The Interplay of Synonymy and Polysemy

3.1 The 400-Sentence Corpus Study 
The main source of data for this work was a 400 sentence data set extracted from the 
Corpus del Español
5
(henceforth CdE) 1900s subcorpus, created by Mark Davies from 
Brigham Young University. For each verb, 100 sentences with finite verb forms were 
selected randomly. 
Arrojar
, 
tirar 
and 
lanzar
 
have nominal forms which are tagged as verbs
6
; 
these were excluded from the sample. Additionally, sentences for each verb were limited to a 
maximum of one per source in order to avoid biases connected to given texts or authors. The 
data was collected in the months of September and November, 2010.
The total occurrences of these verbs are shown in graph 1. Overall, 
arrojar 
is much 
less common than the other verbs; 
tirar 
is the next least frequent verb; 
echar 
and 
lanzar 
are 
the most common in the corpus. 
Each extracted sen-
tence was manually annotated 
for the type of 
INITIATOR
,
MOVANT 
and 
DIRECTIONAL
.
The specific parameters that 
were chosen are discussed in 
the relevant sections. It is 
important 
to 
note 
that 
wherever possible only two or 
three variables were used. 
This was done in order to 
ensure that the resulting data 
5
The CdE corpus was chosen for its useful interface features and its ease of access. It allows searches by 
lemma, and it has a feature for randomizing the sample. This simplified the extraction process, since the 
interface already gave a randomized sample and the data included variations in tense, mood, aspect, person 
and number.
6
The forms were: lanza ‘spear’, arrojo ‘bravery’, tiro ‘shot’ and tira ‘strip’