The Living Machine: A Computational Approach to the Nineteenth-Century Language of Technology

Area of Science:

• Digital Humanities
• History of Technology
• Computational Linguistics

Background:

• The concept of the 'living machine' is a recurring trope in the history of technology.
• Traditional methods struggle to analyze large-scale textual data for nuanced figurative language.
• Advancements in computational methods offer new possibilities for historical inquiry.

Purpose of the Study:

• To investigate the trope of the living machine in 19th-century texts using computational analysis.
• To demonstrate the utility of neural language models for historical research.
• To explore how figurative language shaped perceptions of human-machine interchangeability.

Main Methods:

• Application of a cutting-edge neural language model (masked language model) to large digitized text collections.
• Detection of unusual or surprising turns of phrase indicative of figurative language.
• Qualitative analysis (close reading) of identified sentences to understand contextual meanings.

Main Results:

• The neural language model successfully identified figures of speech portraying machines as alive or automatic.
• Analysis revealed complex ways in which figurative language rendered humans and machines interchangeable in 19th-century discourse.
• Findings highlight the limitations of simple keyword searches for uncovering such linguistic nuances.

Conclusions:

• Neural language models offer powerful affordances for historical research, enabling new analytical perspectives.
• Judicious application of these models can uncover subtle historical meanings and conceptual shifts.
• The study opens new avenues for exploring the history of technology through computational text analysis.