So I’m attending this year’s Cambridge Comparative Syntax conference (CamCoS 9), not as a speaker or student helper this time but only as an audience member. I’ve been participating in CamCoS for a few years now (basically since my first year of PhD) and can report from my personal experience that they are absolutely stunning. This year’s CamCoS is a special occasion for two reasons. First, it is organized in honor of Prof. Anders Holmberg’s retirement. Second, it is held entirely online (on Zoom) due to COVID-19. This is the second online conference I’ve attended in 2020 (the last one was ACT2020). This time the organizers (i.e., my dear old colleagues) have very considerately taken time differences into account, so I’m not forced to stay up till dawn like last time.
I decided to compile my personal notes from the conference into blog posts just like I had done last time for ACT2020, because I find blogging a helpful way to organize my thoughts. And if anyone else finds these posts interesting then that’s a bonus. Usually what happens when I go to a conference is that there’s an awful lot of exciting new information, but things become so exhausting near the end that much of the newly acquired knowledge ends up being forgotten alongside the handouts. These notes are not meant to be read as professional reviews. Rather, they mix what I have learned from the conference and my personal feelings, not all of which are directly related to the talks.
There are altogether five talks on the first day, which are split into two sessions (separated by a 20-min coffee break) and respectively feature Classical Chinese, Japanese/Tagalog, Russian, Czech/Polish/Arabic, and Niger-Congo languages (especially Burkinabé Bambara, aka Dioula) as their main objects of study. Of course, each talk also assumes a more general typological perspective. I think that’s one of the most attractive aspects of CamCoS, in that it really puts crosslinguistic variation in the spotlight and encourages participants to look beyond our linguistic comfort zones. For instance, if it weren’t for the conference I would probably never expose myself to Burkinabé Bambara, but now I not only know it’s spoken in western Burkina Faso but can also say a few things in it, such as I ka mun dumu? ‘What did you eat?’ and N’ ka sise dumu ‘I ate chicken’.
This year’s CamCoS has an overarching theme of “syntax vs. PF” (which, by the way, is a long-standing puzzle in the field); namely, whether parametric variation in human language is best accounted for in (lexico)syntactic or (morpho)phonological terms. Handouts and recordings of all talks can be found on the conference website, which is fairly decently designed by the way.
The first talk, by Edith Aldridge and entitled “Syntactic parameter change in Early Middle Chinese,” is about a salient grammatical change from Old to Modern Chinese. Aldridge’s main proposal is that sentential subjects are licensed at the CP-level (i.e., Spec-CP) in Late Archaic Chinese (LAC, 5th–3th century B.C.E.) but at the TP-level (i.e., Spec-TP) from Early Middle Chinese (EMC, beginning in 1st century B.C.E.) on. CP (complementizer phrase; i.e., the discourse part of a sentence) and TP (tense phrase; i.e., the propositional part of a sentence) are terms used in generative grammar (especially in the Chomsky school). There are a number of observable consequences following from this change, and Aldridge mainly discusses two of them: (i) the CP specifier position is freed up in EMC for object movement in main clauses; (ii) the CP domain (aka the left periphery) itself becomes more elaborate in EMC, with a number of topic and focus positions being added. The overall effect is that we see more scrambling-like patterns in EMC, as in (1) (taken from Aldridge’s handout).
(1) [Early Middle Chinese]
bǎi shén suǒ shí shèng rén wèi dāng yǔ rén děng
“What the various gods eat, saints believe (it) should be the same as for people.”
In this EMC sentence the phrase “what the various gods eat” has been fronted from its logical position (i.e., the position of the parenthesized “it”). The English translation sounds a bit odd, but it’s perfectly natural in Chinese (and presumably also in other languages where scrambling is freely allowed, such as Japanese). That modern syntactic theories can be used to study Classical Chinese interests me a lot, since I have only studied Classical Chinese on a pretheoretical, school-grammar level. Even though I am quite comfortable with Classical Chinese texts, I have had little idea as to what CP or TP corresponds to in it even long after I had stepped into linguistics. So, at least from a pedagogical perspective I am hugely impressed by Aldridge’s work. I think the modern, generative syntactic perspective can actually help pupils learn Classical Chinese more efficiently, because many complex and not-entirely-clear school-grammar rules (especially those concerning function words and syntactic constructions), which used to be (and still are) taught and learned by rote, turn out to be components of a fully logical system.
The second talk, by Kenyon Branan and entitled “Regimes of ordering,” tackles the syntax-or-PF question directly, and the speaker’s answer is that linguistic variation is located in phonology rather than syntax (I use both terms in a broad sense to avoid jargon). The particular phenomenon Branan looks into is scrambling, which I have already mentioned above. In my opinion, scrambling is just a fancy way of saying flexible word order, as if some parts of a sentence, especially its subject and object(s), can “scramble” out of their original positions to other position. Familiar scrambling languages include German and Russian. For example, in German one can say either (2a) or (2b) depending on the context (Wikipedia).
a. …dass der Mann der Frau die Bohnen gab.
“…that the man gave the woman the beans.”
b. …dass die Bohnen der Mann der Frau gab.
“…that the beans, the man gave the woman.”
Branan observes that in Japanese and Tagalog, two other typical scrambling languages, the scrambling operation creates different scope effects, which partly contradict the Lexical Correspondence Axiom (LCA) put forward in Kayne’s (1994) seminal monograph The Antisymmetry of Syntax. Thus, it is okay to put an object in front of a subject that contains a possessive pronoun coreferential with the object in Japanese, as in (3), but not okay to do so in Tagalog, as in (4). Both examples are taken from Branan’s slides, with slight adjustments.
[O mittu-izyoo-no kaisya-o] [S soko-no syain-ga] tyoosasita ✅
“More than three companies, their employees have investigated (them).”
Nag-ma-mahal [O ng bawat anaki] [S ang kanyangi ama.] ❎
“Every child, its father loves (it).”
Branan explains away the contradiction between scrambling and LCA by proposing two PF restrictions, respectively called “exit” and “reentry,” that work in tandem with the LCA and essentially exempt certain phrases from it. Since I’m not familiar with Tagalog grammar (though I do want to learn a bit of it), I don’t have much to say about the particular proposal of the talk. From a general theoretical perspective, however, I feel that the syntax-plus-PF approach is on the right track. Language is a highly complex natural phenomenon, so it’s nothing surprising if it turns out that both syntax and PF play a role in regulating parametric variation. I personally favor such an eclectic approach because if the whole universe isn’t black-and-white, why should language or the human mind be?
The third talk, by Elena Titov and entitled “Syntactic parameters: The case of IS-conditioned argument reordering,” also focuses on the flexible word order of verbal arguments (such as subjects and objects), so I guess it falls under the same empirical domain as the second talk (which may or may not be the reason why they have been put on the same day). The “IS” in the title of the talk is short for “information structure,” which I think is just a fancy (and not necessarily self-evident) way of saying discourse-related grammatical categories like focus and topic. Titov’s view on the syntax-or-PF issue is strongly pro-PF. In particular, she proposes that what lies behind the scrambling effects in languages like Russian is (partly) the notion of interpretive prominence, which essentially creates a non-syntactically-determined asymmetry among the various verbal arguments, whence their word order flexibility. For example, the sentence in (5), taken from Titov’s slides, is only acceptable as an answer to “Who kissed Ivan?” but not to “What did Maria do to Ivan?”.
Ivana pocelovala Marija.
“Ivan, Maria kissed (him).”
Titov’s explanation for this is that when the question being asked is “What did Maria do to Ivan?” both Maria and Ivan are discourse-prominent (in that they are both mentioned in the question), and in this situation the word order isn’t freely alterable, so one could only answer “Maria kissed Ivan” (with an emphasis on “kissed”), if she did so. While I find the analysis largely coherent for the Russian data, I don’t think it provides strong evidence for a pro-PF (or more exactly antilexicosyntax) view on parametric variation. In particular, I don’t think the currently mainstream practice of encoding variation in functional categories equals encoding it in the (narrow) syntax, which is basically just merge in the current Chomskyan view, because functional categories are functional categories and merge is merge—the two cannot be mixed up. As such, I personally don’t think Titov’s argument that feature-based accounts of parametric variation are problematic in view of evolutionary biology (i.e., the view that the syntactic component of the human language faculty has remained unchanged since its evolutionary appearance) is a real argument. The narrow syntax (aka merge) doesn’t change (much as 1+1=2 doesn’t change under normal circumstances), but functional categories and formal features surely can.
During the 20-min break there was an awkward period of silence, and the organizers decided that they’d play some music in the following days. The talks resumed after the break.
The fourth talk (also the first “speed talk”), by Ivona Kučerová and Adam Szczegielniak and entitled “Syntactic variation meets PF uniformity: Underspecification of nominal functional categories,” addresses the interesting phenomenon of structural homophony; that is, homophony across grammatical categories. For example, the element “k” appears in gender, number, and case inflection in a number of Slavic languages (e.g., Czech, Polish), as in (6), and the feminine element appears in categories like division/number and nominalization in Arabic, as in (7).
a. category-changing -k
doplnit ‘to complement’ → dopln-ěk ‘complement’
b. gender-changing -k
diplomat ‘diplomat (masc)’ → diplomat-ka ‘briefcase (fem)’
(7) [Levantine Arabic]
a. individuating feminine element
tabšuur ‘chalk’ → tabšuur-a ‘a piece of chalk (fem)’
b. group-forming feminine element
mtdyn ‘believer’ → mtdyn-i ‘religious group (fem)’
The main proposal in this talk is that both the Slavic -k and the Arabic feminine element are realizations of the same underspecified functional head, which the authors dub as $i^*$ (loosely inspired by Wood & Marantz 2017). This head operates on syntactic features and toggle their values (e.g., from $-$ to $+$).
I find the idea in this talk exciting, but alongside the excitement also come a number of potential technical pitfalls (as is typically the case with highly creative ideas). For example, the almighty $i^* $ needs to operate on some feature(s), but how exactly those features are derivationally selected remains unclear (nor does it seem completely deterministic as the speaker points out in her response to my question in the Q&A period). In addition, since $i^* $ is defined as a polarity-toggling head, the proposal in the talk necessarily relies on a particular version of feature theory, where all relevant features have boolean values (otherwise the polarity toggling won’t work). So what might the theory say about nonboolean features (e.g., features in the [att: val] format) also remains unclear. The above being said, I do think the idea in this talk is worth exploring further.
The fifth and final talk of the first day, by Elena Callegari and entitled “Focus markers in an optionally wh-in-situ language: The case of Burkinabé Bambara,” examines focus marking in several Manding (Niger-Congo) languages, partly in comparison with Italian. Unlike the first four talks, this talk is more empirical than theoretical. It presents a wide range of crosslinguistic data and carefully demonstrates the subtle interpretative differences with regard to the several focus markers. While I’m a total layman in African languages, I do find this talk fairly easy to follow, partly because the slides are colorful and have large fonts. The main takeaway of the talk is that there is considerable crosslinguistic variation as to what type of focus matters for syntax. I think this partly converges with one of Aldridge’s conclusions; namely, the left periphery of human language syntax isn’t universally fixed (pace the cartographic theory).
In sum, four out of the five talks on the first day (i.e., except the fourth talk) are concerned with the discourse domain of human language syntax, especially with topic and focus categories. I personally haven’t studied this domain in detail but might start taking a deeper look into it thanks to all the interesting data I’ve seen today, especially those in Classical Chinese. I look forward to the upcoming days of CamCoS 9. 😃