CONSTRAINTS ON AGREEMENT PATTERNS IN ROMANSH AND TUATSCHIN Géraldine Walther Claudia Cathomas Department of Comparative Linguistics URPP Language and Space & University of Zurich Department of Comparative Linguistics University of Zurich

The Romansh languages are Romance language varieties mainly spoken in the in South- Eastern . In this paper, we explore the agreement patterns in two mutually understandable varieties of Romansh spoken in the North-Western Grisons: Romansh Sursilvan and Romansh Tuatschin. Romansh Sursilvan serves as local standard and is natively spoken by a large part of the local population. It is also taught in regional schools. Romansh Tuatschin is a separate variety within Sursilvan dialectal area and a minority language, still undocumented and unwritten. It is however still natively spoken by approximately 1,500 speakers in the Val and actively transmitted between generations. Most if not all Romansh speakers are multilingual. Speakers in the Sursilvan area speak at least their variety of Romansh and Swiss and Standard German. All varieties are a highly mixed and the insertion of German lexical items or full utterances is ubiquitous. Our work is based on original elicitation and corpus data from ongoing field work in the Val Tujetsch area. The standard word order in Romansh places the subject before the verb. However, inverted orders are also commonly found and give rise to a series of interesting phenomena. (Examples hereafter are from Romansh Tuatschin, but similar phenomena can be observed in Sursilvan.) Some verbs in the Sursilvan varieties of Romansh have special forms for inverted orders, sometimes involving sandhi phenomena with following pronouns: el è ‘he is’ vs. sèl ‘is he’ i è ‘it is’ vs. sai ‘is it’ First person plural pronouns have a special enclitic form in postverbal contexts: nus fagiain ‘we do’ vs. fagiainsa ‘do we’ Pronouns can sometimes be omitted if they occur postverbally. For Sursilvan, Cathomas (2015) shows that this is particularly frequent for (although not exclusive to) second person pronouns. Cu té fevas pins cantavas (té) sut la duscha when you were small sang (you) under the shower “You used to sing in the shower when you were small” Both Sursilvan and Tuatschin present a lot of sandhi phenomena at word boundaries, such as (mainly) adjacent vowel deletions. Cu’l luvrava giu Cuera vev’el duas soras o Mustair cu-el luvrava giu Cuera veva-el duas soras o Mustair when-he worked in had-he two sisters in Dissentis “He used to have to sisters in Dissentis, when he was working in Chur” As can be seen from the example above, some of the sandhi phenomena affect verbal endings. But the absence of verbal endings is not restricted to typical sandhi contexts. Some verbal endings are simply dropped in postverbal subject contexts.

Our collected corpus data so far also shows that: a) Truncated forms with either shortened verb or pronoun forms are an important phenomenon (affecting 1⁄4 of all Tuatschin sequences involving adjacent verbs and personal pronouns) and b) truncations are more frequent in postverbal subject contexts (about 2⁄3 of all such sequences), c) while second person pronouns seem to be indeed dropped more readily in both dialects.

Our study investigates the constraints on the presence and absence of both verbal endings and postverbal pronominal subjects in Tuatschin and Sursilvan. We focus in particular on the differences between the two dialects that are linked to differences in their morphological organisation. Verbal inflection in the two dialects is highly Tuatschin Sursilvan similar. The main difference between Sursilvan 1sg ju conta jeu contel and Tuatschin verbal inflection is that Tuatschin 2sg té contas ti contas presents an -a syncretism between the first and 3sg el/ella conta el/ella conta third person singular endings, while Sursilvan has 1pl nus cantain nus cantein a designated first person singular ending -el (in all 2pl vus cantais vus canteis 3pl els/ellas contan els/ellas contan tenses and modes, for practically all verbs). An example thereof is given in the table across for the present indicative subparadigms of the verb CANTAR/ CANTÀ ‘to sing’. Given this difference between the two otherwise almost identical inflectional systems, we expect that the likelihood of the presence/absence of both the inflectional endings and the pronouns will vary between the two dialects.

Elicitation of existing structures for Tuatschin and Sursilvan speakers so far suggests that : a) Speakers vary individually in their willingness to drop pronouns. b) Tuatschin speakers seem to drop pronouns less readily, although c) younger Tuatschin speakers might be more willing to drop pronouns, d) but when they do, they sometimes insert an additional Sursilvan 1SG ending (possibly as a consequence of increased mixing between the two varieties); this suggests a link between first person inflectional marking and optionality of postverbal pronominal subjects.

Based on a detailed quantitative corpus-based study of agreement patterns in Tuatschin in Sursilvan, this paper will describe the distribution of full and truncated verb forms and their interaction with the optionality of postverbal pronominal subjects: 1: Presence/absence of inflectional endings: a) In sentences with postverbal subjects, Tuatschin 1SG/3SG -a is likely to be absent more often than Sursilvan 1SG -el and 3SG -a. The reason is that in sentences with postverbal subjects, Sursilvan possesses an additional verb form contrast which can be used for early person differentiation, while in Tuatschin, the differentiation is only achieved later through the postverbal pronoun. As a consequence, shortening the verb form allows for an earlier utterance of the discriminating element in Tuatschin. b) However, if the 1SG verb ending -el is dropped in Sursilvan, jeu is likely to drop less frequently than in cases where -el is present, since the task of identifying the 1SG then falls upon the pronoun, thus making it more informative than when -el is present. 2: presence/absence of postverbal pronominal subjects: c) Conversely, the Sursilvan 1SG pronoun jeu is likely to drop more frequently than the Tuatschin 1SG pronoun ju, because Sursilvan relies on the verbal ending to differentiate between 1SG and 3SG forms. In Tuatschin, the paradigms contains an additional syncretism such that the discriminative function of the pronoun is higher. d) Given that both dialects have distinctive second person singular marking, second person postverbal subjects are likely to be more easily dropped (as observed by Cathomas 2015). References Cathomas, C. 2017. Von ‚I dont Know! zu „Kei problem chara!!’ – Eine korpuslinguistische Untersuchung zu rätoromanischen SMS unter besonderer Berücksichtigung verschiedener Formen und Funktionen von Code- Switching. PhD Thesis, Bern University.