Canonical word order is acquired early, and children tend to produce adverbials and topicalised elements after the core sentence (Clancy 1985, 420f), i.e. in the salient sentence position in which second language learners also tend to produce all information that is not part of the canonical word order (see ch. 1. 4.2). Sentence-final particles, which also stand at this salient position, are acquired as early as logically possible, i.e. at the two-word stage (Clancy 1985, 381).
Case particles have obviously no case-marking function in early speech. Clancy assumes that particle choice depends on the noun position in the sentence (e.g. 'ga' for the first noun in sentence; 1985, 390) or on semantic notions such as agency or animacy, or that particles are associated with particular verbs (Clancy 1985, 391). The fact that most errors occur where functional and grammatical roles do not map (as in constructions with receiving and giving verbs and passives) suggests that semantic notions of agency have a strong influence on particle choice. Clancy explains the fact that particles are produced at all by suggesting that they might function as phrase boundary markers. This is supported by the observation that case particles are not produced earlier than the two-word, or two-phrase, stage, i.e. when word delimitation is necessary.
Verb morphology is acquired earlier in JFL than in EFL (relative to MLU; Clancy 1985, 426); some inflections occur before MLU reaches 2.0 (Clancy 1985, 381). This might be due to the fact that Japanese is a pro-drop language, which results in MLU's being shorter because there is no subject. In early JSL learner language, on the other hand, there is a strong tendency to provide the hearer with all information pertinent to the topic, i.e. the subject is almost always produced (Kawaguchi 1996). Inflection for past tense and negation occurs earlier than aspect marking. The first type of aspect marking is V-te irti; other forms with full verbs as auxiliaries (e.g. 'oku' - 'to put', 'kuru' - 'to come') follow later.
The first modifications in the noun phrase occur with noun phrases NP -> N no, then with NP -> N no N. Sometimes there is an intermediate step with the structure NP -> N N (Clancy 1985, 483) before N no N is produced, 'no' is often overgeneralised and used for modifiers such as adjectives and relative clauses as well (Clancy 1985, 458f)2. The relative time of acquisition of adj-N-phrases cannot be clearly defined, but it must be assumed to be after N no N and adj no N-phrases have been acquired.
Clancy suggests a developmental path for the acquisition of adjective morphology (Clancy 1985, 403ff). The crucial point is that children add negation and past tense marker to the adjective, leaving it uninflected, before they inflect the adjective itself (sentences 45-47 from Clancy 1985, 403f)):
45. *Atsui- nni.
hot (neg)
(It) is not hot.
46. *Samui- kunni.
Cold (neg)
It is not cold.
- Analysis of data in this study will show that JSL learners tend to overgeneralize 'no' and link adjectives
to nouns by it as well. Huter (1992) shows that JSL learners also tend to link relative clauses to the head noun by 'no'.
47. *nbunai- knttn. dangerous (past) (It) was dangerous.
Negated past is the last inflection acquired. Clancy (1985, 405) describes an interim stage, in which children place, incorrectly, the past before the negation marker: yo- katta-kunai' - 'It was not good1, na-adjectives are used and inflected like adjectives before their specific morphology is acquired (examples from Clancy 1985, 403f):
48. *Suki- nai.
likable- (neg) I don't like it.
49. *Kirei katta.
beautiful (past) It was beautiful.
Coordination is also acquired after N no N-phrases. Like verb morphology, conjunctions are produced relatively early (Clancy 1985, 439). According to Clancy (1985, 439), the first conjunctions are V-te and V-temo, i.e. verbal morphemes. It seems, however, that conjunctions such as 'kedo' - 'but' and 'kara' - 'because' should for the first phase be interpreted as verb morphology as well, or as information added to the canonical word order in final position rather than as complementizers, because in the beginning phase of acquisition there follows, according to Clancy, no second clause.
An example from Clancy (1985, 504) is illustrative:
50. (h)atte ageru kara, mama.
paste give because mama
Because I'll paste it for her, mama.
The acquisition of relative clauses starts at the same time as the first coordinating conjunctions with one-word (i.e. verb) relative clauses. More complex relative clauses and subordination are acquired later (Clancy 1985, 382). In the acquisition of relative clauses, several reorganisations of the rule system are necessary (Clancy 1985, 466ff); in comprehension, children tend to understand relative clauses as sequential clauses, perhaps due to the lack of relative clause markers. Leftstanding versus embedded position and the grammatical role of the head noun appear to be the main factors influencing the order of acquisition.3 The acquisition of all types of relative clauses therefore stretches over a long period. The same is true for passives and causatives, whose acquisition goes through several stages of reorganisation (Clancy 1985, 463). The onset-time of acquisition for passives and causatives is, in relation to other structures, late.
3 Huter (1992) finds that in JSL acquisition, the order of acquisition of relative clauses follows the order of