On tomlin, and manning and parker on basic word order

On tomlin, and manning and parker on basic word order

Iurqzuage Sciences, Volume 13, Number 1, pp. 89-97, Print& in Great Britain 1991 0388-oOW91 $3.00+.00 Reqamon Press plc On Tomlin, and Manning and ...

555KB Sizes 2 Downloads 30 Views

Iurqzuage Sciences, Volume 13, Number 1, pp. 89-97, Print& in Great Britain

1991

0388-oOW91 $3.00+.00 Reqamon Press plc

On Tomlin, and Manning and Parker on Basic Word Order

JaeJung

Song

Nfftimal University of Singapom

ABSTRACT fn this article, two recent word order studies, i.e. Tomlin I&sic Word Order: Funttionai Principles, London: Groom Helm (1986)1, and Manning and Parker U_unguuge Sriences II,

43-65 f 1989)1, will be examined. Some specific and genera1 problems with these studies will be identified. Firstly, in Torniin’s study, the obse~ed predmninance of S in&I languages is unaccounted for; to remedy this flaw. an afternative differential weighting system is proposed. Secondly, it is argued that Manning and Parker’s (1989) figure/ground interpretation of basic word order is seriously is both irrelevant

problematic

and circular.

in that the support that they provide for the interpretation

In conclusion,

non-linguistic

evidence

in the word order study

is called for.

INTRODUCTION In this note I will examine the two recent word order studies, i.e. Tamlin (f!?%), and Manning and Parker (1989), and will further identify some problems, both specific and general, with these studies. Firstly, I will argue that, in Tomlin, the realization of the ftmctionai principles is calculated in such a way that a most significant thing about the frequencies of the basic word orders - the sheer predominance of S initial languages - is simply obscured. I will demonstrate that the flaw can be easily eliminated under the alternative differential weighting system. Secondly, I will argue that it is doubtful that Manning and Parker can account for the relative frequencies of the basic word orders (and the predominance of S initial languages) when they appeal to the figure-ground hierarchy: their support for the hierarchy with respect to the basic word orders is highly problematic, Finally, in the guise of a conclusion I will make a brief note of the fundamental problem of the word order study in general: lack of independent, non-linguistic evidence for the various ~c~onal principles that both Tomlii and Manning and Parker promote, &eit in an ingenious manner.

90

Language Sciences, Volume 13, Number 1 (1991)

TOMLIN

(1986)

In his data-driven

research on basic word order, Tomlin (T) first establishes

basis of his sample of 402 languages orders (expressed as percentages):

the following

frequencies

on the

of the basic word

sov

svo

vso

vos

ovs

osv

44.78

41.79

9.20

2.99

1.24

0.00

However, T thinks that there is no statistical significance between the SOV and SVO percentages and also between the VOS and OVS percentages. Hence the following pattern of relative frequencies:

sov = svo > vso > vos = ovs > osv T further proposes three ‘functional principles’ to account for the frequencies of the six basic word orders. These principles are: (a) the Theme First Principle (TFP), (b) the Animated First Principle (AFP), and (c) the Verb-Object Bonding Principle (VOB). The first two principles state that theme and agent NPs come before non-theme and non-agent NPs in the basic transitive clause pattern. And the last principle is intended to capture the closer bondedness that exists between the verb and the object NP in comparison with the subject NP. T uses these principles to account for the distribution of the basic word orders found in his sample. The basic idea of T’s explanation (1986: 126) is that the more of the three principles that are realized in it, the more frequent across the world’s languages a given word order will be. And the patterns of realization for the six basic word orders can be presented as in Table 1. TABLE 1

sov svo vso vos ovs osv

44.78% 41.79% 9.20% 2.99% 1.24% 0.00%

TFP

AFP

VOB

+ + + -

+ + + -

+ + + + -

On Tom&, and Manning and Parker on Basic Word Order

91

The plus sign represents realization of the principle in question and the minus sign represents non-realization of it. As the reader can easily see, there is a strong correlation between the frequencies of the basic word orders and the realization of the principles. By calculating the total number of realizations of the principles, T seems to be successful in accounting for the relative frequencies of the basic word orders. However, one most significant thing goes unaccounted for in T’s work. That is, why is it the case that SOV and SVO account for 86.57% of T’s sample in comparison with the other word orders that account for the remaining 13.43% of the sample? Further, the 86.57% is almost evenly divided between SOV and SVO. Although he is concerned with the relative frequencies of the basic word orders, nowhere in his book does T really address the issue of the sheer predominance of the SOV and SVO word orders. It is not clear at all how the three principles can be put to use in accounting for this skewed distribution across the world’s languages. Surely there is a huge difference in the combined frequency of SOV and SVO on the one hand, and that of the remaining word orders on the other, i. e . a substantial difference of 73.14 96? There is no doubt that there is a skewed distribution in favor of S initial languages. T seems to treat the principles as equivalent, since, for instance, he calculates the total value of the realizations of the principles for SOV and OVS at 3 and 1 respectively. In fact, he states (1986: 122): The TFP is realized if it is the case that the S precedes

the 0. The TFP is not realized if the

0 precedes the S. The AFP is also realized if the S precedes precedes

the 0. It is not realized if the 0

the S.

This means that, under T’s view, the S has the value of 1 (each for the TFP and the AFP), as long as it is placed before the 0. Hence it does not matter whether the V precedes or follows the S. The relative position of the S with respect to the V does not have any bearing on the value assigned to the S. It is the relative position of the S opposed to the 0 that counts in T’s calculation scheme. Although this way he can account for the relative frequency of each and every word order in comparison with one another, T cannot explain the substantial difference that exists in the frequencies of the S initial word orders and of the rest. I wish to argue that the three principles should not be regarded as carrying the identical and/or consistent value or weight in (numerically) calculating realization of the principles for each word order. Therefore, I suggest that the TFP and the AFP be given much higher value than the VOB, when the S precedes both 0 and V, and that they be given the same value as the VOB, when the S precedes only the 0. To put it in another way, the TFP and the AFP are optimally realized if the S is placed initially in the basic clause pattern, preceding both 0 and V. In this way, one can explain away the skewed distribution in favor of the S initial word orders, something that T fails to capture and explain with his own ~nctional principles. The following

92

Language Sciences,Volume 13, Number 1 (1991)

table is the re-working of Table 1 along the lines suggested above (the numerical values given in T (1986) are provided in parentheses for the sake of comparison). Arbitrary as it may be, both TFP and AFP will be given the value of 2 instead of the usual value of 1, only if the S is placed initially in the clause pattern. Otherwise, the realization of the TFP or the AFP will be assigned the value of 1. The realization of the VOB will always carry the value of 1.

TABLE TFP

2 AFP

VOB

Total 543) 5(3) 2(2) l(1) l(l)

sov

2(l)

2(l)

l(l)

svo

2(l) l(1) -

2(l) l(1)

l(1) l(1) l(1) -

vso vos ovs osv

-

Now the total values calculated for all the basic word orders under differential weighting system more realistically represent their actual To sum up, the TFP and the AFP seem to be in need of modification that they are viewed as being optimally realized when the S is placed other major constituents of the basic clause pattern, i.e. the 0 and the

O(O)

the alternative frequencies. to the extent before all the V. In fact the

principles of T’s are formulated in such a way as to support the conclusion that I have reached in this note. Unlike the TFP and the AFP, the VOB necessarily involves two constituents, i.e. the V and the 0. The order of juxtaposition is of no significance. On the other hand, the TFP and the AFP both involve the S and its relative position as well. The position of the S within the basic clause pattern is of utmost importance. Hence the S should be treated as opposed to both 0 and V, not just to the 0, as done in T’s work. Why are the TFP and the AFP optimally realized, if the S precedes both 0 and V? More generally, how does T justify the functional principles that he invokes to account for the relative frequencies of the basic word orders? Unfortunately, not successfully. Although he provides linguistic evidence for the tendency of the theme and animated to appear early in the clause, it is uncertain in his work: (a) why S is correlated with animacy and theme; (b) why there exists closer bondedness between the verb and the object NP; and (c) why these correlations are somehow natural or unmarked. I will defer further discussion till the next section; this kind of uncertainty is not unique to T’s work, as will be shown below.

On Tondin, and Manning and Parker on Basic Word Order

MANNING

AND PARKER

93

(1989)

Unlike T’s work, the work of Manning (M) and Parker (P) is built on some of the previous studies on word order: their main aim is to find some underlying cognitive principle that can account for the relative frequencies of the basic word orders established in the work of other researchers. M and P do not allow for the statistical non-significance that T recognizes between SOV and SVO, and also between VOS and OVS; they accept the following frequency hierarchy:

sov > svo > vso > vos > ovs > osv So in this respect M and P differ from T. In fact, they have not at all considered T’s work, in which by far the largest sample in word order typology is used. M and P’s approach to basic word order is a hybrid of Peircean semiotics and Gestalt psychology. In the spirit of Peirce, they (Manning and Parker 1989:52) believe that the basic word order is a semantic diagram: it is a kind of diagrammatic representation of semantic form. Drawing on Gestalt psychology, i.e. figure/ground interpretation, they (Manning and Parker 1989:52) further point out that this semantic form is subject to the same interpretive process as visual form: some order - figure vs ground - is imposed on the diagram during the interpretive process. They (1989:55) adopt the semantic diagram of the basic constituents as in Fig. 1.

Figure 1. According to M and P (1989:55-6), there are two perceptual steps in determining figure and ground: (a) to determine the figure for the entire diagram, i.e. overall figure; (b) to determine the figure for the two remaining regions, i.e. relative figure. The first step is regarded as more important than the second, since it determines the figure for the entire diagram. Further, there are two principles of figure/ground

94

Language sciences,Volume 13, Number l(1991)

interpretation: (a) the smaller, enclosed area will be favored as the figure; and (b) the larger the enclosed area is, the easier it is to perceive the enclosing area as the figure. When it conforms

to the principle

of (a), an interpretation

is called a preferred inter-

pretation, whereas, when it conforms to the principle of(b), an interpretation is called a possible interpretation. Further, when it conforms to neither of the principles, a disfavored interpretation is obtained. When this kind of procedure is followed, the most favored interpretation of Fig. 1 is such that the S is determined as the overall figure and the 0 is the relative figure in the overall OV background. For SOV languages, this is precisely the interpretation carried out: in the first step SOV receives a preferred rating; and also in the second step SOV receives a preferred rating. When the procedure is applied to the other word orders, the following table of comparison of the basic word orders can be produced. TABLE Figure/ground interpretation S before S before V before V before (0 before (0 before

(0 before (V before (S before (0 before V) before S) before

V) 0) 0) S) S V

3

Step(1)

Step(2)

overall figure

relative

Preferred Preferred Possible Possible Disfavored Disfavored

Preferred Possible Preferred Possible Preferred Possible

figure

As the reader can easily verify, Table 3 perfectly correlates with the word order frequency hierarchy that M and P assume in their study. l Although M and P are not really concerned with it, Table 3 may be construed as, albeit implicitly, indicating the predominance of S initial languages; only SOV and SVO receive a preferred rating in the first step; and none of the remaining logical possibilities receive such a preferred rating in the first step. Further, it may seem to account for the relative frequencies of the basic word orders. However, I will argue that, since the validity of the semantic diagram is called into question, their figure-ground hierarchy does not account for: (a) the relative frequencies of the six possible permutations of S, 0, and V; (b) the skewed distribution in favor of S initial languages (i.e. over 7646% of the languages are S initial); and (c) the more or less even distribution of SOV and SVO languages. (Recall that (b) and (c) are exactly the same questions that T also fails to look into.) What is it that allows one to choose the diagram over other possible alternative diagrams, e.g. the V enclosed in the 0, and further the V and the 0 enclosed in the S? In other words, what is the basis of their construction of the diagram? M and P

On Tomlin, and Manning and Parker on Basic Word Order

(1989:58-60)

take pains in drumming

95

up ‘several sources of support’ for considering

the diagram: e.g. differential generality of reference of S, 0, and V; the V as the most independent constituent of the three; ergativity, etc. However, there are serious problems with this kind of analysis. First, M and P’s sources do not offer the expected support. For instance, they argue that since it is the most structurally independent of all, and thus attracts agreement markers of the S and/or the 0 Le. head-marking in the sense of Nichols (1986)1, the V is the overall enclosing region in the semantic diagram in Fig. 1. If they are right, then, one expects SOV languages (representing the most favored interpretation of the diagram) to be head-marking languages. But, in fact, it is the case that V initial languages tend to be head-marking languages, i.e. the V attracting agreement markers of the S and/or the 0 (Nichols 1986). The point is that the agreement marker attached to the V in V initial languages is used for the sake of communicative efficacy, i.e. processing motivation, and that in V initial languages this processing motivation is the strongest (for detail, see Nichols 1986: 8 l-2). The agreement marker has nothing to do with the alleged highest degree of structural independence of the V. This is indeed singly problematic, since the diagram is the very basis of the calculation of the ratings for the six logical word order possibilities. Further, when the whole account provided by M and P hinges on the diagram, one expects much more than support for it, i.e. solid evidence; in fact, their sources of support cannot be even anything close to evidence for their considering the diagram in Fig. 1. Secondly, the kind of support on which M and P draw is but circular. In order to argue for the reality of the diagram, they employ language internal arguments (or what they call sources of support): since linguistically the V is the most independent basic constituent of the three, it must be the overall enclosing domain. M and P are here using as support the same phenomenon that they have set out to explain. What they need to provide for their account is not shaky, circular pieces of support, but non-linguistic evidence. All the more so, since they (1989:60) seem to claim that their ‘explanation’ is psychologically real (cf. their Gestalt approach), and since they (1989:45-9) themselves criticize the previous studies on the same grounds, i.e. circularity, lack of evidence, etc. In sum, the semantic diagram on which M and P’s account rests is given, not argued for. This point leads to the fundamental problem

with the basic word order study in general:

lack of non-linguistic

evidence.

CONCLUSION

Both T, and M and P are mainly concerned with the relative frequencies of the six basic word orders, but not with the predominance of S initial languages. However, it is a significant fact about basic word order that the majority of the world’s languages are S initial; there is a considerable difference between S initial languages and non-S initial languages. This must be accounted for by any theory on basic word order. In

96

Language !Menees, Voiunie 13, Number lQ9!31)

neither of the two studies, it seems, has this been {sa~fa~~~ly) Further,

T, and M and P appeal to such cognitive

processes

discussed. as animacy,

theme,

perceptual preference, etc. in an ingenious and interesting way. But are they successful to the extent that they can account for the relative frequencies of the basic word orders, and the predominance of S initial languages? Initially, they seem to be so (albeit with some modification especially in the case of T). However, all they provide to justify their accounts is language internal arguments for the cognitive processes that they identify, For instance, why should theme and animacy be associated with S? Or why should perceptual preference be such that S is the innermost of the three in the semantic diagram in the first place? One is left in the lurch about why these factors motivate all these correlations. Obviously, i~d~~ndent~ non-linguistic evidence is called for. Although I admit, along with Blake (1988:216), that this is hard to come by, I think that the word order study has come to a point where serious endeavors must be made to seek such independent, non-linguistic evidence. One hopes that word order studies to appear in this decade will be concerned with this very problem.

NOTES 1.

Recall that T does allow for statistical non-signi~&an~e between SGV and SVO, and also between VUS and GVS. If T’s frequency result turns out to be right, M and P’s ~gure-ground theory of basic word order will have to call for some additional overriding principle(s).

REFERENCES Blake, Barry J. “Review of Russell S. Tomlin, Basic Word Order: Functional 1988 JiX.@r& of ~~~~j~t~~s 24. 2 13-17. Greenberg,

Principles,”

Joseph H. “Some Universals of Grammar With Particular Reference to the Order of meaningful Elements, ’ ’ in Utaiversals ~~~~g~ag~, pp. 73-113, Joseph H. Greenberg (ed.), Cambridge, MA: MIT Press. Mallinson, Graham and Barry J_ Blake 198 1 Language Typology: Cross-linguistic Studies in Syntax, Amsterdam: North-Holland. Manning, Alan D. and Frank Parker “The SOV> . . . >OSV Frequency Hierarchy,” Language Sciences Il. 1989 43-65. 1966

On Tondin, and Mmniqg and Parker on Bask Word Order

Nichols, 1986 Tomlin, 1986

97

Johanna “Head-marking

and

Dependent-marking

Grammar,”

Language

56-119. Russell S. Basic Word Order: Functional

Principles,

London:

Croom Helm.

62.