W1 Tuesday: Interfaces and Typologies

I missed writing yesterday’s post, and spent some of today out doing some admin (transporting items from home, getting shorn, etc.).

Two main texts I reviewed in detail, the first being Reuland and Meulen’s introductory chapter (pp.1-20) to The Representation of (In)definiteness.

My interest in the book was simply that I’d not seen noun phrases examined at this level before (basically at the level where the syntax and semantics interface, where the close examination could tell us more about the semantics than the historical discussion, valuable and irreplaceable as that is). Potentially, a direction for my research will be to look at at noun gradeability, and understanding an aspect of noun phrases even for something not directly related to gradeability seemed like a useful exercise when I picked up this book some time ago.

The prior reference point I had for the idea of (in)definiteness was Russell’s theory of definite descriptions, that referring expressions he would call ‘definite descriptions’ rather than indefinite ones are only evaluable if they refer to a real entity, i.e. the reason why saying ‘the King of France’ is strange, since there is currently no such King…

In English, at least, the distinction between definite expressions and indefinite ones is captured by the alternation of ‘a’ and ‘the’, for a broad class of nouns. On a more rigorous analysis, however, the main observation is that there are clear situations where either the definite or indefinite is strongly preferred, for instance in a sentence like, ‘There ensued a/the* riot on Massachusetts Ave.’ They refer to such situations as exhibiting a definiteness effect (DE).

Interestingly, even across languages, R&M note that where DEs are observed, some semantic conditions are observed quite consistently. Likewise, some syntactic conditions also seem to be observable, e.g. the correlation between an expletive subject and an indefinite argument elsewhere in the clause.

R&M proceed to discuss three approaches to analyzing definiteness (Milsark 1977; Barwise and Cooper 1981; and finally Heim 1982). Heim’s analysis I found particularly interesting. R&M characterize Heim’s as a ‘dynamic’ approach (in contrast to Milsark’s and B&C’s static ones) in that the interpretation of an NP is analyzed as a process.

The second text from this book I dug into was David Gil’s chapter (pp.254-269), titled ‘Definiteness, Noun Phrase Configurationality, and the Count-Mass Distinction’.

Gil’s interesting result is that there are at least two clearly distinct and definable NP typologies. His thesis is that ‘NP typology is a joint product of the two covarying parameters of configurationality and the count-mass distinction.’ He notes seven typological correlates, each of which either a Type A language has while the Type B language doesn’t, or vice-versa. His examples contrasted mainly English and Japanese.

To illustrate the count-mass example, he observes that in Type A languages usually a large class of nouns e.g. ‘count nouns’ are obligatorily marked for number, whereas in Type B languages there is either no marking (Japanese, Chinese), it is optional, or it has a special interpretation or occurs in specifically constrained contexts (Turkish, Indonesian). He shows how the parameters that determine this also influence other issues like whether (in)definiteness is obligatorily marked.

The other key term in his thesis, ‘configurationality’, refers to whether noun phrases have a hierarchical structure internally. Whereas Japanese does not, English does. Correspondingly, in English it is the determiner system within which NPs are integrated, whereas Japanese cannot truly be said to have determiners. An example that illustrates this difference is how the English phrase ‘Sam’s three books that Cyril read’ can only occur in this configuration as a noun phrase, whereas in Japanese each nominal or admonimal in the phrase can grammatically appear in any order.

Another interesting piece of data Gil introduces is of adnominal distributive numerals in Japanese (sansatsuzutsu meaning potentially ‘three each’ or ‘three at a time’ or ‘in threes’ – all actually subtly different in English). Gil argues that in the phrase sansatsuzutsu no hon the adnominal distributes over the nominal head, but basically Type A languages cannot have adnominal distributive numerals because numerals are determiners and not really adjectival or nominal modifiers.

I’ve not had much experience with language typology either, but I found Gil’s analysis illuminating. It raises the question of how  or whether I might do a similar analysis for Singlish in my project. I’m interested to find out my supervisor’s opinion about this analysis as well.


The Polynesian /ko/, and a Reflection on Syntactic Categories

Most recently, I’ve been studying Hovdhaugen et al.‘s Handbook of the Tokelauan Language for the preparation of a grammar sketch. My interpretation of the morphemes in the subsequent example from the text are based on their analysis. In particular, when I use the abbreviations ‘VP’ and ‘NP’ below, I refer to their explanatory framework, wherein all sentences in Tokelau may be divided into VPs and NPs only, and whether a phrase is a VP or NP depends on the central word. Each VP and NP has slots in which grammatical particles (prepositions, etc.) can appear.

I ended up looking at one particular example, which involved three variants of a sentence each having the meaning, `The boys built the house,’ in some detail.

  1. The first variant had the phrase-order: [VP Na fau] [NP2 te fale] [NP1 e na tama] .
  2. The second variant had the phrase-order: [NP1 Ko na tama] [VP na ki latou faua] [NP2 te fale] .
  3. The third variant had the phrase-order: [NP1a Ko na tama] [VP na fau] [NP1b e ki latou] [NP2 te fale] .

Some of the elements that appear in the above examples, interpreted:

  • NP2, /te fale/, might be glossed as ‘the house’.
  • The main element of NP1, /na tama/, might be glossed as ‘the boys’. ‘na’ here is analyzed by Hovdhaugen et al. as a plural article.
  • The main element of VP, /na fau/, includes the morpheme /fau/ for ‘build’. ‘na’ in this case is analyzed as a past-tense marker (distinct from the article). The form ‘faua’ we see in the second sentence is necessitated by the inclusion of the pronoun /ki latou/ in the verb phrase.
  • The words /ki latou/ are the third-person plural pronoun (‘them’). It appears in the VP of the second sentence, and its own NP in the third sentence.
  • The word /e/ which we see in the first sentence in NP1 and the third sentence in NP1b was interpreted as a preposition-of-sorts meaning ‘by’. In the above examples, it communicates that ‘na tama’ are they agents.

The element which I did not cover in the above list is /ko/. In Hovdhaugen’s grammar, /ko/ is described as occurring in sentences where an NP is the initial phrase. Ergo, we do not see /ko/ in the first sentence, in which the VP is the initial phrase, but it appears in the second and third sentences.

Regarding the significance of /ko/, one account that the authors present is that between a sentence with a /ko/-initial NP in the sentence-initial and a similar sentence with the VP in the sentence-initial position, the /ko/-initial sentence has the discursive function of giving information about the subject who participated in event x, whereas the VP-initial sentence describes event x, wherein the ‘subject’ was a participant.

One interpretation of this difference is that /ko/ has the function of indicating the subject of the sentence. Relatedly (but not necessarily), one might say that /ko/ marks nominative case.

But the difficulty is precisely in characterizing its function. The category of ‘subject’ itself is difficult to specify, and it is not clear whether the system of markings in Tokelau should indeed be characterized as a system of cases.

The /ko/ morpheme itself is part of several Polynesian languages, and as such has been treated in the academic literature before. Among these descriptions of Polynesian languages, /ko/ has been interpreted as a predicate marker, as well as a ‘copular preposition’. The notion of ‘copular preposition’ is potentially resolvable with Hovdhaugen et al.’s grammar, in that they propose a broadly defined ‘Preposition’ slot in noun phrases. The idea that /ko/ marks a predicate is potentially more difficult to resolve, in that Hovdhaugen et al. analyze it an element of a noun phrase specifically, whereas ‘predication’ itself presupposes a concept of predicate and argument.

Moving away from a discussion of abstract categories for the moment, in the context of the example sentences, one might imagine the effect of ‘ko’-involving constructions to be commensurate with the effect of constructing English sentences beginning with, ‘It is…’ (Credit to my Prof. for this discussion point.) For example, third sentence above (‘Ko na tama na fau en ki latou te fale’) might be constructed as, ‘It is the boys who built the house,’ or even, ‘It is the boys, they built the house.’ (As compared to the default, ‘The boys built the house.’)

The ‘it is’ construction in English itself seems to me to reflect aspects of both the copular preposition and predicate marker proposals, in that they both serve to isolate some function or aspect of the subject, which thereby allows it to be foregrounded in the sentence. (In fact, I think it is fair to say that both the quality of the ‘it’ and the ‘is’ in this construction are recognized as exceptional in some way; we call this particular ‘it’ the expletive ‘it’, and ‘is’ in English what we have traditionally called a copula, and is in any case not a typical ‘verb’.)

On a more fundamental level, I see the concept of predication and the function of a copula as operating on that troublesome boundary between noun-iness and verb-iness.

Earlier today I also happened to attend a senior’s comps talk on ambiguity and syntactic reanalysis. A brief illustrative example of both concepts is with the sentence, ‘The old man the boat.’ While it seems wildly ungrammatical on an initial reading, there is a grammatical interpretation (which I assume you have come to).

One theory of what causes that initial cognitive difficulty is that we construct the syntactic relations between the elements as we read the sentence, and upon encountering a difficulty in continuing the construction, we have to reanalyze the sentence. It was, however, a later point in the talk that seemed relevant to the problem of characterizing the function of /ko/. The presenter quoted a study wherein sentences like, ‘While John hunted the deer ran into the woods,’ were shown to subjects, after which the subjects were asked if John hunted the deer. The result was that many subjects answered in the affirmative. (In contrast, almost all of my friends in the linguistics department at the comps talk answered in the negative; I think there is a simple explanation for this discrepancy, which is that we’d been prompted to process things syntactically by having considered earlier ambiguous examples in the talk.)

making. That most respondents put ‘hunt’ and ‘deer’ together was thoroughly unsurprising, and that this superseded a syntactic judgment which (I believe) they were capable of making suggests that syntax is more superficial than fundamental. For instance, the categories of ‘noun’ and ‘verb’ are treated in our syntax as so clearly distinct that we may decisively call ‘man’ in ‘The old man the boat,’ a verb. By the same theory, we may conceive of the syntactically correct way to put ‘hunt’ and ‘deer’ together.

However, would we be able to make the same claim as decisively about the syntax of another language? Like the copula ‘is’ and the expletive ‘it’ in English, the Polynesian /ko/ seems to occur at that troublesome noun/verb interface, insofar as it seems potentially to mark a subject, suggest an agent (in Theta theory, any verb may be analyzed as having a certain configuration of nominal arguments, for example in how the verb ‘nag’ suggests a nagger and someone who is nagged), or even assign nominative case. Potentially none of these accounts are coherent, if our categories ‘noun’ and ‘verb’ need to be constructed differently.

Predication in an Alien Language

I’m doing ‘Field Methods in Linguistics’ this term, in which we’re working on understanding an unfamiliar language, through direct interviews with our consultant (a native speaker of the language). For our first short paper, we had to propose a basic analysis of the sentence structure. I ended up coming up with an initial theory that I’m quite happy with, and I thought I’d write about how I came to it.

In a nutshell, my theory is that every sentence in the language has an obligatory predicate-marker before the predicate proper. As such, the form of a simple sentence would be something like:

(Subject) – pred – Predicate

By my hypothesis, the predicate marker cannot be omitted, and is always present in some form. The predicate itself may either be a verb, or an adjective (in the sense of ‘is red’). While the predicate marker might be thought of as functioning like ‘be’ in some ways, it doesn’t carry tense; in the language we’re studying it is the verb which realizes the tense/aspect, although the predicate marker is involved in various kinds of agreement. Another interesting feature is that by my hypothesis, it is never absent, so it is akin to having an ‘affirmative’ declaration with every predicate, except, of course, when the sentence expresses something other than ‘affirmative’ (e.g. negative, question, etc.).

But my main motivation for writing this post was to recount some of the ideas and theories that guided my thought-process. What I hypothesized to be a predicate-marker is the so-called [w-]-morpheme that we’ve been discussing in class.

  1. In one of the first few classes, a classmate mentioned a language where both affirmative and negative conditions are marked. (In contrast, in English, we only distinguish ‘waste’ and ‘ do not waste’, without having to say ‘do waste’.)
  2. My lab partner had observed that negation in a sentence patterned with the appearance of the [m-] morpheme and the omission of the [w-] morpheme, in a way that was reminiscent of ‘do’-support in English (e.g. ‘He says.’ vs. ‘He does not say.’).
  3. Some data about how to form questions was also discussed in class, and it was noticed that the [w-] morpheme was also absent.
  4. In a later class, a classmate raised the example of the sentences for ‘I melted the butter.’  and ‘The butter was melted by me.’ I noticed that it was possible to omit both the subject and the object (‘I’/’me’ and ‘butter’), which left only the predicate for ‘melt’ (in its presumably conjugated form) with a preceding morpheme.

I had also observed a few things from the data.

  1. There is a long form of the [w-] morpheme and a shorter form. I noticed that the longer form did not appear with adjectives and verbs with no direct object.
  2. I also noticed that the subject could be dropped if it was either a pronoun, or if the predicate was an adjective. With the butter example, it appeared to me that the subject was omitted in the spoken sentence, but that it was semantically present, in that it seems clearly discernible from the spoken context.
  3. I noticed that the negative marker and the question marker patterned together.

Fragments from these observations and discussions about some points of syntax conspired to bring the notion of predication to mind, and brought me back to my formal logic and computer logic lessons from last term. I was also fortunate to have been looking at adjectives in lab sessions, in that the predication of properties (‘is red’, etc.) was on my mind. After observing how subjects could be dropped, and how question-marker and negation patterned together vis-a-vis the [w-]-marker, I was led to the hypothesis described above.

All in all, this was good fun. On the one hand, the computational and logical paradigm made the patterning comprehensible, but on the other hand more organic questions about register (formality and informality) and semantics (negation is so tricky) also played a part.

Words and Theories

Good starting points from LING 115 class on 25 September 2013.

I. What’s in a word?

  • Sound: Phonetic (and Phonological)
  • Meaning: Semantic
  • Category: Syntactic

I.e. formal linguistics in a nut-shell.

II. What would we expect from a ‘theory of syntax’?

1. Descriptive: If my theory accurately represents my mental representation.

2. Explanatory: If my theory gives a valid analysis of my mental representation.

(Following 2., I can expect to have a hypothesis about how my mental representation was constructed.)

3. Generative: If my theory reliably (or, perhaps, just mostly) generates representations that correspond to my mental representation.