The preceding chapters have traced the development of the two first-appearing and most frequent NPIs in Dutch child language, the modal verb hoeven (have to/need) and the temporal adverb meer (anymore). It was found that children from the onset are sensitive to the limitations on the distribution of these expressions and use them restricted- ly, in a way which is essentially correct. With increasing age, this early present basic restriction is built out in tandem with developing knowledge of ways to express negative meaning.
The investigation covered a wide age range, and natural as well as experimental data were considered, from the first occurrences of NPIs in two-year-olds' spontaneous speech to teenagers' judgments about what constitutes a correct licensing environment. These data were analyzed in light of the issue introduced in Chapter 1: how can children be expected to acquire the restricted distribution of NPIs without receiving systematic evidence about what is not grammatical? Overgeneralizations should be prevented at all cost, since these constitute the type of error for which negative evidence seems indispensable in order to be unlearned. In theory, an error-free progression could be conceived of, to wit by conservative widening of the laws of licensing proposed in Zwarts (1995). It was predicted that if this strategy were employed in the acquisition of NPIs, children would start out with a very narrowly defined licensing rule. Later on, there would be a steady expansion of licensing environments in the order from strong to weaker negations.
The data about the onset of NPIs, discussed in Chapter 2, showed that children indeed start out using hoeven and meer with a highly restricted distribution, in the immediate company of the negation niet (not). Utterances at a somewhat later age, however, discussed in Chapter 3, made it clear that a strategy of conservative widening in the acquisition of NPIs is not tenable after all. Besides many correct utterances licensed by classical negation, the children also produced utterances lacking a correct licenser. One part of these utterances had a negative meaning although there was no negation marker present within the sentence, while the other part had an affirmative meaning. These observations appeared to represent a discrepancy: on the one hand, young children use NPIs correctly and in a highly restricted manner, while on the other hand, there are examples of incorrect usage in which NPIs are overextended to non-licensing environments. Moreover, there appeared to be an internal contradiction in the two categories of ungrammatical utterances, in terms of their opposed meanings.
An experiment was then carried out to test three incompatible hypotheses about the spontaneous speech data. These hypotheses were formulated with specific reference to hoeven. For young children
In this experiment, which after piloting in different age groups was carried out with three year old children as subjects, a new, composite task of reproduction and acting out was used to elicit indirect grammaticality judgments and meaning assignments of sentences containing hoeven in both licensing and non-licensing environments. The experimental results provided no support for any of the three incompatible hypotheses, but pointed in the direction of the fourth, alternative one. The children in general did not accept hoeven in non-licensing environments, as was evident from their modifications to these test sentences in the reproduction component of the task. Also, they showed that their knowledge of licensing went beyond the combination of hoeven with niet (not), since test sentences with hoeven in other licensing environments tended to be accepted.
In the light of these results, the spontaneous speech data were reanalyzed in Chapter 4, in order to give more substance to the alternative `licensing in development' hypothesis. It was here that the various patterns observed in spontaneous speech fell into place, revealing a steady developmental progression, well grounded in a principle which from the onset is essentially correct.
The early supremacy of classical negation as a licenser and the ungrammatical utterances with negative intent but no correct negation marker could be directly linked to early patterns in the acquisition of negation. Classical negation, together with the primitive negation markers that accompanied children's ungrammatical utterances with NPIs, are precisely the constructions which make up children's early vocabulary for expressing negative meaning. The fact that children employ this complete negation vocabulary as NPI environments, even though this sometimes may lead them beyond the adult model (dubbed `pseudo-licensing'), led to the conclusion that children apply their own principled restriction in the acquisition of NPI licensing, steering the course of their developing knowledge of ways to express negative meaning. The postulation of negation as the forerunner for NPI environments also provided an explanation for the observation that the subsequent order in which the wider variety of correct licensers appears is so well in accordance with predictions following from the principle of conservative widening. An investigation of the emergence of negative markers in the children's speech showed that expansion of the negation vocabulary in general proceeds in the order from strong to weaker expressions, an order which is subsequently reflected in the expansion of licensing environments for NPIs.
The other spontaneous speech pattern, the apparently unlicensed utterances with an affirmative meaning, were reanalyzed as another example of pseudo-licensing, as they could also be traced back to a principally correct restriction. It was found that these utterances were either interrogative or expressed a contrastively affirmative meaning, characteristics which closely touch on two possible ways of NPI use in adult grammar, contingent on speaker assumptions and expectations: licensing by questions and by intensifying affirmation. A yet later example of pseudo- licensing was found in utterances with the modal particle nog (still), which alludes to a negative meaning by means of the pragmatic implicatures it induces.
In Chapter 5, it was shown that, with the distribution of NPIs rooted in a principled restriction, children are in an advantageous position for acquiring the exact adult rules for NPI use, even though they may be led on a detour via pseudo-licensing. As children from the onset are aware of the fact that NPIs cannot appear in all configurations in which they according to their membership of a certain syntactic category should be allowed to appear, these expressions are no invisible holes in grammar, and the most important part of the task of acquiring NPIs is already accomplished. From this point onwards, children are not dependent on negative evidence to unlearn overgeneralizations. The early pseudo-licensers - headshaking, anaphoric negation, negative intonation contour, and underarticulation of negation - will disappear when these primitive negation markers die out and are replaced by more mature negative expressions. The later pseudo-licensers - questions, intensifying affirmation, and nog (still) - need not be unlearned at all, but have to be fine-tuned until they are in complete accordance with the finesses that characterize the use of these constructions as licensers in adult discourse. This adjustment process was illustrated by the results of a grammaticality judgment experiment with 7- to 14-year-olds, which showed how the acceptance of pseudo-licensers gradually declined with increasing age.
To sum up with respect to the strategy of conservative widening: this study indicates that the acquisition of NPIs indeed proceeds in the order from narrow to broader generalizations. At the same time, however, the phenomenon of pseudo-licensing has shown that this widening does not occur in a strictly conservative manner: children sometimes use NPIs in a way for which the adult model provides no positive evidence. In all cases, such deviant usage could be associated with negative polarity, which indicates that NPIs from the onset in children's speech have a principled restricted distribution, rooted in knowledge of the essence in NPI licensing: sensitivity to negative polarity.
It thus appears that children, from early on, are sensitive to what NPIs are sensitive to. This does not guarantee a faultless usage of NPIs, though. Acquisition of the range of licensers for NPIs is not an instantaneous event but a prolonged process in which the early present sensitivity to negative polarity has to mature and unfold. As a consequence, the restricted distribution of NPIs in child speech only gradually approaches the licensing system in adult discourse.
Up
to general Table of Contents
Back to previous chapter: `Adjustment of
Restrictions'