Linear B

Linear B

Linear B is the deciphered Aegean syllabary — it writes Mycenaean Greek, the earliest attested form of the language. pyaegean reads it through the same Script plugin model as Linear A: a sign inventory, transliteration to phonetics, a bridge into the Greek track, and the accounting reconciliation. Because Linear B is read, the work here is verifiable rather than exploratory.

import aegean
from aegean.scripts.linearb import word_to_phonetic, greek_reading

aegean.registered_scripts()              # ['cypriot', 'cyprominoan', 'greek', 'lineara', 'linearb']
word_to_phonetic("QA-SI-RE-U")           # 'kwasireu'  (gʷasileus, the ancestor of βασιλεύς)
greek_reading("PO-ME")                   # ('ποιμήν', 'shepherd')

Sign inventory

The inventory is built from the Unicode Character Database — the Linear B Syllabary and Ideograms blocks — so it is authoritative and freely licensed (see Data & Provenance). It holds 211 signs: 74 syllabograms (each with its phonetic value), 14 still-undeciphered symbols, and 123 ideograms/monograms for commodities (grain, wine, oil, people, livestock…). Every sign keeps its Bennett number (B008, B131) and its Unicode name.

from aegean.core.script import get_script

inv = get_script("linearb").sign_inventory
ka = next(s for s in inv if s.label == "KA")
ka.glyph, ka.phonetic, ka.attrs["bennett"]   # ('𐀏', 'ka', 'B077')

Transliteration → phonetics

word_to_phonetic converts a hyphenated transliteration to a phonetic Latin form, with the labiovelar (qa → kwa) and affricate (za → dza) values. The complex signs a2/a3/pu2 are kept distinct from a/pu.

word_to_phonetic("WA-NA-KA")   # 'wanaka'   (ϝάναξ, "king")
word_to_phonetic("TI-RI-PO-DE")# 'tiripode' (τρίποδε, "two tripods")
word_to_phonetic("WO-NO")      # 'wono'     (ϝοῖνος → οἶνος, "wine")

Bridge to Greek

Linear B is Greek, so a transliterated word resolves to its Classical Greek lemma and meaning. greek_reading returns (lemma, gloss) from a 150-entry lexicon of well-established equations — a hand-curated core layered with entries extracted from Wiktionary's Mycenaean Greek pages (via the kaikki.org dump, CC BY-SA): only entries whose etymology states the Ancient Greek equation are taken, so every bridge is source-attested rather than reconstructed. Pass the lemma on to the LSJ backend for the full entry.

from aegean.scripts.linearb import greek_reading, gloss

greek_reading("WA-NA-KA")   # ('ἄναξ', 'king, lord (wanax)')
greek_reading("TE-O")       # ('θεός', 'god')
gloss("DO-E-RO")            # 'slave, servant (male)'   (δοῦλος)

# with the LSJ lexicon active, get the full dictionary entry for the reading:
import aegean
aegean.greek.use_lsj()
lemma, _ = greek_reading("PO-ME")
aegean.greek.gloss(lemma)   # 'ποιμήν: herdsman, shepherd …'

Accounting

Linear B tablets are administrative records — names, commodity ideograms, and numerals, often with a to-so/to-sa (τόσος, "so much") total. The script-agnostic accounting engine reads them directly, using Linear B's total markers in place of Linear A's KU-RO.

from aegean.analysis import balance_check

corpus = aegean.load("linearb")
for doc in corpus:
    for chk in balance_check(doc):
        print(doc.id, chk.marker, chk.computed_sum, "==", chk.stated_total, chk.balances)

The reconciliation is heuristic — section boundaries are inferred — so a balance is evidence, not proof, exactly as for Linear A.

The corpus

No openly-licensed Linear B corpus exists. The most complete one, DAMOS (Database of Mycenaean at Oslo), is CC BY-NC-SA, and LiBER is all-rights-reserved — neither can be redistributed in an Apache-2.0 package. So pyaegean bundles an 18-tablet illustrative sample — PY Ta 641 (the tablet that confirmed Ventris's decipherment) and PY Er 312 hand-curated, plus sixteen one-line excerpts from Pylos, Knossos, and Mycenae tablets taken from sourced quotations in Wiktionary's Mycenaean entries (each cites its tablet and carries a translation; CC BY-SA). These are excerpts to exercise the tools, not editions — the full corpus stays bring-your-own.

From a LiBER selection (interim recipe). LiBER's interface exports a selection by copy-to-clipboard ("e.g., to be pasted into an Excel spreadsheet"). Your own copied selection — your use, under LiBER's terms; pyaegean fetches and re-hosts nothing — can be loaded through Corpus.from_records: save the paste as CSV with columns like id,site,text, then:

import csv

import aegean

with open("my-liber-selection.csv", encoding="utf-8") as f:
    records = [
        {"id": r["id"], "text": r["text"], "meta": {"site": r.get("site", "")}}
        for r in csv.DictReader(f)
    ]
corpus = aegean.Corpus.from_records(
    records, script_id="linearb",
    provenance=aegean.Provenance(
        source="My LiBER selection (manual export)",
        license="© CNR Edizioni — all rights reserved; personal research use",
        citation="LiBER — Linear B Electronic Resources (Del Freo & Di Filippo, CNR).",
    ),
)

Point pyaegean at your own licensed EpiDoc export (e.g. a DAMOS download) and it parses it locally, never re-hosting:

pip install "pyaegean[epidoc]"                 # the EpiDoc reader (lxml)
export PYAEGEAN_LINEARB_CORPUS=/path/to/damos   # a file or directory of EpiDoc XML

aegean.load("linearb")                          # now loads your corpus
# or explicitly:
from aegean.scripts.linearb import load_epidoc_corpus
load_epidoc_corpus("/path/to/damos")