on intransitive verbs depends upon whether the subject is agentive/ergative or patientive/absolutive.

15) Possessor Raising The phenomenon of 'possessor raising' is found across several linguistic families in the Pacific Northwest. We first briefly examine possessor raising in the Salishan family, which exhibits a relatively 'unproblematic' behavior akin to that found in the Bantu languages. Next, however, we examine possessor raising in the Wakashan and Na-Dene languages, which exhibit properties that are more typologically unusual and analytically challenging. The work of Christine Ravinski, for example, seeks to account for the puzzling requirement that possessor raising in Nuu-chah-nulth can target *only subjects*, a property that most theories of possessor raising would predict to be impossible. Furthermore, we will examine the surprisingly unconstrained process of possessor raising in Haida, where possessors are permitted to raise from any position within the clause, even from within PPs.

16) Control and Backwards Control The phenomenon of 'backwards control' has recently received extensive study by Maria Polinsky and Eric Potsdam, who argue that the phenomenon (as it occurs in languages such as Tsez, Kabardian and Malagasy) is best analyzed within a 'movement theory' of control. Interestingly, backwards control also occurs in the Na-Dene language Haida, and has been extensively documented by John Enrico in his reference grammar for the language. We will examine Enrico's detailed description of the phenomenon, paying special note of aspects that appear problematic for a 'movement-based' analysis such as Polinsky and Potsdam's. (It is worth noting here that Enrico himself develops an analysis that is somewhat in line with the competing theory of control developed by Idan Landau.) In addition, we will examine work by Donna Gerdts and James Thompson on control in the Salish language Halkomelem, where it appears that control requires an infinitive to be 'restructured' (in the sense of Wurmbrand 2001).

17) Matrix Subordinates In both the Salishan and Na-Dene families, there are languages in which it is possible to use subordinate clauses as 'free standing, independent utterances'. That is, in these languages, one often encounters independent clauses - ones bearing no discernible syntactic relationship to any preceding or following clauses - bearing morpho-syntactic properties that are otherwise characteristic of subordinate clauses. In some cases, such 'matrix subordinates' have a clear semantic contrast with normal matrix clauses; in Tlingit, for example, they often appear translated in texts as exclamatives. In other cases, however, it is difficult to determine what - if anything - the semantic contribution of such subordinative morphology is. We will examine this phenomenon in both the Na- Dene language Tlingit and the Salish language Halkomelem. We will also examine some potentially related phenomena in more widely-studied Pacific languages such as Japanese and Korean, as well as in European languages such as Spanish.