COGNITIVE EXPLANATIONS OF BASIC WORD-ORDER VARIATIONS
Nanchang University, China
This paper proposes that many major word order phenomena can be reduced to effects of basic cognitive principles. First, what is called ‘word order’ should be titled ‘Chunk Order.’ By ‘chunk’ we refer to the unit that is sensitive to the span of short-term memory and attention, i.e., the number 7 ( 2) (Miller 1956). If we take the lexical head and all its dependents (arguments, adjuncts and all modifiers) as the word-order unit, it turns out that the total number of such units in a construction can never be greater than about 7 (Lu 2001).
Furthermore, many major phenomena of word-order can be reduced to the result of the interplay of two fundamental iconicity principles, each being semantic and pragmatic in nature.
This conclusion is based first on the following parallelism between clauses and NPs with respect to the cross-linguistic variation of basic word-order. Specifically, in both clauses and NPs, while the basic orders of head-final structures are relatively consistent, those of head-initial structures are rather diverse, as illustrated below (‘Head’ here means only the lexical head).
(1) a. Orders of S, O, V in clauses:
If V-final, the basic order is almost always SOV.
If V-initial, both VOS and VSO are likely to be basic orders.
b. Orders of D(eterminer), A(djective) and N in NPs:
If N-final, the basic order is always DAN.
If N-initial, both NDA and NAD are likely to be basic orders.
Related to this parallelism is the following observation.
(2) When the two dependents are separated by the head:
a. The clause basic order is overwhelmingly SVO.
b. The NP basic order is always DNA.
Tomlin (1986) explains the ordering in (1a) with the combination of the following three principles: A. The theme first principle. B. The verb-object bounding principle. C. The animated first principle. He claims, ‘those constituent orders will be more frequent which permit more of the principles to be realized.’
This paper extends Tomlin’s explanation of (1a) to (1b) mainly via the following two revisions:
First, we regard Tomlin’s Principle B as an instance of Behaghel’s First Law (Haiman 1983: 782; see also the ‘proximity principle’ of Givón 1991, and ‘relevance criterion’ of Bybee 1985). In the current study we restate this principle as the Semantic Proximity Principle (SPP), which says that the closer a dependent is to the head semantically, the closer it stands to the head structurally. The reason for this restatement is that the head is the defining factor of the word-order unit and the primary reference point of word-order.
Second, we merge Tomlin’s principles A and C, based on the observation that definiteness and animacy constitute a natural class (Comrie 1979), which we call ‘identifiability’ (Lambrecht 1994). The mergence of principles A and C results in a more general principle, named the ‘Identifiabilty Precedence Principle’ (IPP), which states that the more identifiable a unit is, the stronger its tendency is to precede.
The interplay of IPP and SPP is enough to cover the clausal word ordering shown in (1a). Tomlin’s three-principle approach may further explain the difference of the numbers between VOS and VSO languages. However, Dryer (1997) argues that this difference is unreliable and insignificant.
Numerous studies have already shown that SPP is also responsible for the NP internal order (e.g. Hetzron 1978, Sproat and Shih 1991 about adjective ordering within NPs, Rijkhoff 1990, 1998, 2002). However, few efforts have been made to apply IPP in NP internal ordering. To this end, we only need to take a naturally extended version of IPP (EIPP), which states that the more identifiability a modifier contributes to its mother NP, the stronger its tendency is to precede.
Given that the hierarchy of identifiability contribution to their mother NPs by D, Q (numeral) and A in as in that order, the interplay of EIPP and SPP explains (1b) and the following observations (Hawkins 1983, Rijkhoff 2002) as well: 1.When any or all of the D, Q and A precede N, they (i.e., those that do precede) are always found in that order. 2. For those that follow, no predictions are made. 3. In no case does A precede N when D or Q follows.
The current paper provides a detailed description of Chinese NPs, which further confirms EIPP.
IPP explains as well the following three word-order universals: Greenberg (1966)’s universal 5 (if a language has SOV as the dominant order and the genitive follows the governing noun, then the adjective likewise follows the noun), universal 25 (if the pronominal object follows the verb, the nominal object does so likewise), and Ultan’s universal of genitive positioning (when the pronominal genitive follows the governing noun, then the nominal genitive likewise follows the noun as well) (Ultan 1978: 24). These universals in essence are correlated to the hierarchy of preference to precede. In each case, a unit less likely to follow (i.e. more identifiable) follows necessarily implicates the unit that is more likely to follow follows, if everything else being equal.
The interplay of SPP and EIPP can explain other word order phenomena, such as ordering of V, O(bject) and manner Adv(erbial). Cross-linguistically, when both O and Adv following V, the basic order is overwhelmingly V O Adv, as illustrated in English. Given that the nominal expression O is more identifiable than non-nominal Adv, and that O is closer to V than Adv is, the phenomenon can be explained readily.
In fact, SPP and/or IPP can explain majority of word-order universals posed by Greenberg.
Comrie, Bernard. 1979. Definite and animate direct objects: a natural class. Linguistica Silesiana 3: 13-21.
Dryer, Mathew S. 1997. On the six-way word order typology. Studies in Language 21:69-103.
Greenberg, J.H. 1966. Some universals of grammar with particular reference to the order of meaningful elements. Universal of grammar. ed. by J.H. Greenberg, second edition, 73-113. Cambridge, Mass: MIT Press.
Hawkins, J. A. 1983. Word order universals. Academic Press.
Hetzron, Robert. 1978. On the relative order of adjectives. In Hansjakob Seiler ed., Language Universals: 165-184. Tübingen: Gunter Narr Verlag.
Lambrecht, Knud. 1994. Information Structure and Sentence form. Cambridge University Press.
Lu, Bingfu 1998. Left-right asymmetries of word order variation: a functional explanation. University of Southern California Ph.D. diss.
Lu, Bingfu 2001. What is the chunk in linguistic structure? Proceedings of the Third International Conference on Cognitive Science. Beijing, China.
Miller, G. 1956a. The magical number seven, plus or minus two. Psychological Review No.63.
Miller, G. 1956b. Human memory and the storage of information. I.R.E. Transaction on Information Theory Vol.IT-2, No.3.
Rijkhoff, Jan. 2002. The Noun Phrase, Oxford University Press.
Tomlin, Russell S. 1986. Basic word order: functional principle. London: Croon Helm.