The third day of CamCoS 9 was longer than the previous two days, with altogether seven talks (including two invited talks) running till 1 a.m. in my time zone. But I stayed in the Zoom room till the end (well, except for a couple disconnections due to my poor bandwidth), partly because I wanted to make this blog series complete 😃. As usual, I tried to ask questions, at least for the first few talks (i.e., before I got dozy), and also took some notes. Unfortunately I couldn’t concentrate very well from time to time because I had had another day of compulsory admin training (crazy bureaucracy!!). So, this post may be a bit cursory in science and filled with a lot of my personal whimsical thoughts/feelings, but I hope I’ve at least got the key information correct. All handouts can be found on the conference website (2020-09-12 update: except that for the second talk, which for some reason is no longer available).
Six out of the seven talks today are about the nominal domain of human language grammar. This forms a nice contrast with the previous two days, which mainly focused on the verbal/clausal domain. The first talk (invited) today, given by Beata Moskal and entitled “How to be inclusive,” is about the morphological process of suppletion, in particular the absence of “ABA” patterns in certain grammatical domains. The ABA gap in suppletion was first observed for adjectival degree comparison morphology (Bobaljik 2012). The letters A and B here stand for word stems. For example, in English there are positive–comparative–superlative patterns like cool–cooler–coolest (AAA, with a single stem cool) and good–better–best (ABB, with two stems good and be(tt)) but nothing like good–better–goodest (ABA). Bobaljik’s team had examined a large number of languages and the gap seemed to be crosslinguistically robust.
Other grammatical domains showing the ABA gap include several pronominal categories, such as case, number, and clusivity (e.g., the inclusive “we”). This talk mainly discusses the latter two categories, with abundant data drawn from a huge number of languages (e.g., Limbu, Ayiwo, Mangarayi), most of which are unheard-of to me. More specifically, the speaker tries to explain away potential counterexamples for the no-ABA hypothesis. A key concept she uses is “markedness.” To be honest I have found this concept fundamentally elusive for long, both in phonology and in morphosyntax. I’m not even sure if there’s a rigorous definition for it, since what counts as marked and what doesn’t all sound very subjective to me, and hence I think theories designed to describe or derive markedness are also subjective. So yes, the proposed theory does explain the data (and presumably does it well), but no, I don’t think the foundations of the theory have been made clear enough, though of course it’d be unfair to say that’s the speaker’s obligation, for markedness-based analyses have been around for decades after all.
The second talk of the day was jointly given by M. Teresa Espinal and Sonia Cyrino (both authors were present) and entitled “A syntactic approach to the variation in the expression of indefiniteness, (anti-)specificity and partitivity in Romance.” The speakers started off by presenting a range of indefinite nominal phrases in Italian and Brazilian Portuguese, each with a different grammatical function, such as sheer indefiniteness (1a), indefiniteness with specificity (1b) (in that it is some specific boys that the speaker has seen), indefiniteness with partitivity (1c), etc.
(1) [Brazilian Portuguese]
a. Eu vi meninos.
“I saw boys.”
b. Eu vi uns meninos.
“I saw some boys.”
c. Eu vi alguns dos meninos.
“I saw some of the boys.”
The purpose of the talk is to provide a unified syntactic account for these and other similar expressions, and the account proposed by Espinal & Cyrino is a semantically oriented one, though they call it syntactically driven. For instance, for sheer indefiniteness they postulate a semantic operator of type $\langle e,\langle e,t\rangle\rangle$, which is spelled out as de in some Romance languages (e.g., Italian) and covert in others (e.g., Brazilian Portuguese). The speakers continue to postulate semantic operators for other types of indefiniteness, and much of the talk is dedicated to justifying these operators with empirical evidence. Since the postulated operators are assumed to be involved in the syntactic derivation, Espinal & Cyrino conclude that variation resides in syntax; and since the operators can have different realizations in different languages, they conclude that variation also resides in the mapping from syntax to phonology (so, at PF in a broad sense).
I guess what they want to suggest is that some or all of the indefiniteness-related operators exist in Romance languages but don’t necessarily exist in other languages, because otherwise this talk de facto presents us with no variation in syntax but merely shows that the same syntactic structure can have various morphophonological exponents. So, this makes me wonder what the speakers mean by “variation” when they conclude that “variation is at syntax,” and more crucially, in what sense such variation is parametric, because in my understanding parametric variation is crosslinguistic by definition; that is, it should concern switched-on/switched-off grammatical options across languages. For example, the null subject parameter says a language either has or doesn’t have null subjects in a certain properly restricted domain, and given such a properly restricted domain, a language cannot simultaneously allow and disallow null subjects, for that would be a parametric contradiction. However, that doesn’t seem to be the case in the theory advanced in this talk, because the several indefiniteness-related operators can apparently coexist in an individual Romance language. They simply underlie different constructions.
The third talk, by Ellen Brandner and entitled “PF identity is not syntactic identity” (though this title isn’t reflected in the slides), is about a determiner doubling phenomenon in Southern German dialects. For example, while in Standard German one says either (2a) or (2b), in southern dialects like Bavarian and Alemannic one can actually say something like (3), with two indefinite determiner eins.
(2) [Standard German]
a. ein so lieber Mensch
“a so nice person”
b. so ein lieber Mensch
“such a nice person”
(3) [Bavarian German]
ein so ein lieber Mensch
“a such a nice person”
The speaker asks three questions: (i) What syntactic positions do the two eins occupy? (ii) What meaning does the second ein contribute? (iii) What parameter can be formulated to cover this crossdialectal variation? She hypothesizes that indefinite determiners can either occupy a “bounded mass” head (BM) position or an “individual” head (Ind) position. BM semantically picks out a subset from some larger set denoted by its sister (in which sense it is similar to the function of a partitive), while Ind picks out an individual. She assumes that which of the two positions the indefinite determiner occupies is a point of crosslinguistic variation.
That being said, Brandner doesn’t ascribe the doubling phenomenon to the (un)availability of the BM/Ind positions. Instead, she views it as the result of misanalysis, either of the particle so ‘so’ or of weak quantifiers like ein wenig ‘a little’ and ein bisschen ‘a bit’. Overall, she argues that some of the doubling cases (in particular the ein so ein case) are PF-induced while others (in particular the ein ganz ein case) are “PF-identical with the (nominal) weak quantifiers (ein wenig),” though I don’t really understand what that means. Maybe she did clarify this in the talk but I probably was dozing off 😪. Overall, however, I did enjoy this talk a lot. I may even try and speak to my German-speaking friends with indefinite determiner doubling and see how they react. 🤓
After the third talk there was a 20-min break as usual. I quickly grabbed some food and made a cup of coffee. Then I sat back in front of the computer and listened to the pop music played by the hosts, though my Internet was so slow that the music constantly got choppy.
The fourth talk (another speed talk) began before long, and I also felt better after eating and drinking. On that note, I do miss the coffee breaks at previous CamCoS conferences, where there were always nice coffee/tea/cookies and sparkling conversations. I guess we won’t have that privilege again until the corona crisis is truly over 😞. Anyway, the fourth talk was entitled “Multiple loci of variation: A case study of negative concord in English varieties” and given by Mary Robinson. It’s about negative concord in different English varieties (Standard English, Scotland English, Reading English, African American Language, etc.). Robinson observes that speakers of many English varieties actually possess grammatical knowledge of negative concord, whether they regularly use it or not. As an example she presents a snippet from a podcast featuring a man from Ontario:
Wasn’t that deep, 30 feet deep roughly. Couldn’t come up with nothin’. Didn’t see anythin’.
Below are a few more examples mentioned in the talk:
a. They’ve nae got nae choice. [Scotland]
b. Nobody didn’t touch that but her. [Appalachian]
c. I don’t believe there’s no God. [African American]
Robinson’s overall conclusion is that crosslinguistic variation regarding negative concord exists both in syntax and at PF. On the one hand, she analyzes negative concord by a movement-based theory, with syntactically conditioned allomorphy. On the other hand, she adopts Heidi Harley’s root theory and deduces from the fact that negative concord is sensitive to noun roots that the root-conditioned aspect of its variation must be holding at PF (according to Harley the content of a root only becomes available postsyntactically).
I personally find the negative concord phenomenon very interesting, and I also believe that any adequate theory or implementation of human language grammar must take it into account. I remember raising a question about it to an NLP researcher working on negation in this year’s Applied Category Theory conference, because apparently her intricate type assignments didn’t leave room for negative concord (nor had she intended to do so judging from her response to my question). Overall, I get the impression that NLPers’ knowledge and modeling of grammar is a lot more simplified than linguists’. But if so, why don’t we see much interdisciplinary collaboration (or even nonbiased communication)? It’s not like we don’t need each other’s expertise. Then what’s the reason?
The fifth talk was given by Allen Asiimwe, Jenneke van der Wal, and Maria Kouneli and entitled “Rukiga augments and Greek articles: one analysis?” The talk had a single empirical focus: a peculiar morpheme optionally prefixed to modifiers of noun phrases in a Bantu language called Rukiga. For example, e- can be optionally used in (5a) but é- can’t be used in (5b). Here e and é are presumably allomorphs of the same grammatical morpheme.
a. Naabugan’ ékihugúhúgu (e)kihango. ✅
“I have found a big butterfly.”
b. Éízóób’ élíri hale. ❎
“the sun that is far”
The talk was delivered by Jenneke on behalf of her team, and as usual her slides are very colorful, very economic in words, and had many pictures. I like her style of presentation because it is very audience-friendly. And I’ve been trying to learn from her in the past few years, though with little success—I always ended up cramming way too much content in my slides. I think I’ll try harder in the future. After presenting some Rukiga data, Jenneke turned to compare Rukiga with Greek, observing that a similar “determiner spread” phenomenon exists in the latter, as in (6).
to kokino (to) podilato
the bike (the) red
“the red bike”
In the end, Jenneke proposed not a unified structure but three separate structures for the Rukiga modifier prefix (which she calls an “augment”).
The sixth talk, by Daniel Greenson
(unfortunately I can’t find his online profile) and entitled “Revisiting Montalbetti’s effects: Evidence from Brazilian Portuguese, Spanish and English,” is about the crosslinguistic variation re the “overt pronoun constraint,” which is formalized in Montalbetti (1984) as “overt pronouns cannot link to formal variables iff the alternation overt/empty obtains.” To be honest I don’t understand this statement at all (can’t even parse it) out of context. I gather from Greenson’s data that this has to do with pronoun binding, though I certainly need to read the handout more carefully, which contains an impressive amount of data.
EDIT (2020-09-13): Montalbetti’s constraint more exactly has to do with the referential interaction between quantified DPs and embedded arguments, as in the Spanish sentence Toda persona cree que ella es inteligente ‘Every person thinks that she is intelligent’, where “every person” and “she” cannot refer to the same person. Here coreferentiality is only possible when ella ‘she’ is omitted, as in Toda persona cree que es inteligente ‘Every person thinks that they are intelligent’. While Montalbetti attributes this contrast to the properties of null arguments, the current talk demonstrates that it’s not as simple as that and that in fact a non-null-argument-based analysis is more advantageous. (Thanks to the speaker for clarifying this in the comment section!)
For some reason For logistics reasons the talk was a prerecorded one, but the speaker was present for live questions. And both the talk and the Q&A were quite impressive, especially considering the speaker is still an undergraduate student, which makes me wonder what I had been doing when I was an undergrad—surely nothing properly linguistic!
The seventh and final talk of the day (i.e., the second invited talk) was given by Klaus Abels and entitled “The absence of Foc in the clausal spine.” For me personally this was the most refreshing talk of CamCoS 9, because the speaker boldly challenged the cartography theory (which has now become a highly popular enterprise). I like hearing different voices both in research and in life, but more importantly I think academia should be a safe space for different theoretical views, so that researchers holding nonmainstream/unpopular perspectives wouldn’t need to hide or disguise themselves for fear of losing opportunities. So, regardless of the content of the talk, it definitely won me over in spirit.
Of course, its content was also incredibly good. Abels started by directly taking the side that information structural features like [+focus] aren’t encoded in the clausal spine (contra cartography). Then, he went on to present arguments from several different angles and grammatical domains to justify his bold claim (which, by the way, engendered a few more bold claims). Finally, he concluded that “assuming FOC0 in the left periphery of the clausal spine stands in the way of developing an explanatory theory of syntax.” If you have missed this talk—and especially if you are also unsure about the foundations of cartography—I highly recommend it to you. The recording is on the conference website and there are both slides and a more elaborate handout.
In sum, the third day has been a sea of information! I can see that the conference has largely made its “categorial turn” today (from verbal to nominal) and can’t wait to see what else it has to offer on the fourth day. I’ll make sure to complete my blog notes. 😃