61 |
Secondly, both Divjak & Gries (2008) and Liu (2010) mentioned adapting their
annotation process based on the data they had available; this type of bottom-up analysis has
been adopted before. Finally, the results of the statistical measures show that there is a
statistically significant difference in the behavior of the verbs across these three variables.
This indicates that the three variables are good predictors of the behavior of the
throw-
verbs.
The results from the data are shown in table 3. The p-value for the entire matrix of
data is 6.697E-14 (with a X-squared of 73.83 and a df of 6). The Cramer’s V is 0.3133,
showing a medium effect size (King & Minium 2008:327-329). Each of the verbs shows a
different preference in the type of
MOVANT
. The table also shows how each verb varies from
the expected values. The goodness-of-fit p-value indicates that the observed data differs from
the expected. The arrows in the table show in which direction each verb differs (See §3.1).
Table 3
. Distribution of
MOVANT
types across all four
throw-
verbs
14
Physical inanimate
Physical animate
Nonphysical
GOF p-value
GOF p-value
GOF p-value
arrojar
50 ↑
6.94E-03
23 ↓
1.72E-02
27
echar
24
37
27
lanzar
11 ↓
7.47E-08
41
48 ↑
2.04E-05
tirar
54 ↑
2.13E-06
28
6 ↓
5.59E-06
Beginning with physical inanimate
MOVANTS
,
the number of examples for
echar
is not
significantly different from the expected value (approximately 1/3 physical animate
MOVANTS
).
Both
arrojar
and
tirar
have significantly more physical
MOVANTS
than expected.
In fact, (over) half of their total sentences include physical inanimate
MOVANTS
.
Lanzar
in
contrast has significantly fewer
MOVANTS
in this category.
With physical animate
MOVANTS
, only
arrojar
has a significant result, having fewer
physical
MOVANTS
than expected. The remaining three verbs have approximately 1/3
MOVANTS
. In the final category, nonphysical
MOVANTS
, the number of examples for both
arrojar
and
echar
fall within the expected values, with approximately 1/3 nonphysical
MOVANTS
.
Tirar
has significantly fewer nonphysical
MOVANTS
, while
lanzar
has significantly
more of this
MOVANT
type.
Table 3 shows that the verbs can be distinguished based on the
MOVANT
type. The
goal now is to understand why the verbs show this distribution. In the following, I will
discuss each
MOVANT
type, describing the most common uses and meanings that the verbs
14
The numbers for
echar
and
tirar
only add up to 88. There are 11 sentences with each verb that do not have
an overt
MOVANT
(see
Do'stlaringiz bilan baham: