Not informers but guides to memory: who told researchers about Allika, and how far can we trust them?

Four roles: language informant, collector, correspondent and archive

A language informant lived in the locality or knew its dialect well and supplied pronunciation and local memory. A collector interviewed and wrote. A correspondent contributed material to a scholarly institution over a longer period. EKI now preserves, describes and digitises the cards; the archive gives us access, but did not witness the events recalled.

EKI's guide shows how carefully a keelejuht could be documented: name, age or birth year, origin, education, occupation, parents' origins, places of residence and command of dialect. Not every record is complete, but the scholarly logic is clear — researchers need to know how a person acquired local knowledge.

A person named inside a story has another role again. E. Tärküll appears on the Allika card as a craftsman and smith. The card is about him; it does not show that he spoke to the collector, and still less that he legally owned the land.

Alma Tärno: the nearest identified local voice to Allika

The “5 Allika talu” card of 12 July 1968 bears the name Tärno. EKI's official dialect survey identifies Alma Tärno of Oandu as a major Lüganuse keelejuht, with 7 hours 10 minutes of recorded speech. This is not merely a stray name in a margin but a person deeply connected with Oandu's language memory.

Care requires us to stop one step short of a perfect attribution. The Tärno signature and the archive entry strongly connect the sheet with Alma Tärno, but one scan does not reveal the whole interview. It cannot conclusively tell us whose hand wrote each sentence or what the signature signified within the collection's working method.

The content can be tested separately. It places Allika east of the river, calls it a former farm, identifies the land as manor land and connects the place with the smith E. Tärküll. The river clue is especially useful because it can be compared with maps and differs from the Karjaste maja card west of the river.

Aino Källo and B. Brambat: turning conversations into a corpus

Aino Källo was not an occasional interviewee but a correspondent. The Lüganuse survey attributes 1,598 pages and 15,487 cards to her, while the catalogue identifies her 1968 collection of 811 place names, including Oandu. That scale suggests systematic, experienced work, but it does not remove the need to read each sheet critically.

B. Brambat is named on the title card of the 1948 collection: 683 cards from Maidla municipality. It contains Allika, Tärklase and a separate Allika/Lipu strand. The displayed sheets do not identify the local speaker, so the precise formulation is that Brambat collected and organised the corpus; the person who supplied these particular forms remains unknown.

An experienced collector generally improves structural reliability: place, object type and name variants are recorded consistently. The content can still depend on one person's memory, the wording of a question and what the writer chose to omit.

Why they did it — and what we do not know about motive

The collection's institutional purpose is visible: preserve popular place names, pronunciation and dialect words before they disappeared. For linguists this is speech evidence; for historians, a map of human use of landscape; for descendants, sometimes the only trace of a name that never entered a register.

Personal motive is harder. Someone may have wished to help, felt pride in home, trusted the collector, or wanted to save fading memory. A correspondent may have regarded collecting as valuable cultural work. All are plausible, but none can be assigned to a named person as fact without a letter, diary or interview.

The cards date from 1948 and 1968, in Soviet Estonia. That context should not be hidden, but neither should it be turned into a conspiracy without evidence. The sheets belong to a linguistic and toponymic collection; they contain no sign of political reporting, loyalty assessment or an operational task.

How far to trust them: judge the claim, not the person

The cards answer “what was this place called?” best. The reciprocal Allika ↔ Tärklase references of 1948 strongly support a local connection between the names. The separate Allika/Lipu card is equally valuable because it prevents us from merging every match into one convenient farm.

Relative location clues — a side of the river, a neighbouring place, manor land — are also useful. They are not coordinates, but they can be compared with maps. Decades-old recollections of people and dates are weaker: a 1968 memory of Karjaste maja in 1922 is a valuable lead, not a final register of residents.

A place-name card is weakest as proof of ownership. “Smith E. Tärküll” beside Allika shows a strong local association, but not whether he lived, rented, worked or owned land there. That requires a land register, deed, tenancy file, resident list or another record created specifically to document rights.

This is not distrust; it is calibrated trust. We accept the card within the purpose for which it was made, seek an independent repetition and leave a blank where it is silent. That is how someone else's memory becomes testable Allika history rather than a ready-made legend.