NLP at Stanford before the Stanford NLP Group

Joyce Friedman

The first computational linguistics/NLP faculty at Stanford was Joyce Friedman (1928–2018; at Stanford 1965–1968). She joined Stanford Computer Science the year the department was founded (although several CS faculty had been at Stanford earlier in the Department of Mathematics) and was the first woman on the CS Faculty. However, she moved in 1968 to the University of Michigan and then later to Boston University. She was President of the Association for Computational Linguistics in 1971 (again, as the first female president). Joyce worked on topics such as implementing Chomsky’s Aspects-era transformational grammar and formal reasoning. She was the advisor of several people who later became prominent including David Scott Warren and C. Ray Perrault (who in turn was the PhD advisor of James Allen).

Joyce Friedman

C. Ray Perrault   David Scott Warren

Raj Reddy

The other prominent early language technology person was Raj Reddy (1937–; Stanford PhD student 1963–1966; Stanford CS faculty 1966–1969; thereafter at Carnegie Mellon University). Raj came to Stanford before the CS department was created and worked with John McCarthy. He has described how at that time either “you did numerical analysis or AI with McCarthy”. He began working on speech recognition during his PhD at Stanford. You can watch the Hear Here video on this project from 1968 on YouTube. Raj continued on with this topic, later receiving a Turing Award for his pioneering research on speech recognition. PhD students of Raj Reddy include James Gosling, Kai-Fu Lee, Roni Rosenfeld, and Alex Waibel.

Raj Reddy

Roger Schank

The next NLP faculty at Stanford CS was Roger Schank (1946–2023; Stanford CS faculty 1968–1973).

Roger Schank

Roger Schank was a controversial and mercurial figure in NLP, but he left a broad legacy. Three of the 2025 Stanford NLP faculty are academic descendants of Roger Schank, as shown in the below partial family tree.

  • Jacob Mey (a linguist specializing in pragmatics)
    • Roger Schank
      • Eduard Hovy
        • Diyi Yang
      • Jaime Carbonell
        • Manuela Veloso
        • Masaru Tomita
        • Hieu Pham
      • Gerald DeJong
        • Jude Shavlik
        • Ray Mooney
      • Wendy Lehnert
        • Claire Cardie
          • Yejin Choi
      • Robert Wilensky
        • Peter Norvig
        • James Martin
        • Marti Hearst
        • Dan Jurafsky

Terry Winograd

Roger Schank was followed by Terry Winograd (1946–; Stanford CS faculty 1973–2012). Winograd’s most famous NLP system was SHRDLU, the blocks world with language understanding, planning, and reasoning that he wrote for his PhD prior to coming to Stanford. Here is a video of SHRDLU. Terry became disillusioned with the symbolic AI methods he had been trained in, and by around 1984, he had stopped doing AI and hence NLP, and began to reinvent himself as a human-computer interaction researcher. He founded the Stanford HCI Group and for many years was its only faculty member.

Terry Winograd

The (Stanford) NLP Winter 🥶

There was no speech/NLP research at Stanford for 15 years, 1984–1999.

There was NLP Research at nearby labs: SRI, Xerox PARC, and HP, and at further away places like IBM Research (Yorktown Heights) and Bell Labs. Martin Kay was 50% in Stanford Linguistics but his research was done at PARC. A number of Stanford Linguistics and Computer Science Students did complete PhDs on computational linguistic topics during this period by working at one or other of those labs. They include: Mark Johnson (1987), Dan Flickinger (1988), Stuart Shieber (1989), and David Magerman (1994).

Martin Kay

NLP was institutionally weak in the US overall during this period. There were no NLP faculty at MIT, UC Berkeley, Princeton, U. Washington, or UIUC either. The only places with multiple NLP faculty were U. Penn, Carnegie Mellon University, Johns Hopkins, and Georgetown. There were individual NLP faculty at a bunch of other schools including Columbia, Cornell, and Rochester.

Our Linguistics-Heavy Origins

As well as the Schank academic family tree originating from linguistics, this is also true of the other main family tree in our group (somewhat generously counting a formative postdoc relationship between Percy and Tatsu):

  • Noam Chomsky
    • Joan Bresnan
      • Christopher Manning
        • Dan Klein
          • Percy Liang
            • Tatsunori Hashimoto

Chris Potts is himself a linguist descendant of linguists:

  • Neil Smith (a phonetician)
    • Geoffrey K. Pullum
      • Christopher Potts

This leaves Monica Lam as the only Stanford NLP faculty (up to 2025) who is not descended from a linguist, as far as we know:

  • Joseph Traub (a physicist)
    • Hsiang-Tsung Kung
      • Monica Lam