Package 'leadeR'

Title: Profiling Leaders at a Distance
Description: Profiles political leaders at a distance from text data such as speeches, interviews, press conferences, and other public statements. Computes Leadership Trait Analysis scores for seven personality traits -- including need for power, conceptual complexity, and self-confidence -- and classifies leaders into one of eight leadership styles. Also computes Operational Code Analysis scores summarising a leader's beliefs about politics and the use of power.
Authors: Mitsuru Mukaigawara [cre, aut] (ORCID: <https://orcid.org/0000-0001-6530-2083>), Joshua Kertzer [aut] (ORCID: <https://orcid.org/0000-0002-7358-0638>)
Maintainer: Mitsuru Mukaigawara <[email protected]>
License: MIT + file LICENSE
Version: 0.1.0
Built: 2026-05-18 21:15:57 UTC
Source: https://github.com/cran/leadeR

Help Index

Clean text for analysis

Description

Removes editorial annotations in square brackets, parentheses, and curly braces from political speech transcripts. Also normalizes whitespace and collapses double em-dashes left behind by removals.

Usage

clean_text(x)

Arguments

x

A character vector of text to clean.

Value

A character vector with annotations removed and whitespace normalized.

Examples

clean_text("We must act now [applause] for the future.")
clean_text("The president (speaking loudly) left the room.")

Compute affiliation scores

Description

Measures the affiliation orientation of a leader's speech by classifying verb-level actions as affiliative (A) or other-affiliative (OA) based on the speaker's own-entity references and sentence-level context.

Usage

get_aff(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character vector of entity names representing the speaker's own country or group.
text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are A and OA. When bootstrap = TRUE, columns are meanA, meanOA, varA, varOA.

Compute conceptual complexity scores

Description

Counts high-complexity (HC) and low-complexity (LC) markers in the text, using word stems, phrases, and exact words with negation-aware counting.

Usage

get_complex(
  text,
  bootstrap = FALSE,
  B = 1000,
  quote_strip = TRUE,
  window_chars = 40
)

Arguments

text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.
quote_strip

Logical; if TRUE, remove quoted text before counting. Default is TRUE.
window_chars

Integer; number of characters to look back for negation context. Default is 40.

Value

A one-row tibble. When bootstrap = FALSE, columns are HC and LC. When bootstrap = TRUE, columns are meanHC, varHC, meanLC, varLC.

Compute self-confidence scores

Description

Classifies first-person pronoun occurrences in the text as self-confident (SC) or other-self-confident (OSC) based on sentence-level context patterns covering three conditions: self as instigator, self as authority, and self as recipient of positive recognition.

Usage

get_conf(text, bootstrap = FALSE, B = 1000)

Arguments

text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are SC and OSC. When bootstrap = TRUE, columns are meanSC, meanOSC, varSC, varOSC.

Compute control (need for power) scores

Description

Classifies verbs associated with first-person subjects as instrumental-control (IC) or other-control (OC) using dependency parsing to link subjects to their governing verbs.

Usage

get_ctrl(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character vector of entity names representing the speaker's own country or group.
text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are IC and OC. When bootstrap = TRUE, columns are meanIC, meanOC, varIC, varOC.

Compute distrust scores

Description

Performs per-entity classification of references in the text as suspicious (S) or other-suspicious (OS) based on distrust verbs, harmful nominals, and contextual modifiers.

Usage

get_dist(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character vector of entity names representing the speaker's own country or group.
text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are S and OS. When bootstrap = TRUE, columns are meanS, varS, meanOS, varOS.

Leadership Trait Analysis (LTA)

Description

Runs all eight LTA trait functions at once and returns a single combined tibble. The eight traits are: power, affiliation, distrust, conceptual complexity, task orientation, self-confidence, nationalism, and control.

Usage

get_lta(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character string identifying the speaker's country or entity (e.g., "United States").
text

A character string containing the speech text to analyse.
bootstrap

Logical. If TRUE, returns bootstrap means and variances instead of raw counts. Default is FALSE.
B

Number of bootstrap iterations. Default is 1000.

Value

A one-row tibble::tibble.

When bootstrap = FALSE, columns include raw counts (P, OP, A, OA, S, OS, HC, LC, TI, IP, SC, OSC, N, ON, IC, OC) plus trait proportions (Pp, D, C, Ta, Ss, Na, B).

When bootstrap = TRUE, columns include bootstrap means and variances for the raw counts (meanP, varP, meanOP, varOP, ...) plus trait proportions and their delta-method variances (Pp, varPp, D, varD, C, varC, Ta, varTa, Ss, varSs, Na, varNa, B, varB).

Examples

# Requires spaCy to be installed; see spacyr::spacy_install().
spacyr::spacy_initialize()
get_lta(own_entity = "United States", text = "We will defend our nation.")
get_lta(own_entity = "United States", text = "We will defend our nation.",
        bootstrap = TRUE, B = 500)

Compute nationalism scores

Description

Classifies entity references in the text as nationalistic (N) or other-nationalistic (ON) based on own/other entity detection and contextual modifiers at the referent level.

Usage

get_nat(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character vector of entity names representing the speaker's own country or group.
text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are N and ON. When bootstrap = TRUE, columns are meanN, meanON, varN, varON.

Compute operational code (VICS) scores

Description

Performs full VICS (Verbs in Context System) operational code analysis on a speech text, classifying verb-entity pairs as Self/Other and scoring them on a -3 to +3 scale (Punish to Reward). Returns P1–P5 and I1–I5 indices.

Usage

get_oca(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

A character vector of entity names representing the speaker's own country or group.
text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates for all indices. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble with P1–P5, I1–I5, and raw Self/Other counts by score category.

Compute Power Score (LTA)

Description

Classifies verbs in text as reflecting Power (P) or Other Power (OP) based on the Leadership Trait Analysis codebook.

Usage

get_power(own_entity, text, bootstrap = FALSE, B = 1000)

Arguments

own_entity

Character vector of the speaker's country/entity name(s).
text

Character string of the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrap means and variances.
B

Number of bootstrap replications (default 1000).

Value

A tibble with columns P and OP (or meanP, meanOP, varP, varOP if bootstrap = TRUE).

Compute task orientation scores

Description

Classifies lemmas in the parsed text as task-instrumental (TI) or interpersonal (IP) based on codebook word lists, with token-level quote and negation handling.

Usage

get_task(text, bootstrap = FALSE, B = 1000)

Arguments

text

A character string containing the speech text to analyse.
bootstrap

Logical; if TRUE, return bootstrapped mean and variance estimates. Default is FALSE.
B

Integer; number of bootstrap replicates. Default is 1000.

Value

A one-row tibble. When bootstrap = FALSE, columns are TI and IP. When bootstrap = TRUE, columns are meanTI, meanIP, varTI, varIP.

JFK Speech: Inaugural Address (January 20, 1961)

Description

Full text of JFK's Inaugural Address on January 20, 1961.

Usage

jfk19610120

Format

A character string containing the full text of the speech.

Source

https://www.govinfo.gov/app/details/PPP-1961-book1. These original source texts are U.S. government works and are not subject to copyright protection in the United States under 17 U.S.C. § 105. The package authors do not claim copyright in the original presidential speech texts.

JFK Speech: Address Before the UN General Assembly (September 25, 1961)

Description

Full text of JFK's address before the UN General Assembly on September 25, 1961.

Usage

jfk19610925

Format

A character string containing the full text of the speech.

Source

https://www.govinfo.gov/app/details/PPP-1961-book1. These original source texts are U.S. government works and are not subject to copyright protection in the United States under 17 U.S.C. § 105. The package authors do not claim copyright in the original presidential speech texts.

JFK Speech: Commencement Address at American University (June 10, 1963)

Description

Full text of Senator John F. Kennedy's commencement address at American University on June 10, 1963.

Usage

jfk19630610

Format

A character string containing the full text of the speech.

Source

https://www.govinfo.gov/app/details/PPP-1963-book1. These original source texts are U.S. government works and are not subject to copyright protection in the United States under 17 U.S.C. § 105. The package authors do not claim copyright in the original presidential speech texts.

Classify LTA Traits into Hermann's Leadership Typology

Description

Takes per-speech LTA output (from get_lta()) and aggregates trait scores across speeches, then classifies the leader along three dimensions (constraint, openness, motivation toward world) and maps the first two plus task orientation to one of eight leadership styles.

Usage

type_lta(
  lta,
  precision_weighted = FALSE,
  need_for_power = 0.5,
  control = 0.44,
  complex_high = 0.56,
  confidence_high = 0.81,
  task = 0.59,
  distrust = 0.41,
  ingroup = 0.42
)

Arguments

lta

A data frame with one row per speech, as returned by get_lta(). Must contain columns Pp, B, C, Ss, Ta, D, Na. When precision_weighted = TRUE, must also contain varPp, varB, varC, varSs, varTa, varD, varNa.
precision_weighted

Logical. If FALSE (default), aggregate traits with a simple mean. If TRUE, use inverse-variance (precision) weighting via random-effects meta-analysis (metafor::rma()).
need_for_power

Threshold for the need-for-power trait (Pp). Default 0.50.
control

Threshold for belief in ability to control events (B). Default 0.44.
complex_high

Threshold for high conceptual complexity (C). Default 0.56.
confidence_high

Threshold for high self-confidence (Ss). Default 0.81.
task

Threshold for task orientation (Ta). Default 0.59.
distrust

Threshold for distrust (D). Default 0.41.
ingroup

Threshold for in-group bias / nationalism (Na). Default 0.42.

Value

A one-row tibble::tibble with aggregated trait values, standard errors (when precision_weighted = TRUE), and classification columns: constraint, openness, motivation_toward_world, task_orientation, typology, and method.

Examples

# Simple-mean aggregation works on any data frame with the seven trait
# columns; here we use a small illustrative input.
example_lta <- data.frame(
  Pp = c(0.45, 0.52, 0.48),
  B  = c(0.40, 0.48, 0.43),
  C  = c(0.60, 0.55, 0.58),
  Ss = c(0.75, 0.82, 0.79),
  Ta = c(0.62, 0.58, 0.65),
  D  = c(0.35, 0.42, 0.38),
  Na = c(0.38, 0.45, 0.41)
)
type_lta(example_lta)


# Precision-weighted aggregation needs per-trait variances, which come
# from get_lta(..., bootstrap = TRUE). Requires spaCy to be installed.
spacyr::spacy_initialize()
res <- data.table::rbindlist(
  lapply(c(jfk19610120, jfk19610925, jfk19630610), function(x)
    get_lta(own_entity = "United States", text = clean_text(x),
            bootstrap = TRUE, B = 1000))
)
type_lta(res, precision_weighted = TRUE)