LDC is not an AI lab and not a developer that trains models: it takes corpora in for distribution and licenses them out. Every figure is LDC's own statement with its check date, by the same method as every profile in this independent reference on AI training data licensing.
Each number carries its source and check date, and how DataPacts sources and dates every figure is set out in the methodology.
What the Linguistic Data Consortium Takes In and Licenses Out
Corpora reach the Linguistic Data Consortium from data providers and leave it under licenses to members and non-members, so LDC sits between the people who make language data and the organizations that use it.
- In: corpora from providers, through the submissions portal.
- Out: membership, which gives access to the catalog.
- Out: per-corpus licenses under the LDC User Agreement for Non-Members.
The catalog covers speech, text and lexical corpora, and the entries are finished, documented corpora and not raw recordings or company records. Switchboard-1 Release 2 shows the scale of one entry: LDC states about 260 hours of telephone speech from 543 speakers in around 2,400 conversations, distributed under the non-member user agreement.
The question for a holder of recordings or transcripts is what this route is. It is a distribution channel in which LDC licenses a corpus to many users, and it differs from a sale in which one developer acquires data.
A holder weighs this route against a sale to a developer. A distributor licenses a finished corpus to many users, while selling speech and audio data for AI training to a developer is a bilateral deal with its own terms.
Linguistic Data Consortium Identity: A University-Hosted Consortium
The Linguistic Data Consortium is a consortium hosted by the University of Pennsylvania rather than a separate company, and it has no registry number of its own.
| Attribute | Value | Source |
|---|---|---|
| Name | Linguistic Data Consortium (LDC) | ldc.upenn.edu join page |
| Host | University of Pennsylvania | ldc.upenn.edu join page |
| Separate company or registry number | None: a consortium hosted by the university | ldc.upenn.edu join page |
| Contracting legal entity | Not stated here | n/a |
| Website | ldc.upenn.edu | ldc.upenn.edu |
| Country | United States | ldc.upenn.edu |
| Type | Distributor of licensed speech and text corpora | catalog.ldc.upenn.edu |
| Founding date | Not stated | n/a |
| Non-profit status | Not stated | n/a |
Table as of 2026-11-09: the identity attributes of the Linguistic Data Consortium and the page each comes from.
Because LDC is hosted by the university, agreements run through the host institution and not through a company with its own register entry. The founding date, a non-profit status and the number of members are not stated in the sources checked, so none is given here.
The same hosting question arises when a university licenses its own data; how universities and research groups licensing data to AI companies proceed is covered on its own page.
Linguistic Data Consortium Program Terms and Membership Fees, Dated
LDC states four membership fees on its join page, all paid by the organization that uses the corpora, as checked on 2026-11-09.
| Term | What the record shows | Who pays | Checked | Source |
|---|---|---|---|---|
| Not-for-profit and US government Standard membership | USD 2,400 | Corpus user pays LDC | 2026-11-09 | ldc.upenn.edu join page |
| Not-for-profit and US government Subscription membership | USD 3,850 | Corpus user pays LDC | 2026-11-09 | ldc.upenn.edu join page |
| For-profit Standard membership | USD 34,000, year not labeled | Corpus user pays LDC | 2026-11-09 | ldc.upenn.edu join page |
| For-profit Subscription membership | USD 40,000 | Corpus user pays LDC | 2026-11-09 | ldc.upenn.edu join page |
| Eligibility | For-profit organizations and any organization planning commercial technology development need the For-Profit Membership Agreement | n/a | 2026-11-09 | ldc.upenn.edu join page |
| Rights | For-profit members retain commercial rights to data acquired during membership; perpetual rights to data licensed during membership years | n/a | 2026-11-09 | ldc.upenn.edu membership benefits page |
| Non-member route | LDC User Agreement for Non-Members (example: Switchboard-1 Release 2); per-corpus fee not recorded | Corpus user pays LDC | 2026-11-09 | catalog.ldc.upenn.edu |
| Provider route | Submissions portal | n/a | 2026-11-09 | ldc.upenn.edu join page |
| Provider payment or revenue share | Not recorded in the sources checked | n/a | 2026-11-09 | n/a |
| Referral fee | None recorded | n/a | 2026-11-09 | ldc.upenn.edu join page |
Table as of 2026-11-09: LDC's membership terms, who pays each amount, and what the record holds on the provider side.
The fees run from corpus users to LDC, not from LDC to anyone, and the join page does not label the year they cover. Figures from different buyers use different units, so the fees sit beside other program terms compared without conversion or averaging.
As LDC states on its join page (checked 2026-11-09), for-profit organizations, and any organization planning commercial technology development, need the For-Profit Membership Agreement. That is the agreement behind the two highest fee levels, and a company cannot treat the lower not-for-profit levels as an option when it plans commercial use.
The provider-side rows read "not recorded" because the sources checked state no provider payment, and no referral fee appears in them either.
Member and Non-Member Licenses: What an LDC Corpus User Receives
An LDC membership gives for-profit members commercial rights to the data acquired during membership, and every member keeps perpetual rights to data licensed in its membership years.
Both statements are LDC's own, on its membership benefits page, checked on 2026-11-09, and neither is a promise about every corpus in the catalog. The commercial-rights line applies to for-profit members, which is why that group signs its own agreement, and the perpetual-rights line applies to data licensed during the years a membership is held.
Non-members license individual corpora. Corpora such as Switchboard-1 Release 2 are distributed under the LDC User Agreement for Non-Members, and the per-corpus non-member fee is not recorded here.
A perpetual grant matters to anyone comparing terms because the right to data licensed in a membership year does not lapse when the membership does, and a provider whose corpus is licensed this way cannot expect the use to end.
The wider rules on license term and perpetual licenses cover what that means for a holder.
How Data Providers Contribute Corpora to the LDC
Data providers submit corpora to the Linguistic Data Consortium through its submissions portal, and the sources checked here record no payment or revenue share to providers and no statement of the rights LDC takes.
- Route: the LDC submissions portal.
- Data wanted: speech, text and lexical corpora.
- Payment or revenue share: not recorded in the sources checked.
- Rights granted to LDC: not recorded in the sources checked.
For a provider, "not recorded" means the terms cannot be compared before contact. A provider asks LDC for the distribution terms in writing before submitting a corpus, and treats the answer as the only authority on payment and rights.
Other distributors state more. ELRA says the provider grants a distribution license, with compensation set by contract, and Mozilla Data Collective states that uploaders keep 100% of revenue; each is a party claim checked on 2026-11-09.
LDC's own distribution terms are the ones to read before any submission, and selling a corpus directly follows different steps, set out in the guide on how to sell a dataset.
Membership fees are payments by users to LDC and are not income of a provider.
Linguistic Data Consortium Reputation: The Record as of 2026-11-09
No complaint or incident involving the Linguistic Data Consortium from a primary or named-person source appears in the record checked on 2026-11-09.
Library guides and forum threads that rank for LDC queries describe how users obtain corpora; they are not evidence of a complaint and are not counted. The absence of a recorded incident is a statement about the sources checked on that date, not a rating of LDC.
Documented incidents across buyers and intermediaries are logged in the dated log of data buyer incidents, which holds no LDC entry.
Benefits, Risks, Downsides and Alternatives for a Data Holder Considering the LDC
For a data holder, the Linguistic Data Consortium offers a distribution route for language corpora with published user fees but no recorded provider payment, so the practical alternatives are distributors and platforms that state provider terms, and the option of not licensing at all.
Benefits (as the record supports them)
- An established catalog with published fees and member agreements.
- A portal with a defined route for providers.
- Catalog reach to members across sectors, with separate fee levels for for-profit and not-for-profit organizations.
Risks
- Provider payment and the rights LDC takes are not recorded, so a provider cannot compare them before contact.
- Members keep perpetual rights to data licensed during membership, so a released corpus stays in use.
Downsides
- No payout or referral fee is recorded.
- The route suits finished language corpora, not company records in general.
Alternatives (alphabetical, each a party's claim checked 2026-11-09)
- Defined.ai supplier program: no exclusivity and no minimums (section 3.5).
- ELRA: the provider grants a distribution license, with compensation by contract.
- Gengo Shigen Kyokai (GSK): provider compensation not stated.
- Mozilla Data Collective: uploaders "keep 100% of revenue".
- Not licensing at all, which keeps the data and every later option open.
These are claims by each party, not payments made. The full set of alternatives to selling data includes keeping the data.
The Linguistic Data Consortium Among Corpus Distributors and Vendors
The Linguistic Data Consortium is a language-resource distributor on the vendor side of the reference, profiled next to ELRA and GSK.
It is profiled with the other dataset vendors and data labs, and the hub on who buys AI training data places it beside every other buyer. When a distributor states no provider terms, choosing a buyer for your data covers the next step.
- Distributors: ELRA, the European language-resource distributor, and Gengo Shigen Kyokai (GSK), the Japanese one.
- Catalogs and vendors: Defined.ai and its Defined.ai alternatives, Mozilla Data Collective, Nexdata and SpeechData.ai.
- Data types: text corpora for AI training, translation memories and parallel text and multilingual and low-resource language data.
- Prices: AI training data prices across data types.
More About the Linguistic Data Consortium
Frequently Asked Questions About the Linguistic Data Consortium
Does the Linguistic Data Consortium pay data providers?
No payment or revenue share to providers is recorded in the sources checked as of 2026-11-09. LDC's published fees are paid by the organizations that use its corpora.
How much does LDC membership cost?
LDC states USD 2,400 (Standard) and USD 3,850 (Subscription) for not-for-profit and US government members, and USD 34,000 and USD 40,000 for for-profit members, checked 2026-11-09. The year is not labeled.
Can a company use LDC corpora commercially?
LDC states that for-profit members retain commercial rights to data acquired during membership (checked 2026-11-09). For-profit organizations sign the For-Profit Membership Agreement.
Is the Linguistic Data Consortium an AI lab or a company?
Neither: it is a consortium hosted by the University of Pennsylvania that distributes corpora, with no separate company registration.
How does a provider submit a corpus to the LDC?
Through LDC's submissions portal for providers. The distribution terms a provider signs are not recorded here, so ask LDC for them in writing first.