Today is the fourth and final day of this year’s CamCoS conference. I am already starting to feel nostalgic, but as the cliché goes: all good things must come to an end. I know that the original CamCoS series had already reached its end a while ago (not long after the ReCoS project had ended), but since there is a “new” in the official title of this year’s conference (i.e., CamCoS 9 New), I assume this could be the beginning of a splendid new era?
Following a long day yesterday, the schedule for today has returned to its normal size, with five talks distributed into two sessions separated by a 20-min break. During the break Michelle dedicated an online wall of messages to Anders, where other linguists (me included) expressed appreciation for his decades-long contribution to the field, thanked him for the kind help he had given out, and wished him all the best for his postretirement life. It’s always heart-warming to experience moments like this, and I hope that by the time I retire I will feel happy about the good I’ve done to the world.
Alright, before I get all corny and melancholy, I’d better start noting down what I’ve learned from the five remaining talks of CamCoS 9. Similar to yesterday, today’s categorial theme is also the nominal domain. The first talk, given by Marta Ruda and entitled “Analyses of null arguments meet a bare NP, consistent NS, partial NO language,” is about the way null subjects are licensed and derived in consistent null subject languages. The speaker began by presenting two previous approaches (that of Holmberg/Roberts and that of Barbosa) side by side and then developed a middle-way approach based on them, using Polish as a main source of data but also drawing crosslinguistic evidence from Kashubian and Russian—she explained the relevant variation in terms of the (un)interpretability of a [+person] feature on T. Some of her observations are that in Polish the definite interpretation of null subjects doesn’t hinge on the presence of a topic, as in (1a), and that topic change doesn’t require an overt pronoun either, as in (1b).
a. ø Rozmawiałam z mamą Tomka. ø Powiedziała, że ø nadal jest w szpitalu.
“I talked with Tomek’s mom. She said that he/she is still in hospital.”
b. Jan czekał. Anna przyjechała. ø Otworzył jej drzwi. ø Weszła bez słowa.
“Jan waited. Anna came. He opened the door for her. She entered without a word.”
In (1a) there are three null subjects, which are indicated by the empty set symbol ø. The second ø refers to Tomek’s mom and the third one refers to either Tomek or his mom, and neither piece of referential information depends on a topic in the preceding discourse (for the postverbal “Tomek’ mom” isn’t a sentential topic). Similarly, in (1b) there are two null subjects (again indicated by ø). The first one refers to Jan and the second one refers to Anna, and the referential relations aren’t disturbed by the topic shift (from Jan to Anna) in the preceding discourse.
Though Ruda didn’t say this explicitly, I gather based on her argumentation that she’s on the syntax side of the syntax-vs.-PF debate. Null subject languages and pro-drop phenomena more generally have been a long-standing hot topic in the field as well as a motivating example for the principles and parameters framework. My personal impression is that transformational-generative approaches to the topic (including all three approaches mentioned/developed in the current talk) tend to differ from one another only in technical details, which makes me wonder how a fundamentally different theoretical framework (e.g., generalized phrase structure grammar, categorial grammar) might tackle it on the one hand and how the whole pro-drop issue is perceived in computational linguistics and NLP on the other hand (say, whether it is treated as a thing at all). We as generative syntax pupils are so used to the significance and technicalities of pro-drop that I am a wee bit curious as to what the bigger picture looks like beyond our own research field.
The second talk, given by Silvia Terenghi and entitled “Structural differences in the syntax of person indexicals,” is also about person features, so it partly crosses paths with the first talk. More specifically, Terenghi’s research interest lies in the diachronic asymmetry of person features: they tend to be stable in personal and possessive pronouns (e.g., I–my) but unstable in demonstrative forms (e.g., this, here). While the ternary system (i.e., first, second, and third persons) has stayed stable in the Romance languages in the former domains, it hasn’t in the latter domain, with the erstwhile ternary system (e.g., here, there, and over there) frequently being reorganized into various binary systems (e.g., here and there). Terenghi illustrates this by the examples below:
a. (erstwhile) aicí ‘here’, aquí ‘there’, allí ‘over there’
b. (now) aquí ‘here’, allí ‘there’
(3) [Rioplatense Spanish]
a. (erstwhile) acá ‘here’, ahí ‘there’, allá ‘over there’
b. (now) acá ‘here’, allá ‘there’
The main proposal of the talk is that the above-mentioned diachronic asymmetry is structurally derivable via the notion of “structural salience” (Polinsky 2018). Terenghi’s analysis of demonstratives in terms of person features (embedded in Harbour’s 2016 framework) is quite novel, because as Anders Holmberg mentioned in the Q&A period, conventional analyses of demonstratives are usually based on spatial locations rather than persons. The way Terenghi translates structural salience into (in)stability is by assuming that “the most salient (hence: stable) feature is the first to compose with its hosting head,” whatever that means. I think I need to read Polinsky (2018) to understand the salience-based analysis better, because the slides for this talk don’t provide enough background information on this. I also think an explicit modeling of structural salience can potentially benefit many of us, because it’s one of those ideas that have occurred to many but haven’t been given enough serious attention in a minimalist (i.e., adhoc-assumption-free) context. So, I’ll add Polinsky (2018) to my reading list. 📚
The third talk (the last speed talk), given by Andreas Blümel and entitled “A uniform NP-approach to (poly-)definiteness and nominal concord,” proposes a new theory for nominal phrases based on Oishi (2015) as well as Chomsky’s (2019) introduction of the Hilbert epsilon operator into syntax. I can’t say I have understood the talk very well, because I don’t know anything about Oishi (2015) or Chomsky (2019) and there isn’t enough background information in the talk to help me fill in the blank. So, I probably need to read both before I can figure out why they are so important for Blümel (as well as how well grounded they themselves are).
Unfortunately, I haven’t found much information about Oishi’s article or the edited volume it is in, so perhaps I’ll never be able to find out what he did in that article. I did ask the speaker about the Hilbert epsilon operator but didn’t get a thorough answer. According to the reference list in the slides, Chomsky had introduced it in a lecture series he had given last year at UCLA, and the good news is that the lecture recordings are available online. 📌 So what I’ll do is watch Chomsky’s lectures, do some background reading, and then write another post or even a proper squib to report what I’ve learned about this new operator. I feel it a bit odd that no one has written anything about it yet, because usually Chomsky’s new ideas catch on fairly quickly among his followers.
The fourth talk, jointly given by Madeline Bossi, Emily Drummond, and Zachary O’Hagan and entitled “Two case studies in morphological person restrictions,” was the second recorded presentation of the conference. The talk was about person combination restrictions on the direct and indirect objects (more exactly on their clitic/agreement realizations) of ditransitive verbs in world languages. For example, in French the direct object cannot be second-person when the indirect object is third-person, hence the ungrammaticality of the sentence in (4a). Compare it with (4b), which has a third-person indirect object plus a third-person direct object and is grammatical.
a. Je te lui ai présenté. ❎
“I introduced you to her.”
b. Je le lui ai présenté. ✅
“I introduced him to her.”
While previous approaches to the phenomenon are mainly syntactic, the speakers propose a novel morphological analysis in this talk, because the syntactic approach can’t explain the data they have found in two underdescribed languages: Caquinte and Kipsigis. The more general goal of the talk is to show that syntactic and morphological perspectives must be combined to yield a more complete typology. I think a good thing the speakers did is that they remembered to come back to French in the end. This was good because there were more people in the audience familiar with French than people familiar with Caquinte/Kipsigis, so coming back to French could help the audience better understand the predictions of the talk.
The fifth and final talk of the day, which is also the final talk of the conference, was given by Zahra Mirazzi (unfortunately I can’t find her online profile) and entitled “Microvariation in aspect splits in nominative-accusative system.” This talk studies the semantic contributions of two series of agreement markers (classified based on their syntactic positions) in four Northeastern Neo-Aramaic languages: Senaya, Jewish Zakho and Christian Barwar, Jewish Amadiya, and Shaputnaya. What’s peculiar about the two series of agreement markers in these languages is that while their word orders are fixed, their functions aren’t. Sometimes one series of markers mark subject agreement while the other series mark object agreement, as in (5a), and sometimes it’s the other way around, as in (5b).
(5) [Shaputnaya] (red: subject marker, blue: object marker)
a. ana ourzey xazennun
“I will see those men.”
b. ana Zahra xezyali
“I saw Zahra.”
In (5a) the first suffix on the verb -en marks subject agreement (i.e., it agrees with “I”) and the second suffix -nun marks object agreement (i.e., it agrees with “those men”). In (5b), by contrast, the first suffix on the verb -a marks object agreement (i.e., it agrees with “Zahra”) and the second suffix -li marks subject agreement (i.e., it agrees with “I”). By carefully examining the data, Mirazzi singles out the grammatical category of aspect as the cause of the agreement marking split.
Overall, I get the impression that today’s talks are still agreement-heavy, so it seems agreement is really a long-standing popular topic in syntax. Personally, though, I kind of wish there had been some more talks studying other linguistic phenomena, such as word formation, event structure, speaker-addressee interactivity, and so on, because after listening to so many talks on agreement my brain is getting a bit numb. This doesn’t mean the talks on agreement aren’t good, of course. All talks have been fantastically delivered and impressively insightful. Besides, I remember from previous years that there used to be several talks on Asian languages, but this year there have only been two (both on the first day). I’m not sure whether there’s any association between this and my foregoing observation re agreement-based talks, but it is a fact that many Asian languages lack agreement phenomena.
So, what have I learned from CamCoS 9? Well, I’ve learned more about the discourse domain (aka left periphery, aka information structure) and the debate over its place in grammar; I’ve learned more about various agreement phenomena and how they could be analyzed; I’ve learned more about the structure of nominal phrases as well as the specific functionality of each nominal functional category; and I’ve learned about a bunch of languages that I had never heard of.
And what about the theme question of the conference? Namely, where is the variation: syntax or PF? To be honest I don’t think the question has been seriously touched on a lot throughout the conference. We have been demonstrated that some variations are best analyzed syntactically while others are more likely to be PF phenomena, so at the end of the day we’re still left with the same question: syntax or PF? Or maybe this shouldn’t be a question at all. As a few speakers pointed out, crosslinguistic variation surely exists both in syntax and at PF.
My personal take is that we can ask the syntax-or-PF question, but only in an adequately restricted theoretical context. The first thing that needs to be clarified is what “syntax” and “PF” respectively mean. I don’t think researchers have unanimous definitions for these terms. For instance, some seem to assume a broader definition of (narrow) syntax than others. My own definition is certainly a very narrow one, because for me the narrow syntax is simply merge, so it surely doesn’t contain any categories or features (how can an operation store things after all?). Thus, a lot of the proposals from this year’s CamCoS (and from the literature more generally) are de facto attributing variation to the lexicon in my eyes, for that is the only storehouse in the human language faculty.
But of course that is just my personal opinion, which I’m sure many others disagree with. That’s why I think we must first agree on our definitions of “syntax” and “PF” as well as our fundamental assumptions of the architecture of grammar (e.g., which modules can store things and which are purely operational/algorithmic) before we can fruitfully discuss the syntax-or-PF question. I don’t think we should gloss over or stop thinking about these definitional issues even when we are all amazed by and focused on the multitudinous new data that have been discovered.
All in all, I’ve definitely enjoyed this year’s CamCoS. I just wish my Internet connection had been better so that I could have shown my face/voice and made new acquaintances, which is what we typically do at real conferences. For some reason Zoom hasn’t been very friendly to me whenever I have international meetings, especially when there are multiple participants and when the meeting is long. So, I should say thank-you to the chairs for reading out my questions. And I’m already looking forward to the next CamCoS! 😃