“Language... has created the word "loneliness" to express the pain of being alone. And it has created the word "solitude" to express the glory of being alone.”

Link for Research Paper Network

Supposed DP tree of Assamese (till now)...

No updates today.... feeling so lazy to write down the gist of todays class on (In)definites and Genericity!! Roughly todays class was about how is the notion of plurality concieved in natural language. In simple terms, plurality is a set denoting concept. If we have 2 objects a & b, the plural of a & b is concieved as {a, b, a+b}. But there are some deviations from this definition, for example: in Hindi

"main kitab kharidti hu", this sentence doesnot mean I buy only one book, though the head Noun is in singular form. Same kind of example is "maine bacche ko kitab di" (not necessary one book). But when we specify with a numeral like "maine do kitaben kharidi", then number morphology is reflected on the head. If we consider the first sentence, its interpretation is generic, its a general statement. Again lets look at some English sentences: "he is a house painter" or "he is stamp collector" it doesnot mean he paints only one house or collects only one stamp, but its not written as houses painter or so. These kinds of compound nouns also denote a more complex notion of plurality.

Another interesting example to look into is Mass nouns, it seems they are inherently plural (Inherent Plural Hypothesis, Chierchia 1998), they need some measure words to count them, like 3 sacks of rice or 6 glasses of milk. But we can have examples where mass nouns can directly be counted by a numeral like "three beers" or "four teas please". So many aspects to look into. Then Prof Dayal explained some theory of Atom, Taxonomy, Kind interpretation etc. which I'll write later on, have to read the handout and class notes again, carefully... 

Meanwhile, trying to draw the tree structure of Assamese DP, though it will change for sure. The ?P, I guess is the locus of specificity, have no idea about naming this projection. Lets see...

 

Merge and Feature Checking in Minimalism

Lets begin with some assumptions:

Lets say naltural language specifies a universl feature list [F] and each language selects a subset {F1……Fn}. These features are assembled into a Lexicon like,

[FH] for Hindi

[FB] for Bangla

[FA] for Assamese

Now lets make another assumption that there will some possible set of universal features which will be assembled onto each word in the lexicon and will be true for every lexical category. For example, the features of [±Tense] and [±Aspect] will be true for Verbs and features like [±Gender], [±Number] and [±Person] for Nouns. Those features which are assembled into the lexicon are called Interpretable feature.

Now take this example and represent each word with the letters below:

The cat came

 B     D    E

How will Operation Merge take place to derive the sentence structure? First, it will merge D and E to make C and then C will be merged with B and finally B and C will be merged into A. So Operation Merge is reducing syntactic objects into one unified tree.

Next question is what drive merge operation? Who is going to merge with what? X-bar theory says its Theta roles of the lexical category and principles of Theta criterion drives merge. But not everything can be explained with Theta criterion. Rather categorial features should trigger theta role selection. Some verbs like ‘feel’ seems to have an expeiencer and a theme/source. But the them/source can be of different syntactic categories. So theta role doesnot determine syntactic category nor syntactic category determines theta role.

Pat felt a tremor.

Pat felt uncomfortable.

Pat felt Chris had not performed well.

So ‘feel’ is averb and it has a [V] feature. It needs a Noun as its argument and Noun has a [N] feature. But we need to distinguish between being and needing. This difference is to be referred to as a difference in interpretability. Being a verb, feel has a interpretable [V] feature and needing a Noun it has an uninterpretable [N] feature. If a syntactic object has an uninterpretable feature, it must merge with a syntactic object that has a matching feature and once it is done the requirement is met. The uninterpretable feature is checked. Checking is done under sisterhood. An uninterpretable feature [F] on a syntctic object Y is checked when Y is sister to another syntactic object Z which bears a matching fetuare [F].

To distinguish interpretable feature from uninterpretable features, we will write uninterpretable feature with a ‘u’ in front of them. D has an uninterpretable feature [F]. E has an interpretable feature [F]. If we merge them the uninterpretable feature can be checked under sisterhood.

             C

   D                  E

V [uF]            [F]    (imagine a tree, line connecting D & E to C, dn't know how to draw them here)

Kick          me

[uN,V]     [N,acc,I,sg]

“Kick” is a verb (has an interpretable [V] feature) and c-selects a noun (has an uninterpretable [N] feature).

“Me” is a noun (has an interpretable [N] feature) and other features like Accusatice Case, first Person, Singular number.

The head is the needy one, the one that had the uninterpretable feature that was checked by merge.

After Watanabe, moving on to Bangla... started with The Nominal Left Periphery in Bangla and Asamiya ..... lets explore.....

Useful link: Symposium on Indian Morphology, Phonology & Language Engineering

Still confused regarding the main issue...

Now that I have a rough idea regarding how to go about this dissertation of mine (if at all I am able to write it... wish I took this break a year ago I could have submitted my thesis by this semester, at least could be eligible for extention!! now left with no choice, but to deregister). Only good thing is that within 10 years I can come back any time, register again and submit in one year. Anyways, no point thinking about it now. What I need to focus on is how all these linguistic analysis will be used for the computer wala thing? These DP internal movement and all can not be written in the form of production rules.... how will I write the grammar then? And the lexicon design part, I have to decide a binary feature set (like number, specificity, count, mass etc.) which will work like principles supposedly and apparently all the DP internal elements like CL, Dem, Num, Quantifier, Adjective, Noun etc. will have these features. Now there will be some parameters which will say under what conditions which feature will be true and how movement will happen and to where? for example:

moi   du-khon kitaap porhilu

I        2-CL       book    read

I read two books (means any two books)        whereas

moi    kitaap   du-khon   porhilu
I         book     2-CL         read

I read the two books (means two specific books already mentioned in discourse)

Suppose in a Eng-Assamese MT system or vice-versa, when we have to insert 'the' or CL to the noun will be predicted by the grammar.

Few more sentences to test Complex CL construction

1. Two bottles of wine were thrown into the soup
2. John drank up 4 bottles of wine
3. Three big bowls of soup
4. Three pound meat
5. Roger ate four big bowls of rice
6. I bought 4 card boxes of linguistics books into my office

For examples like 5, Hindi construction would be "Roger-ne chaar bare katori chaaval khaye" is grammatically correct and acceptable too. I guess I have rightly written the sentence like in TV cookery shows I have heard "chaar bare chammach haldhi powder" sort of things. In Assamese, this kind of construction is odd:

Roger-e          daangor     saari    baati      bhaat    khaale

Roger-Nom   big              4           bowl       rice       ate

The following is the right one with the insertion of Instrumental case marker and this is the strict order, no DP internal movement, rather scrambling of constituents (meaning Roger need not eat from 4 different bowls, it can be one big bowl, 4 times):

Roger-e          daangor    baati-re        saari    baati     bhaat   khaale

Roger-Nom   big              bowl-Instr   4          bowl      rice      ate

Its like, if we have to say in Hindi: "bare katori-se chaar katori chaaval"... can we say that?

Is similar construction possible in Hindi (without using 'baar' as in 'chaar baar' )?

Again when we say -- "two sacks of rice" either way its possible:
"do bori chaaval" or "chaaval ke do bori". But there is a difference in interpretation. The first one denotes some amount of rice whereas in the second one it refers to two sacks meant for filling with rice, not necessarily amount of rice. However, English translation and interpretation is same (amount of rice). Is it making sense?

Readings for DP in Japanese

Oh... finally done with Watanabe!!! So what is Watanabe trying to say? that Classifier is a parametric manifestation of Number morphology and in a syntactic representation, Classifiers and  Number morphology occupy same head position. Hence, Singular-Pural distinction in CL languages is just a matter of morphology and languages with CL can also have Singular-Pural distinction, like Japanese. Watanabe is saying CL is the head of #P because CL is manifestation of Number morphology and #P is the locus for that. So in CL languages, CL is the head of #P (if at all numeral is present) and number is in the Spec of #P and the argument he is giving for positing number in Spec of #P : modification in number like 'at least 3 books'. Then, # head udergoes agreement with Noun and move to Spec #P.
Then, in Section 4.1 on page 19, he talks about Complex CL constructions or Pseudopartitives like "2 boxes of books" and also that Complex CL and Massifiers are of same manifestation. However there is a difference in interpretation between the two if we take this assupmtion to be true of Assamese, for example: Assamese: two boxes of books

du-baakos kitaap
2-Box         book          (here it means 2 boxes full with books meaning amount)

kitaap-or     du-ta   baakos
book-Gen    2-CL    box         (here it means 2 boxes meant for keeping books)

But this kind of construction is not possible for count classifiers or true measure words:

tini-khon kitaap             * kitaap-or tini-khon    (3 books)

tini-litaar   gaakhir          * gaakhir-or tini-litaar (3 litter milk)
3-litre         milk

According to him, a CL and 'de/bun' with mass interpretation is in complementary distribution and occupies Spec #P position. But, how to describe the Assamese phenomenon in this light?

Again there are some points in his dissertation like only Chinese massifiers allow adjectival modification, Count CLs donot.  Supporting this argument, he says (on page 25), being functional heads ordinary classifiers resist adjectival modification. How? In Assamese, its not so.

There are such more questions... have mailed them to Ayesha, lets see what she has to say... so wait for Monday......

Passing Thought: came  across this sight http://gigapedia.com/. I guess books can be downloaded from it. Great!!

Useful reading: Universal Quantifiers

"Quantification, The Nominal Mapping Parameter, and DP Structure in Korean and Japanese" is a simplified version of Watanabe. I find this CaseP projection a bit dicy from argumentation point of view. Watanabe justifies its projection and phrasal movement to it Spec position to satisfy EPP (Extended Projection Principle) feature. Confused!!!

Findings till now in a nutshell....

Basic Assumption:

• Numeral in language either need Number morphology or Classifiers to mark countability on Nouns. Suppose languages with number morphology are called Number Language (NL) and languages with classifiers are called Classifier Language (CL)
• If Noun is a count noun then use of number morphology in NL and use of classifier (will combine with numeral or the noun) in CL.
• If Noun is a mass noun, then use of Measure words in NL and use of Massifiers in CL. They make mass nouns countable by creating a unit by which they can be counted.

Why Numerals need any of these two categories?

Since semantic denotation of the partitioning of the count nouns must be syntactically visible, Numerals require the presence of a syntactic marker. In some languages, number morphology does that, while other languages use classifiers. (Doetjes, 1996)

Count and Mass Noun Distinction:

• Count noun occur in both singular and plural form, Mass nouns do not mark number at all.
• Count nouns can be modified by a numeral, Mass nouns cannot be so modified (without a measure phrase)
• Count noun can co-occur with both articles, Mass nouns cannot co-occur with an indefinite article.
• Count nouns can co-occur with Qs like every and several, Mass nouns cannot co-occur with them.
• Count nouns cannot combine with expressions like 'too much', Mass nouns can combine with such expressions.

Apart from these differences Count and Mass noun differ in aspects like:

       - Cumulativity and Quantization

       - Semantic Partitioning of Nouns

Count nouns in CL are like mass nouns because they need a counting unit (classifier). So, Nouns in CLs generally have status analogous to mass nouns in Number Languages (NL).

Languages with [+arg, -pred] NP i.e., e-type languages, allow bare NP to appear as arguments. These bare NP arguments denote kinds, which are converted to mass properties for the purposes of quantification. Being mass, nouns in [+arg, -pred] languages resist plural morphology and require a classifier as the counting unit. (Chierchia, 1998)

Language Universals and Chomskyan Account

One of the main tasks of linguistic theory is to define what a possible human language is. In the pursuit of this end linguists seek to identify those facts, which are true of all human languages (attested or possible); generalizations that correctly describes all human languages are known as Language Universals. Any characterization of a possible human language needs to comply with such generalizations.

In this sense, the search for the universals is central to the enquiry of the nature of language. The quest for language universals should be restricted to those statements that tell something interesting about the nature of human language; but this begs a good many questions: among other things, the level of abstractions at which research should be conducted becomes an important issue. This particular issue has an important role in language universal studies. The question of explaining language universals has also contributed to the methodological and theoretical differences between the two main approaches of linguistics study: Chomskyan and Greenbergian. While Chomsky advocates the innate principles as the only comprehensive explanation, Greenbergian approach has an open mind on possible explanations considering, psychological and functional (discourse & pragmatics) factors. The Chomskyan approach to language universal is a particular description based on a theory specific meta language and constrained universal meta language as explanation. The idea can be traced to Roger Bacon's observation that all languages are built upon a common grammar, substantially the same in all languages, even though it may undergo in them accidental variations, and the 13th century speculative grammarians who, following Bacon, postulated universal rules underlying all grammars. The concept of a universal grammar or language was at the core of the 17th century projects for philosophical languages. Later linguists who have influenced this theory include Noam Chomsky, Edward Sapir and Richard Montague.

Chomskyan account:

The linguistic theory of Chomsky is based on one of the empirical facts that a subject knows linguistic facts without instruction or even without direct evidence. This predetermined knowledge is some "notion of structure", in the mind of the speaker, which guides the subject in acquiring a natural language of his own. His important hypothesis about Universal Grammar (UG) is a theory of linguistics postulating principles of grammar shared by all languages, thought to be innate to humans. It attempts to explain language acquisition in general, not describe specific languages. This theory does not claim that all human languages have the same grammar, or that all humans are "programmed" with a structure that underlies all surface expressions of human language. Rather, universal grammar proposes a set of rules that would explain how children acquire their language(s), or how they construct valid sentences of their language. In this context the quest for universal properties of language takes a very important character. It is the search for the universally valid principles of linguistic structure which innate UG makes available to the child at the onset of the acquisiton process.

In the context of Generative linguistics, a theory of linguistic structure should aim for explanatory adequacy (Chomsky, 1965) which incorporates an account of language universal and it attributes tacit knowledge of these universals to the child. Understood in this sense, implications for language universals are:

• They are part of internal language or I-langauge and cognitively represented.
• Since universals are part of I-langauge, they must be abstract.
• Linguistic universals will not be falsifiable by the actual diversity of languages and will be sufficiently rich and explicit to account for the rapidity and uniformity of language learning.
• The study of language universals is the study of the properties of any generative grammar for a natural language. Particular assumptions about language universals may pertain to either the syntactic, semantic, or the phonological component, or to interrelations among the three components.
• In Generative linguistics there is no sense to say that a statement is near universal. Universals are either part of UG or not. Therefore, universals must hold absolutely and be true of all languages.
• In Aspects of the Theory of Syntax (1965), Chomsky classifies language universals as formal or substantive. Substantive universals concern the vocabulary for the description of language. They generalize about language categories: 

           - Categories that must be present in each human language (e.g. vowel, noun)

           - Set of all possible categories available in any human language (e.g. ±voiced, ±uvular)

In his 1981account of linguistic theory, Chomsky states that absolute universals need not to be learned. Mastery of a language is not learning these principles. Rather, being equipped with UG (with its parameter) and exposed to language, the child constructs the grammar of his language. Our ability to speak is based on- the innate principles, parameters available in UG and the triggering experience of exposure to a specific langauge. When the parameters are fixed in one of the permitted ways, a particular grammar is determined: the ‘core’ grammar of that language. In a highly idealized picture of language acquisition UG can be taken to be a characterization of the child’s pre-language initial state. As the child grows up, he/she goes on learning new words through out their lives. In addition, we also learn certain less usual construction of the language. These exceptional or marked patterns of the language are not taken to be part of core grammar. They belong to the marked ‘periphery’ of the language and may be acquired later.

The next stage of development in Generative paradigm is the Minimalist Program (Chomsky, 1995). Here the main concern of theoretical linguistics is to devise a theory of grammar. According to generativists, a theory of grammar must satisfy four criteria. One obvious criterion is universality, in the sense that a theory of grammar must enable us to devise a descriptively adequate grammar for every natural language: in other words, the ultimate goal is to develop a theory of Universal Grammar. The other conditions to be fulfilled are explanatory adequacy, restrictability (the theory should be constrained so that it can be used only to describe natural language) and learnability (theory must provide grammar which are learnable by young children via relatively short period of time). A related requirement is that linguistic theory should provide grammar which makes use of minimal theoretical apparatus, in other words, grammar should be as simple as possible. In fact, Minimalist Program is motivated to minimize the complex structure and principles of 1980s syntax and the acquisition burden placed on the child and thereby maximize the learnability of natural language grammars. The program assumes that the langauge (I-language) has two components: a computational component and a lexicon. The computational system takes elements from the lexicon and form structural descriptions (SDs). The SD provides information about the properties of each linguistic expression, including its sound (PF) and meaning (LF). These two levels must satisfy three basic conditions of adequacy:

• It must be universal in the sense that any actual or potential human language or meaning of an expression is representable within it.
• It must be an interface, in that its elements have an interpretation in terms of sensory motor systems for PF and for LF, elements have an interpretation in terms of other systems of the mind/brain involved in thought.
• It must be uniform, in that its interpretation is uniform foe all languages, so as to capture all and the only properties of the system of languages as such.

Besides this the program also seeks an explanation for the uniformity and rapidity in the pattern of child language development. The answer is the claim of possessing a biologically endowed language faculty, the defining characteristics of which are the universals of human language, because, if we are making the assumption that the language faculty does not vary from one human being to another, those innately determined aspects of language faculty will also be universal. As it has been stated above that a theory of grammar should be restrictive. Universal principles should constrain the range of grammatical operations permitted in a language. One such universal is Structure Dependency Principle (all grammatical operations are sensitive to the grammatical structure of the sentences they apply to). These universals have important implications for the nature of the language acquisition process. This model vastly reduces the complexity of acquisition process. The learning task is further simplified since parameters are characterizable in a series of binary choices. The model also explains why children manage to learn the relative order (suppose, of head & complements) in a language in such a rapid and free fashion.

Explanation:

• Innateness is the only explanation for language universals having assumed that to speak a natural language we need:

              - Deep Structure (DS)

              - Intermediate structures

              - Surface Structure (SS)

              - Rules of transfer from DS to SS

• Innateness is justified as an explanation for universals on the ground that the evidence children have available through the language they hear around them is insufficient for them to develop a complex, abstract grammar, which underlies any language.
• Rules are subject to complex and abstract constraints and having observed that children acquire any language (as L1) easily it seems to be a necessary conclusion that crucial abstract principles of Transformational-Generative Grammar are innate.

These innate principles are neutral with respect to differences among languages, i.e. they are universal. That is, language universals are innate ideas that facilitate L1 learning.

My first contribution to Wikipedia!!!

Classifiers and Number Morphology

Wikipedia people are very generous I guess... considered my not so significant piece of writing to be posted on their site! Infact, was just trying to see how to upload a page in Wikipedia and that time had this paper only (a section of the paper). Though, a long way to go to take the shape of a well written article. Still struggling with the Japanese dissertation and data!!! This PhD seems to be a distant dream now... even if I have to deregister... at least will not have the regret that I didn't try.....