TensorAgent: TensorAgent: a model of bilingual lexical selection.
In tyrannomark/bldR: Bilingual-Led Diversity Library

Description Usage Format Value Fields Methods Examples

This class implements the model of individual lexical selection described in Ellison & Miceli (2017 - Language 93(2):255-287) - hereafter EM. It is implemented using the TensorA, because that class facilitates linear calculations parameterised over a variable number of parameter dimensions.

1	TensorAgent

R6Class object.

Object of class TensorAgent.

Documentation: Here are TensorModel's fields. EM abbreviates Ellison & Miceli (2017).
$LanguageMode: This is the language mode parameter on a scale [0,1].
$MonitoringLevel: The is the effort put into monitoring on a scale [0,1].
$Meanings: The meanings currently in the lexicon (a list of strings).
$Languages: The languages modelled in the lexicon (a list of strings).
$NumberOfLanguages: The number of languages modelled in the lexicon (integer).
$LexicalTensor: The tensor with the frequency of occurrence for each language-meaning-form combination.
$version: An integer tracking whether updates to values to avoid repeated calculation of the same values.
$delta_lt: Kronecker delta (1 if l=t zero otherwise), see Equation 2 in EM.
$p_l_t__b: Probability of using language l given a target language t and language mode b.
$p_f_sl: Probability of using form f to represent meaning s in language l, see Equation 3 in EM.
$p_f_st__b: Probability of using form f to representing meaning s when trying to use language t and the bilingual mode is b, see Equation 4 in EM.
$p_l_fst__bm: Probability of identifying language l as the source of form f when trying to express meaning s in language t, see Equation 8 in EM.
$p_f_st__bm: Probability of using form f to express meaning s when aiming to speak language t, given bilingual mode b and monitoring level m, see Equation 11 in EM.
$p_l_t__b_version: The version number associated with the current value of p_l_t__b.
$p_f_sl_version: The version number associated with the current value of p_f_sl.
$p_f_st__b_version: The version number associated with the current value of p_st__b.
$p_l_fst__bm_version: The version number associated with the current value of p_l_fst__bm.
$p_f_st__bm_version: The version number associated with the current value of p_f_st__bm.

$new(): Creates a new, empty TensorModel object.
$clearLexicon(): Empties the lexical memory, removing all meaning-language-form triples.
$setLanguageMode(languageMode): Set the level of interaction between target and other languages.
$setMonitoringLevel(monitoringLevel): Set the level of monitoring exerted by the agent.
$addExample(meaning,language,form,ct=1): Adds meaning-language-form triple to the model if it does not exist already. It then adds ct to the frequency recorded for this triple.
$normalise(t,overIndices): Normalise a tensor t over some vector of indices overIndices.
$constructDataTensor(): Uses the tuples entered using $addExample(..) to build a tensor representing the distribution over experienced meaning-language-form combinations.
$makeMeLanguagePairs(): Constructs a rank-2 tensor, both of whose indices range over languages, and whose values for each dimension combination is 1.0.
$make_p_l_t__b(): Calculate the probability of a language given a target language.
$make_p_f_sl(): Calculate the probability of using form f to represent meaning s in language l.
$make_p_f_st__b(): Calculate the probability of using form f to representing meaning s when trying to use language t and the bilingual mode is b.
$make_p_l_fst__bm(): Calculate the probability of identifying language l as the source of form f when trying to express meaning s in language t.
$make_p_f_st__bm(): Calculate the probability of using form f to express meaning s when aiming to speak language t, given bilingual mode b and monitoring level m.
$as.data.frame(): Create a data frame with columns Meaning, Language, Form and probability of form p, giving the distribution p_f_st__bm.

library(bldR)
ta <- TensorAgent$new();
ta$clearLexicon();
ta$addExample("PHOTO","English","foto",    ct=0.5);
ta$addExample("PHOTO","English","picture", ct=0.5);
ta$addExample("PHOTO","Dutch",  "foto",    ct=1.0);
ta$constructDataTensor();
ta$setLanguageMode(0.4);
ta$setMonitoringLevel(0.8);
ta$make_p_f_st__bm();
df <- ta$as.data.frame();
print(df);