Talks on the second day of CamCoS 9 are mostly concerned with agreement phenomena. There’s again a diverse range of languages being examined, including Basque, Mansi/Hungarian, West-Germanic dialects (e.g., Stadskanaal, Frisian, Limburgian), Icelandic/Galician, and Gungbe/Amahuaca/Tundra Nenets. I’ve generally found these talks slightly harder to follow than those on the first day, maybe because (person/number-)agreement isn’t my cup of tea. Anyway, below I try to summarize what I have learned, and again all handouts can be found on the conference website.
The first talk of the day, by Maia Duguine and entitled “Clausal embedding in Basque: Syntactically disharmonic but harmonically realized,” raises the question of whether the final-over-final condition (FOFC) put forward by the ReCoS group (i.e., initiators of the CamCoS conference series) should be treated as a syntactic or a PF phenomenon. The purpose of the talk is to argue for a PF-based approach, with evidence from clausal embedding patterns in Basque. Specifically, the discourse domain of Basque clauses is head-initial, as in (1a), but when a clause is embedded by a lexical head, say a noun, a verb, or a preposition, the embedding-point merger (i.e., the merger of the clause and the lexical head) is head-final, as in (1b). The examples are take from Duguine’s slides.
a. Ez du emakumeak leihoa ireki.
“The woman didn’t open the window.”
b. leihoa ireki ez duen emakumea
“the woman who didn’t open the window”
In (1a) the negative particle ez precedes the auxiliary du and the rest of the sentence (hence the head-initiality of ez with regard to the proposition it negates), while in (1b) the noun phrase emakumea ‘the woman’ follows its modifier leihoa ireki ez duen ‘that who didn’t open the window’ (hence the head-finality of emakumea with regard to its modifier). According to Duguine, this in theory creates a final-over-initial scenario and violates FOFC, but Basque employs a variety of PF strategies to save the derivation from such violation, by way of allowing only one overt realization between the base-generated version of the (potentially complex) head of the embedded clause and its movement-induced copy—and the eventually spelled-out version is always the base-generated (aka “lower”) version. This leads Duguine to conclude that the evaluation of FOFC happens at PF, because if FOFC is a phonological constraint, then one could simply treat non-spelled-out constituents as not there and thereby treat Basque examples like (1b) above as FOFC-compliant.
I did learn a bit of Basque during my PhD, and I still want to learn more. As a language learner I find the generative analysis of Basques grammar in the talk and the references therein very helpful. Regarding the theoretical proposal of the talk, however, I am less sure how much I agree. I remember that the definition of FOFC explicitly states that it is only meant to hold for categories in the same extended projection. That means the whole issue of clause embedding falls outside the applicability domain of FOFC, because the embedding category (basically a lexical category) and the embedded clause obviously do not belong to the same extended projection of the same lexical head.
The second talk, by Katalin É. Kiss and entitled “The varying relation of case and verbal agreement in the Ugric languages,” argues for a biconditional relation in Ugric languages between grammatical relation–discourse role (G-D) fusion and case-agreement (C-A) correlation; that is, case and agreement marking are correlated in a Ugric language if and only if its grammatical functions and discourse roles are fused. By case-agreement correlation É. Kiss means the cooccurrence of case marking on nominal arguments and agreement marking on verbs; for example, in Hungarian a verb may agree with its subject and/or object, with the subject/object being case-marked, as in (2). By grammatical function–discourse role fusion É. Kiss means the necessary overlapping of the two grammatical notions; for example, in Eastern Mansi subjects are necessarily also topics, as in (3). The examples are taken from É. Kiss’s slides, with slight adjustments.
A szintaxis dollgozatot megírta Mari.
“The syntax paper, Mary has written (it) up.”
(3) [Eastern Mansi]
Näγ kɔnə tiγ tatwəsən?
“You were brought here by whom?”
In (2) the -a ending of the verb megírta ‘has written’ marks an action from a third person singular subject (here “Mary”) to a definite object (here “the syntax paper”), with the latter being marked with the accusative case -(o)t. In (3) the sentence is more naturally expressed in English as Who brought you here? but in Eastern Mansi there’s no way to say this with “you” being both the topic and the direct object, hence the passivization.
The above-mentioned biconditional leads É. Kiss to propose a three-way typology for Ugric languages: (i) languages with both G-D fusion and C-A correlation (e.g., Eastern Mansi), (ii) languages with neither G-D fusion nor C-A correlation (e.g., Hungarian), and (iii) languages with partial G-D fusion and partial C-A correlation (e.g., Northern Mansi, Eastern/Northern Khanty). The bulk of the talk is dedicated to presenting data from these language varieties, about which I have little to comment on. I guess a more general question I have (which I remember was also raised by Anders Holmberg in the Q&A period) is about the wider prediction of the proposal beyond Ugric languages. In other words, is the three-way typology only meant for Ugric languages or does it hold more generally in human language?
The third talk (also the second speed talk of the conference), by Astrid van Alem and entitled “Intervention effects on complementizer agreement: Revisiting the agree vs. PF debate,” tackles a commonly observed “complementizer agreement” phenomenon in many nonstandard varieties of West-Germanic languages. The basic observation is that in these varieties some clausal complementizers can take a quasi-agreement suffix (mainly in second person singular), as in (4)–(6). The examples are taken from van Alem’s slides.
Ik wait dastu de woarheit zegst.
“I know that you are telling the truth.”
…darre wiej de westrijd wint.
“…that we win the game.”
…datstich/toe de westrijd geis winne.
“…that you are going to win the game.”
In the three West-Germanic dialects above the complementizers are the three words beginning with da-, and the complementizer agreement markers are respectively -st- in (4), -e in (5), and -s- in (6). Following these markers are the subject pronouns, either as clitics (e.g., -u ‘you’ and -tich/toe ‘you’) or as standalone words (e.g., wiej ‘we’).
The purpose of the talk is to demonstrate that the above phenomenon is neither syntactic agreement nor a PF phenomenon but doubled cliticization via syntactic movement. A key advantage of this theory is that it can explain the intervention effect accompanying the phenomenon; namely, the effect that when the complementizer and the subject of the complemented sentence are separated by an extra element (usually an adverb like “also” and “even”) the quasi-agreement suffix disappears in some West-Germanic language varieties, as in (7).
a. …datst ek do fegetarysk ytst. ❎
“…that you, too, eat vegetarian.”
b. …dat ek do ytst al fegetarysk. ✅
“…that you, too, sometimes eat vegetarian.”
As (7a) shows, when the subject is modified by the adverb ek ‘also’ the complementizer agreement pattern is unacceptable, and the only way to express the intended meaning is as in (7b), where a different word order is employed.
I always like learning new stuff about Germanic languages because they are cool in a number of grammatical aspects that interest me (such as compound/particle verbs). I also think the three-way pronoun classification (originally put forward in Déchaine & Wiltschko 2002) the current talk draws inspiration from is appealing. The only component in the proposed theory that I feel uncertain about is the nature of the functional projection dominating DP in the extended nominal phrase. Van Alem seems to identify it as a focus phrase (FP), presumably following van Craenenbroeck & van Koppen (2019), but she doesn’t specify whether/how this FP is related to the FP/FocP in the extended verbal phrase (i.e., are we dealing with the same category). Also, it seems to me that this projection is mainly used here as a landing site, which means the pronoun movement is neither triggered by the F head nor associated with any [+focus] feature. As a result, the complementizer agreement marker (i.e., the moved constituent) doesn’t really bear any focus interpretation or prosody. So, I think it might be worthwhile to explore some alternative analyses and compare them with the FP-based analysis before drawing final conclusions.
After the third talk there was a 20-min coffee break as yesterday, which I used to grab some dinner because I hadn’t had time to eat earlier (due to a long training day in the institute I’m currently working for XD). In today’s coffee break the organizers played some pop music (reportedly chosen by Anders Holmberg). The music was nice—a bit rock’n’roll and a bit blues. It was called We Gotta Get Out of This Place if I remember correctly. There were also other songs, but I can’t remember them, and since I kept losing Internet connection and being forced to leave and reenter the Zoom room, I also didn’t manage to save their names from the chat window in time. 😞
I struggled a bit to follow the fourth and the fifth talk, perhaps because they were both data-heavy and it was already quite late in my time zone, and all my limbs were sore after the full-day staff training. Anyway, the fourth talk was given by Brian Gravely and entitled “Only syntactic movement will do: Morphosyntactic variation in Galician and Icelandic.” Its purpose is to show that postsyntactic movement-like operations should be dispensed with and that syntactic movement alone is enough to account for the relevant data that have motivated postsyntactic movement theories. The empirical focus of the talk is on cliticization. This is yet another research area outside my purview—“clitic” is one of those terms that everyone talks about but I have never really understood the gist of. Not that the phenomena behind it are uninteresting; I just don’t get along with the term for whatever reason, to the extent that I sometimes find it easier for my brain to make sense of the relevant data by replacing every “clitic” with “thing.” 😂
I won’t comment on the empirical details in the talk due to my lack of background knowledge, but I do feel that its overall conclusion that “all movement operations must occur in the syntax proper” might be too strong—not that I’m against this view; I’m a wholeheartedly minimalist person and would happily avoid all quasisyntactic yet nonsyntactic operations, including but not limited to lexical rules, PF operations, and covert/LF movements. I just feel that that conclusion cannot be rigorously reached by the reasoning in the talk, because even if the two case studies on allomorphy are both on the right track (they seem to be) we still cannot conclude that all cases of allomorphy are syntactic in nature; and even if all allomorphy phenomena are indeed syntactic, we still cannot safely conclude that all (quasi-)movement operations are syntactic. Each step of the above reasoning seems to be an overgeneralization from “some” to “all.”
The fifth and final talk of the day was given by Kenny Baclawski and Nico Baier and entitled “Morphological evidence for syntactic information structural features.” I think this talk could have been placed on the first day, because it is mainly about the discourse domain of sentence structure (i.e., topic, focus, etc.). The speaker (K.B.) began by laying out two approaches to “information structure” phenomena and then went on to present crosslinguistic data (from Gungbe, Amahuaca, and especially Tundra Nenets) that might corroborate one or the other. She suggested that among the three case studies the Tundra Nenets data are only compatible with a syntactic information structure feature–based analysis and therefore count as categorical evidence for the syntactic approach; namely, the approach that treats information structure features like [+topic] and [+focus] as formal, syntactic features. This talk doesn’t directly touch on the theme question of CamCoS 9 (i.e., whether parametric variation is in syntax or at PF), but it might lay groundwork for a potential answer to it, for if information structure features are syntactic, then parametric variation in information structure is also likely to be syntactic in nature.
In sum, talks on the second day mainly address the syntax-vs.-PF question (either directly or indirectly) by investigating agreement phenomena. Agreement is a fascinating aspect of human language grammar, and linguists love it (though many language learners hate it). There’s a good balance between data and theory throughout the day, so both the empirically minded and the theoretically minded can find something to appreciate. As for me I am most impressed by the wide range of languages the speakers have looked into. Actually I often wonder how linguists who study multiple exotic languages manage to acquire their data. Both large-scale surveys and fieldwork trips require money and people skills, so I guess they must have both. Anyway, I think all five talks today are great (despite my sporadic challenging remarks above), and I’ve learned a lot from them all. I look forward to the remaining days of the conference! 👨💻