Parametric theory of word order, language acquisition and historical change

Oct 21, 2011 by

In the previous posting, I outlined an alternative theory that treats the order of major sentential constituents — subject, object and verb — not as a primitive, but as a result of several binary factors (called “parameters”) that determine pair-wise order of various elements (e.g. of the verb and the object, or of the verb and the subject, etc.). As pointed out in that previous posting, such a theory can handle better the relative frequencies of various word orders in the world’s languages. Moreover, this theory can also provide a better explanation for two other phenomena: the acquisition of word order by small children and the evolution of word order (the very topic of Gell-Mann & Ruhlen’s work). Here, I will discuss these two issues in more detail.

Let’s start with children acquiring their native language. Numerous studies show that children who are just beginning to combine words together into sentential utterances do not get the word order of their target language correctly. If, following G-M&R we assume that the order of major constituents is a primitive, we should expect children to get all aspects of word order correctly at the same time! In other words, children should go from the initial incorrect word orders to the correct word order expected in their target language in one jump. But this is far from truth. In fact, children acquire different aspects of their target language word order gradually, one by one.

Interestingly, the parametric theory, outlined in the previous posting, can be used quite effectively to account not only for word order variation and frequencies, but also for the way children acquire aspects of word order. Recall the three parameters that we relied on to account for the VSO order: Head Directionality Parameter, Verb Raising Parameter and Subject Placement Parameter.

Experimental studies with small children show that they learn these different aspects of their target language syntax at different points in their linguistic development. The Head Directionality Parameter — which makes English or French different from, say, Japanese or Basque — is acquired relatively early, at the tender age of 18-22 months. Already at the two-word stage, English toddlers request Give cookie! and their Japanese peers — (the Japanese counterpart of) ‘Cookie give!’…

The Verb Raising Parameter — which determines the position of a (tensed) verb with respect to certain adverbs and negation markers and hence accounts for the contrast between the French order Jean mange souvent le chocolat (literally ‘John eats often the chocolate’) and its English counterpart John often eats chocolate) — comes next. According to studies by Déprez & Pierce and others, children who are about 21 months old exhibit evidence of knowing that the tensed verb appears in the position before the negation marker pas in French but after the negation marker not in English. Thus, English children of that age say things like Not have coffee, while their French peers place the tensed verb before pas, as in Me plaît pas monsieur là (literally ‘me pleases not man there’), Veux pas lolo (literally ‘want not milk’). At the same time French children know to place the infinitive verb after pas, as is appropriate for adult French; thus, they say things like Pas casser (literally ‘not to-break’), Pas attraper papillon (literally ‘not to-catch butterfly’).

Finally, these studies also show that children acquire the word order patterns controlled by the Subject Placement Parameter later, at about 24 months. Thus, even though English and French children figure out the proper placement of the verb with respect to the negation markers not/pas at an earlier age, it takes them an addition three months to realize (not consciously, of course!) that subjects in their target languages must come before the tensed verbs (or auxiliaries) and negation markers. Thus, smaller children, who have not yet reached this stage, place the subject after rather than before the negation marker, with English children (less than 24 months old) saying No I see truck and No Leila have a turn. French toddlers who are less than 2 years old too still place the subject after rather than before the tensed verb, as in Tombe Victor (literally ‘falls Victor’), and even after adverbs, as in Veut encore Adrien du pain (literally ‘wants more Adrien bread’). This latter utterance is also in contradiction to Tom Givon’s claims that SOV word order comes most naturally to humans; here, the child produces a VSO rather than SOV utterance!

The non-simultaneous acquisition of the Verb Raising and Subject Placement Parameters is revealed in particular by the following utterance from a 20-month old French child (already mentioned above): Me plaît pas monsieur là (literally ‘me pleases not man there’). Notice that here the verb is already placed appropriately before the negation marker pas, while the subject is not yet in its proper pre-verbal position.

To recap, it appears that children do not learn the order of the subject, object and verb all at once but rather deal with different aspects of sentential word order separately, as would be expected from the parametric theory.

Similarly, distinguishing discrete parameters responsible for ordering various major sentential elements (subject, object, verb, adverbs, negation, etc.) with respect to each other in a pair-wise fashion provides an explanation for the theory — advocated by G-M&R — that only certain types of word order changes are possible. Indeed, if the order of the three major constituents (subject, object and verb) is taken to be a primitive parameter, with six possible settings, nothing in G-M&R’s theory explains why certain changes are possible and others are not. In other words, G-M&R provide a list of possible word order changes, based on typological and historical observations, but they do not tell us why these and not other changes are possible.

Note that the more frequently encountered types of word order changes correspond also to the more frequently encountered types of mixed word order languages (see Table 1 in G-M&R’s paper): SOV/SVO, SVO/VSO, VSO/VOS, SVO/VOS and SOV/OVS. According to G-M&R’s data, 46 languages with mixed word order patterns (37% of the total mixed word order languages) have both SOV and SVO (according to WALS figures, these numbers are 29 languages, 43% of all mixed word order languages). The next most common combination is either the combination of SVO and VSO (according to G-M&R, 24 languages or 19%; according to WALS, 13 languages or 19%) or the combination of VSO and VOS (according to G-M&R, 17 languages or 14%; according to WALS, 14 languages or 21%). Next comes the combination of SVO and VOS (according to G-M&R, 11 languages or 9%; according to WALS, 8 languages or 12%), followed by the combination of SOV and OVS (according to G-M&R, 9 languages or 7%; according to WALS, 3 languages or 4%). All other mixed combinations are found much more rarely, and according to G-M&R “may be due in part to errors in analysis of these languages”. The correlation between the most common patterns of change and the most common types of mixed word order languages can be explained by the often-made assumption that languages with a mixed word order pattern are really languages in the midst of (gradual) word order change. Here’s how G-M&R describe the process of switching from SOV to SVO:

“It should be obvious that a language cannot change its basic word order overnight. What is required is a long gradual process during which it is the frequencies of different word orders that change. A language may begin with a high frequency of SOV and a low frequency of SVO. As the language changes, the frequency of SVO may increase at the expense of SOV until there emerges a stage referred to as ‘free word order,’ in which the frequencies of both orders are similar. A final stage may occur when the frequency of SVO becomes high and that of SOV low.”

By why are these specific mixes and these specific types of change possible (or more frequent)? Here again the parametric theory can provide an answer: these commonly found types of word order changes correspond to switching the setting of just one parameter! For example, by switching the setting of the Head Directionality Parameter an SOV language is converted into an SVO language (this is what happened in the history of English). Similarly, a switch in the setting of the Subject Placement Parameter turns an SVO language into a VSO one (or vice versa).

Two additional changes — from SVO to VOS and from SOV to OVS — can be accounted for by another parameter, the so-called Subject Side Parameter, which determines whether the subject appears to the left of the verb phrase or to the right of it. The verb phrase itself can be either head-initial, as in the case of both SVO and VOS orders, or head-final, as in the case of SOV and OVS orders. In other words, it is the Subject Side Parameter parameter which distinguishes the head-initial SVO language like English from an also head-initial VOS language like Malagasy:

English: The children buy the rice for the baby.
Malagasy: Mividy vary hoan’ ny azakely ny ankizy (literally ‘buy rice for the baby the children’)

(The switch from VSO to VOS or vice versa involves a switch in the setting of a parameter that would be too complicated to explain here.)

Languages with a mixed word order (or “no dominant word order”, as WALS formulates it) are, according to this theory, indeterminate with respect to a certain parameter, with both settings being possible. Gradually, one or the other of the parameter settings “wins over” the other and the historical change is complete.

This brings me to the last point I would like to make, which concerns G-M&R insistence on separating historical change due to external influences (essentially, grammatical borrowing from other languages) from historical change due to what they call “natural drift”. Thus, they insist that change into SOV is possible only as a result of external influences and never as a result of the “natural drift”. However, it is not clear at all how this distinction is supposed to work. After all, grammatical borrowing is as subconscious as “change by natural drift” (and unlike lexical borrowing, which can be fully intentional). After all, speakers of a language do not get together and decide to switch to the word order used by their neighbors (why would they?!). Whether the change is triggered by external influences (from another language) or by language-internal, structural factors (i.e. by “natural drift”), the change procedes in stages of gradually increasing frequency of the “new” word order and the corresponding decline in the frequency of the “old” word order. So why would diffusion-induced and natural-drift-induced changes procede according to different paths? No explanation is apparent from G-M&R’s work…

Related Posts

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below: