2.5 Conclusion
3.2.1 Description of Causative Errors
This section describes the causative errors made by children in more detail. The causative
errors observed by Bowerman 1982 occur with verbs belonging to a wide range of seman-
tic subclasses. However, at the same time, there are also similarities between the kinds of
verbs that the learner erroneously causativizes. This point is crucial as children extend
causative form on intransitive verbs across a diverse range of meanings. The verbs and
their association with a narrow semantic class, as described by Pinker 1989, is provided
below. The following table of overgeneralized verbs is from Pinker 1989, who summarizes
the errors made by Bowerman’s children.
Table 3.1: Summary of verbs overgeneralized in a causative frame by the children ob- served by Bowerman 1982. The semantic subclass and number of errors noted with each type of verb is also provided (Pinker 1989:303, Table 7.8)
.
Subclass Verbs # of Sentences
Directed motion come, go, fall, rise, drop 30 (28%)
Going out of existence die, disappear, vanish 12 (11%)
Being/staying stay, be, spell, sound, wait 16 (15%)
Possession have, take 13 (12%)
Psychological remember, watch, guess, 12 (11%)
wish, feel, ache, learn
Involuntary emission sweat, blood 3 (3%)
Internally caused state change bloom 1 (1%)
Semivoluntary expression of emotion laugh, cry, giggle 5 (5%)
Voluntary action eat, drink, sing, talk 14 (13%)
swim, climb
Table 3.1 provides an overview of the kind of verbs that children erroneously causativize.1
As can be seen from this table, these verbs belong to a variety of semantic subclasses, but
the property they have in common is that all of these verbs lack intentionality. At first
glance, it appears that all but one subtype of errors, verbs of voluntary action, belong
to unintentional verbs; however, upon closer look it appears that the voluntary action
subclass of verbs is only used causatively in cases when the object is inanimate. Some
1
The work on acquisition done to date primarily refers to these original set of errors (e.g., Ambridge,
Pine, Rowland, and Young 2008). However, the production data from these children are not available in
the public domain, and therefore, the analysis presented in this chapter will largely be based on child data available in CHILDES. These errors are primarily intended to provide an overview of the type of verbs that children make causative errors with.
examples of causative errors with these “voluntary action” verbs are shown below.
(32) Examples from Bowerman 1982 as cited by Pinker 1989:
a. I wanta swim that [Holding an object in the air and wiggling it as if it were
swimming] (E, 2;1)
b. I’m talking my birdie [Pulling string on bird-shaped music box] (E, 2;2)
c. I’m singing him [Pulling string on cow-shaped music box] (C, 3;1)
The examples in (32) are compatible with the idea that the learner is aware of the dis-
tinction between intentional and unintentional verbs. In these cases, it is likely that the
learner is still following the causativization rule for unintentional verbs, while treating
certain typically intentional verbs as unintentional given the circumstance. Foreshad-
owing the analysis presented in this chapter, I argue that children use intentionality in
acquiring verb classes in their language. I return to this point in later sections.
3.2.2.
Experimental Evidence for the Productivity of the Causative
Alternation
In the previous section, we saw that children produce causative errors with verbs that
generally lack intentionality. Up to this point, we have also primarily focused on the fact
that children generalize the transitive causative form to verbs that do not occur in that
frame in the adult grammar. However, an important aspect of the causative alternation
is that it is productive in the adult grammar for certain semantic subtypes of verbs. We
focus on this point in this section.
This section summarizes previous experimental results that find that intransitive (un-
accusative) verbs are productively causativized when they involve external causation in
the adult grammar. The productivity of the causative alternation is important because
any analysis of the acquisition of causatives must also account for the productivity of this
As we have seen in previous sections, studies on the causative alternation have of-
ten emphasized that children use the causative rule productively. To supplement these
findings, Maratsos, Gudeman, Gerard-Ngo, and DeHart 1987 tested whether children and
adults would generalize novel intransitive verbs as a transitive causative. In an experimen-
tal setting, they introduced a novel intransitive verbfud, and examined the spontaneous
speech of adults and children when they used this novel verb in the testing phase of the
experiment. The experiments were conducted on adults and on children aged 4;6 – 6;2.
The novel verbfud in the causative referred to a dough-like substance being converted
into strands by the means of a machine. In the intransitive, it referred to a dough or clay-
like state. In the experiment, a total of six sessions with story-telling and demonstration
tasks for spontaneous production were used overall, in addition to a sentence completion
task. 40 children in total, who were assigned to four groups, participated in the exper-
iment. Group I heardfud only as a transitive (control group), while group II heardfud
only in an intransitive frame. Those in group III were exposed to sentences that were
intransitive with a locative or benefactive, and Group IV also heardfudas an intransitive,
but the children here were not encouraged to use the verb overtly in any way; they were
simply told that they were going to learn aboutfudding. Moreover,fudwas never used as
a periphrastic causative in the exposure phase of the experiments.
The results of the study found that children usedfudas a verb ranging from 8 times
by one child and 107 times by another. If a verb was used with an agent, it was counted as
transitive even when the object was omitted. Although the results varied between groups,
preference for the use of the causative form was found overall. Group I, the control group,
produced transitive causative sentences 98% and 97% of the time in story and demonstra-
tion tasks when they heard the novel verb used in the transitive causative frame. Those
in group II produced the causative form 23% of time in spontaneous production and 37%
of the time in the sentence completion task even when they did not hear the novel verb
speech and the sentence completion task respectively, and group IV did so 21% and 30%
of the time. Even when the children did not hear the novel verb used in a causative frame,
children produced the verb transitively 26% of the time overall. The number of transitive
causatives produced by each subject in groups II through IV Maratsos et al. 1987’s exper-
iment is shown in Figure 3.1. There were a total of 28 participants in these groups where
the novel verb was only used as an intransitive. This figure is taken from Maratsos et al.
1987 (p.101, Figure 4.1). These results are indicative of the productivity of the causative
alternation rule.
Figure 3.1:Number of transitives produced by each subject in Maratsos’s et al. 1987’s ex- periment. The results of the participants illustrated here are from the non-control groups where only the intransitive form of the novel verb was used (N = 28).
The results from Maratsos et al. 1987’s study illustrate that some children are will-
ing to causativize a novel verb, but there is also individual variation where a number of
participants did not causativize the novel verb. Additionally, Gropen, Pinker, Hollander,
and Goldberg 1991 also find evidence for the productivity of the causative rule in En-
glish. They find that children causativize intransitive verbs when the action depicts a
that children produce these causative errors in a systematic way.
The results from Maratsos et al. 1987’s study and Gropen et al. 1991’s study show that
children and adults both readily and productively causativize verbs denoting direct exter-
nal causation from their intransitive frame. The reverse was not found. Thus, throughout
this chapter, I assume that children also causativize from a base intransitive form, and the
generalization under question is that from an intransitive to a causative transitive frame.
Thus far, we have seen that the causative rule is productive when direct external cau-
sation is depicted. There is also evidence in the literature that other semantic subtypes of
verbs productively undergo the causative alternation (e.g., Brooks and Tomasello 1999).
For instance, Brooks and Tomasello 1999 tests two novel verbs, one with manner of mo-
tion semantics (e.g.,roll) and the other with directed motion semantics (e.g.,come). They
found that children over the age of 4 causativized the novel manner of motion intransitive
verb, but not the novel directed motion intransitive. These results indicate that children
eventually learn that certain semantic subclasses of verbs can productively causativize,
even if not all intransitive verbs undergo the alternation.
Throughout this section, we have seen evidence for the productivity of the causative
alternation. We bear this productivity in mind as a theory that assumes a strict conserva-
tive form of lexical learning would not predict that these forms should occur productively
in the first place. The results also indicate that participants were almost always unwilling
to use a novel verb that they had heard in a transitive causative frame as an unaccusative
without positive evidence indicating that the verb could occur as an intransitive. In the
following section, I discuss lexical semantic and usage based approaches that provide an
account of the productive and unproductive aspects of the causative alternation rule, and