The Acquisition of Japanese Focus Particles Dake (Only) and mo (Also)

The Acquisition of Japanese Focus Particles: dake (only) and mo (also) The acquisition of semantic interpretation for adverbs such as only, also, even should be a challenge for young children, since it requires combining information from multiple sub-modules of language: syntax, semantics, and pragmatics. In addition, there is a cross-linguistic difference in how the possible alternative sets are determined. This paper presents novel findings from an experimental investigation of the acquisition of focus particles dake (only) and mo (also) in Japanese. Given that characterizing the nature of the syntactic-semantic interface is one of the important issues in the generative-based acquisition study, the focus phenomenon is a reasonable topic to investigate. The current study provides important empirical data for the syntactic theory of focus items. 1. Japanese Focus Particles The interpretation of too/also in English is not necessarily determined syntactically. In the following examples, the interpretation of the adverb also varies, even though it appears in the same syntactic position in both sentences. (1) a. John also introduced [Bill]F to Sue (There is someone other than Bill, whom John introduced to Sue.) b. John also introduced Bill to [Sue]F (There is someone other than Sue, to whom John introduced Bill.) (Rooth 1996) On the other hand, the range of the alternative set in the interpretation of mo is syntactically determined, as demonstrated in the following examples: (2) Subject+mo: Yusuke-mo jitensha-o kat-ta Yusuke-also bicycle-ACC buy-PAST ‘Yusuke also bought a bicycle (in addition to other people)’. (3) Object+mo: Yusuke-ga jitensha-mo kat-ta Yusuke-NOM bicycle-also buy-PAST ‘Yusuke bought a bicycle, too (in addition to other belongings)’. The particle dake (only) behaves similarly to only in English; as indicated in the English gloss in (4) and (5) below, dake is typically associated with the noun phrase immediately preceding it. (4) Subject+dake: Yusuke-dake jitensha-o kat-ta Yusuke-only bicycle-ACC buy-PAST ‘Only Yusuke bought a bicycle.’ (5) Object+dake: Yusuke-ga jitensha-dake kat-ta Yusuke-NOM bicycle-only buy-PAST ‘Yusuke bought only a bicycle.’ Following Aoyagi (1999), the focus particle mo is referred to as a K-particle, while dake is an F-particle (see Teramura 1991 and references cited therein for a discussion of the traditional classification of the two types of focus particles in Japanese linguistics). It is widely assumed in Japanese syntactic literature that the association between focus particles and the focused items is established by covert or overt movement. Considering the scope interaction between negation and mo, as well as the crossing effect with an NPI and a wh-phrase, Hasegawa (2005) argues that mo phrases undergo overt movement to the Spec of TP. Aoyagi (1999) argues that the association of DP and K/F-particles is licensed by covert movement of the focus particles to a head of different functional projections (T for K-particles, v for F-particles)1. On the other hand, Hoshi (2005) claims that the distribution of K- and F-particles can be captured in the same projection (FocP), though the K-particle can be the Focus head, while the phrase containing the F-particle and the focused element moves overtly to the Spec of FocP. Wherever the landing site of the focus particles and their associated focused items is, there should be no difference between the focused subject and the focused object. Nevertheless, it has been observed in recent language acquisition research that a group of young, Japanese-speaking children exhibits a subject-object asymmetry in their interpretation of sentences that includes focus particles. Endo (2004) reported that a group of Japanese children gave a non-adult response, in which dake was always associated with the object, regardless of its syntactic position. She did not observe any ‘subject-only’ responses. However, the relationship between the number of non-adult responses and age was inconsistent in Endo’s study. Furthermore, some of her test sentences included both dake 1 For a different minimalist approach to focus particles, see Sano (2001a, b) and references cited therein. and the Nominative Case-particle ga. (6) (6) is a representative example: Taroo-dake-ga ringo-o tabe-ta. Taroo-only-NOM apple-ACC eat-PAST ‘Only Taroo ate an apple.’ The sentence-initial Nominative NP often induces the ‘exhaustive listing’ reading in a simple clause. Hence, we predict that if the experiment is appropriately designed, a different picture would emerge. To verify the prediction, a study was conducted to see if Endo’s results could be replicated. Considering that the semantic contribution of the particle mo (also) is similar to that of dake (only), in that both of them can be associated with focused materials, we also conducted separate sessions with sentences which included mo. 2. Subjects and Method A total of 120 Japanese-speaking children from Sapporo and Osaka (4;7-6;10, mean: 5;10) participated between September 2004 and February 2005 (62 in the dake session, 58 in the mo session). The Truth-Value Judgment task (Crain and Thornton 1998, Matsuoka et al. 2005) was conducted in a quiet room of the day-care center or kindergarten that they normally attend. Test sentences with dake did not appear with any Case particles, so that they did not induce the ‘exhaustive listing’ reading. Moreover, each target sentence began with the phrase ‘kono ohanashi de wa (in this story)’. This was done to control the strong preference among Japanese speakers to place the topic marker (wa) at the beginning of a simple sentence2. The following is an example: (7) kono ohanashi de wa, omawarisan-ga kyoryu-dake this story in TOP policeman-NOM dinosaur-only ‘In this story, ｔhe policeman patted only the dinosaur.’ nade-mashi-ta. pat-POL-PAST3 There were three tokens of each of the ‘subject+focus particle (mo/dake)’ and the ‘object+focus particle (mo/dake)’ (see Appendix for the complete list of the test sentences and fillers). Stories and fillers were ordered in such a way that more than one story of the same type was not presented consecutively. The majority of the children participated in either the dake or mo sessions, but not both. 2 We are indebted to Satoshi Oku for suggesting this to us. TOP: Topic, NOM: nominative, ACC: accusative, PERS: personalizer, POL:polite, DIM: diminutive 3 3. Results Three children did not complete the session (one for the mo experiment, two for the dake experiment). Results of the two experiments are summarized below: Chart 1: Number of Adult-like Responses (mo) Subjects Total: 57 Mean age Subject + mo Object + mo Adult-like (N:11) 6;1 88% (29/33) 82% (27/ 33) All-Yes (N:44) 5;10 8% (10/132) 8% (10/132) Subject Orientation (N:2) 5;7 67% (4/6) 17% (1/6) Object Orientation (N:0) n.a. n.a. n.a. Chart 2: Number of Adult-like Responses (dake) Subjects Total: 60 Mean age Subject + dake Object + dake Adult-like (N:29) 5;11 94% (82/87) 91% (79/87) All-Yes (N:12) 5;5 14% (5/36) 8% (3/36) Subject Orientation (N:18) 5;10 91% (49/54) 24% (13/54) Object Orientation (N:1) 5;9 33% (1/3) 100% (3/3) We found a very different pattern from what Endo reported. First, children’s non-adult responses decreased as the age of the subjects increased, both with dake and mo (from 100% to 81% for mo, from 56% to 36% for dake). Moreover, both for dake and mo, we found a larger number of subject-only responses. In the case of dake, the subjectorientated group accounts for 30% of the subjects (N: 18, mean age: 5;10), while there is only one participant who consistently gave the object-oriented responses (age 5;9). At the same time, children treated dake and mo differently: 17% (11 out of 57) of the mo subjects (mean age: 6;4) gave adult-like responses (as shown in Chart 1), while 48% (29 out of 60) of the dake subjects (mean:5;11) gave adult-like responses (as shown in Chart 2). 4. Discussion Children’s response patterns indicate that those two focus items (mo and dake) actually have different syntactic-semantic properties, which provides support to the claim that mo and dake have different syntactic derivations. At the same time, though, the existence of the subject- and the object-oriented responses 4 given to mo and dake, consistently observed in previous studies and the current study, suggest the possibility that human grammar treats the two types of focus items in a similar fashion as well. Namely, our data possibly provide support for Hoshi’s (2005) argument that the distribution of K- and F-particles is captured by postulating the FocP. The non adult-like response patterns reported in our study are not language-specific phenomena; very similar patterns have been observed with English-speaking children. Crain et al. (1993) investigated young children’s interpretation of sentences which contain only. See the following examples: (8) (9) Only the bird is holding a flag The bird is holding only a flag For (8) to be true, there should be no other character (other than the bird) who is holding a flag. On the other hand, for children to accept (9) as true, the bird should be holding nothing other than a flag. Crain et al. reported that young children gave either the subject-only (mean age: 4;8) or an object-only (mean age: 5;0) interpretation of only, regardless of its syntactic position. The subject-only group of children always associated only with the subject for both sentences (8) and (9); the object-only group interpreted only as if it were associated with the object in both sentences. The response patterns obtained in the current studies, as well as Endo (2004) and Matsuoka (2004), match the patterns reported in children’s non adult-like interpretation of only. This strongly implies that both subject-oriented and object-oriented interpretations of focus items are universally allowed options in language development. Our results, however, differed from those reported by Endo; unlike Endo’s subjects, a group of children gave the subject-oriented responses. Actually, they outnumbered the children who gave the object-oriented responses. As discussed in Section 1, Endo’s test sentences contained the Case particles (ga and o). Hoshi and Miyoshi (2005) argued that sentences which contain both dake and a Case particle are syntactically derived in a different fashion from sentences with dake not followed by any Case particle. The different response patterns observed in Endo’s and our studies might provide empirical support to their analysis. A follow-up study is in progress to address the issue. 4 Matsuoka (2004) reported that a group of children of similar ages gave the object-oriented responses to sentences that included mo. Based on her observations, we assume that all four response patterns are observable both in the case of dake and mo. There is an indication of a possible developmental path in the following graph, which shows the ratio of the four response patterns in the dake part of the experiment, with the children divided into three age groups (4, 5, and 6-year-olds). Graph 1: Response Pattern by Age (dake) 100% 80% 60% OBJ SUBJ All-Yes Adultlike 40% 20% 0% ４-Years-Olds (N:9) 5-Years-Olds (N:26) 6-Years-Olds (N:25) Children start with the grammar which allows focus items to be associated with either of the subject or the object (the ‘All-YES’ pattern). Then a group of children go through either the subject-oriented or the object-oriented stage, before giving the adult-like responses. The pattern is very different in the case of mo, as shown in the following graph: Graph 2: Response Patterns by Age (mo) 100% 80% OBJ SUBJ All-Yes Adultlike 60% 40% 20% 0% 4-Years-Olds (N:5) 5-Years-Olds (N:25) 6-Years-Olds (N:27) Even though adult-like responses can be seen in older children (5-6 years olds), there seems to be no clear indication that the likelihood of this response increases with age. There are very few children who gave the subject- or object- oriented responses, either. The all-YES pattern persists throughout the three age groups. Obviously, interpreting the association of focus with mo is more challenging than dake to young children5. It is unlikely that those six-year-olds were not able to comprehend the additive meaning of mo. The following is a summary of the appearance of the focus items from the CHILDES database (MacWhinney 2000). Chart 3: First Occurrence of Focus Particles dake6 AKI (Miyata 2004) mo7 2;7;26 2;3;26 RYO (Miyata 2004) 2;5;1 2;2;29 TAI 2;1;30 1;6;04 TARO (Hamasaki 2004) 3;2;3 2;11;11 JUN (Ishii 2004) 2;6 2;2;20 SUMIHARE (Noji, et al. 2004) 2;2 2;2 (Miyata 2004) At a glance, we can see that mo appears earlier than dake. Except for TARO, who seemed to have taken longer to start using both focus particles, the first productive occurrence of mo is observed between 1;6–2;3, while the first occurrence of dake is between 2;1-2;7. An important observation here is that children’s spontaneous usage of mo and dake starts at the early stage of language development. This means that the non-adult interpretations of focus particles of 4 to 6-year-olds are not simply the result of delayed acquisition of the items in question. It is not clear which of the subject- or the object-oriented interpretation is the ‘standard’ or ‘dominant’ path in children’s language development. However, there is an 5 The difference observed here could be related to the fact that the K-particle controls the felicity of the sentence, while the F-particle changes the truth-value of the sentence (Diane Lillo-Martin, p.c.) For example, sentence (2), repeated below, seems to be awkward (but not false), even if there is no one other than Yusuke who has a bicycle. (2) 6 Yusuke-mo jitensha-o kat-ta Yusuke-also bicycle-ACC buy-PAST ‘Yusuke also bought a bicycle (in addition to other people)’ Dake in ‘kore dake’ or ‘kon dake’ could be interpreted either as focus (i.e. ‘only this’) or amount (i.e. ‘about this much’). For that reason, those phrases were excluded from our analysis. 7 The ages are when children uttered mo with more than one type of DP in the same file. Namely, we excluded the files in which the child used mo with the same DP repeatedly (e.g. ‘koko mo.’ ‘Akichan mo.’) indication that the subject-orientated responses could be the dominant option. Hüttner et al. (2004) noted that German-speaking children (2;11-7;8) performed better with interpreting the stressed auch (an equivalent to English also), as opposed to the unstressed auch. Interestingly, the stressed auch induces subject-oriented interpretation. (10) a. Max will AUCH Boot fahren. Max wants to go by boat like other people go by boat. b. Max will auch Boot fahren. In addition of other vehicles Max want to go by boat. Hüttner et al. (2004) Hüttner et al. interpreted the data differently to argue that children acquire the stressed auch earlier than the unstressed one. However, this seems to contradict the finding that young Dutch-speaking children (Mean age 5;5) have difficulty using prosodic information to interpret sentences with alleen (an equivalent to English only) (Szendrői 2003). Gualmini et al. (2003) also reported that contrastive stress did not reliably act as a cue for children (mean age: 4;9;26) to interpret possibly ambiguous English sentences that include only. An alternative account of Hüttner et al’s data is that children ignored the contrastive stress and assigned the subject-oriented interpretation more often. 5. Summary Focus items mo and dake in Japanese show different syntactic behavior, which led Japanese researchers to classify them into two different categories: K-particle (mo) and F-particle (dake). Nevertheless, our research showed that young Japanese-speaking children go through a similar developmental path as they learn the focus items of different types. The path does not seem to be completely uniform, though: when children interpret sentences which include a focus item, one group of children assigned the subject-oriented interpretation, while the other exhibited the object-oriented interpretation. The focus phenomenon is an exception to the general observation in children’s language development: comprehension precedes production. Hence, investigating the developmental sequence of the acquisition of focus items will provide important insights into the nature of the innate knowledge of language.