Tweeting in Roman Imperial Coinage

Thomas Martin, Thomas Posillico, Neel Smith

College of the Holy Cross

Harl gift

OCRE: a familiar resource on the Web

OCRE

But also a downloadable dataset

Downloadable

Crucial features

  • openly licensed
  • defined data structure
  • stable identifiers

Apply computational tool of your choice

  • geographic information system
  • statistical packages
  • natural language analysis

Tweeting?

Official, sanctioned texts:

  • frequent
  • widely circulated
  • terse
Issues per year

Frequent

  • 97 issues per year
  • ~ every 4 days for 500 years!
Issues per year

But varies by authority

Numbers of issues per mint

Widespread

Tweet-like texts

  • average text ~ 21 characters
  • limited vocabulary
    • e.g., RIC 1-3: only 424 distinct tokens!
  • highly abbreviated
    • e.g. R XL = remissae quadragensimae (LOL)

Building a text corpus

RIC 3, Commodus 171

Diplomatic transcription (from RIC)

LIBERT P M TR P XIII IMP VIII COS V P P

Expanded and normalized version

libertas pontifex maximvs tribvnicia potestate XIII imperator VIII consvl V pater patriae

Standard orthography

23 alphabetic and 4 punctuation characters:

ABCDEFGHIKLMNOPQRSTVXYZ -•←|

An example: libertas

Expanded corpus resolves:

  • orthographic ambiguity (LIB = libertas or 1iberalitas?)
  • morphological ambiguity (nom., gen., dat. all appear)

Context

  • legends of more than 200 issues refer to libertas
  • 29 distinct legends
Color key for libertas classes
Frequency over time of libertas classes
Color keyFrequency over time of libertas classes

Concluding observations

beyond search → exploration of multidimensional data using variety of instruments
replicable scholarship → all analyses available online

Thank you!

Slides and further links:

http://neelsmith.info/current-projects/tweeting/aha/

QR code