Standard grammaticalization theory assumes unidirectionality. The process whereby a lexical item is coopted into a grammatical pattern, becoming bleached, reduced, and eventually cliticized, is deemed virtually irreversible. If grammaticalization continues to act upon the gram in question, eventually it will be reduced to nothingness.
This thesis distinguishes between the process of grammaticalization and its results. Under the process of gramaticalization linguistic units change from:
(1) relatively free to relatively bound,
(2) more motivated to less motivated,
(3) optional to obligatory,
(4) lexical to grammatical.
The process of grammaticalization is the transition from a less `grammatical' state to one that is more so. But the results of the process, if we follow the history of any given linguistic unit as it undergoes grammaticalization over and over again, may involve a return to a state similar to one that was in effect at an earlier stage of its history.
Unidirectionality of change is not inconsistent with cyclicity of result. If we set out on a trip on this planet and consistently maintain the same heading, we will eventually return to our starting point. If we ask a computer to continuously increase a byte by one, starting with `00', after it reaches `FF', it will return to `00'. This dissertation does not seek to negate the unidirectionality of change: it merely clarifies what we can or cannot infer from the fact of unidirectionality. We cannot infer, for instance, that because a unit presents as isolating at one stage in the language, it is less grammaticalized than one that presents as affixed: the independent unit might have undergone more grammaticalization than the affix.
The particular phenomenon explored in this dissertation is the grammaticalization of pronouns into copulas and copulas into pronouns.
A crosslinguistic examination of copulas in ten languages, Chinese, English, Finnish, French, Hebrew, Hungarian, Korean, Russian, Turkish, and Vietnamese, reveals the following cognitive tendencies:
(1) one morpheme is used for possession and existence,
and
(2) another morpheme is used for identity and class membership.
Of the ten languages, only English did not fit into the above pattern, using is for existence, identity and class membership, but have for for possession.
Significantly, in nine of the languages, identity and class membership constructions expressed the two participants (the token and the class, or the two items whose identity is being asserted) as grammatically coequal. That is, in languages with morphological case, they were both in the nominative. In languages without overt case, examination of sentence pattern reveals that neither participant was subordinated to the other. (The exception was French, whose object pronouns were used in equative sentences: while colloquial English displays the same pattern, for purposes of the study, standardized dialects were used.)
It is from these equative copulas that third person pronouns are developed, and by the same token it is from subject pronouns that we get newly formed copulas.
In this dissertation the following instances of grammaticalization of pronoun and copula are demonstrated:
(1) Chinese (Classical to Modern) pronoun to copula
(2) Hebrew (Biblical to Modern) pronoun to copula
(3) Finnish (pre-attested to Modern) pronoun to copula
(4) Turkish (pre-attested to modern) copula to pronoun
(5) Hebrew (pre-proto-Semitic to Biblical) copula to pronoun
The Hebrew example, if we combine (2) and (5) above, provides us with a full cycle, from copula to pronoun to copula.
A pronoun is no less grammatical than a copula, nor is a copula less grammatical than a pronoun. How, then can a transformation from one to the other be deemed an example of grammaticalization?
The key point is that in judging whether something is an example of grammaticalization we must look to the process, not the results. At each point along the continuum of change, the unit in question was becoming more opaque, dependent, bleached and grammatical. In the case of Hebrew, the path followed went like this:
xaja/xawa ---> haja/hawa (--->) howo --------> hu ---> huThe changes as they are diagramed above are each an example of grammaticalization, because they are examples of the following processes:A B C
`To live' --> `to be'(e.g.->) `his being'----->`he' --->`be.present'
A. Semantic and morphological bleaching
B. The change from an independent lexical form with grammatical morphology, to a member of a closed set of grammatical morphemes (deictics)
C. The change from a member of a closed set of grammatical morphemes to membership in a paradigm for the conjugation of a highly bleached, syntactially privileged verb.
At each step of the way, the linguistic unit underwent a further degree of abstraction. In step A, the semantic bleaching changed from a concrete description to an abstract relation, in step B, from a fully parsable derived form to a member of a deictic pronoun set, and in step C from a deictic, referring to an entity, to a member of a verbal paradigm referring to the most bleached relation expressed in the language.
From this we can infer that while grammaticalization progresses along a unidirectional cline from concrete to abstract, the history of a particular linguistic unit may reveal that it has travelled the same path more than once.
The implications for grammaticalization theory from the above discussion are as follows:
(a) an independent linguistic unit is not necessarily less grammaticalized than one that is dependent,
and
(b) an isolating language has not necessarily undergone fewer cycles of grammaticalization than one that is synthetic.
The process of grammaticalization is a constant force that drives language onward, and it can account for all change that moves from iconicity to formalism. The study of grammaticalization can encompass anything from nonce formation to paradigm building to syntactic change to typological evolution. The same processes that we can see taking place over a few stretches of discourse can also be traced back as operating over thousands of years. But the linguistic resources upon which the process operates are limited. This calls for a certain amount of recycling. Most cycles are so long and convoluted that returning to the point of departure in any recognizable shape is entirely out of the question. But for those items that are already highly grammaticalized, the radius of the cycle is small, and so the closing of the circle is recognizable, even over a considerable stretch of time.
© Aya Katz