It’s the 50th anniversary of the moon landing, a good time to take a tour of lunar linguistics:
Corpus of space and lunar conversation: Over four years, twelve human subjects made a total of 9 trips to the moon lasting 6 to 13 days: the talk during this time, most of which we have on tape, would be a satisfyingly bounded spoken language corpus, you might think. Talk between astronauts and between astronauts and Mission Control was however just the capstone of the heap of texts preserved from the 400,000 workers on the effort, including 100,000 hours of Mission Control tapes – 19,000 hours for Apollo 11 alone – as well as memos, reports and interviews (Sangwan et al. 2013: 1135). Space archivists have been slow to isolate the space talk since the research interest of NASA and its contractors was teamwork and crisis solving.
Both the lunar module and the command module were in constant radio communication, except where no link was possible, for example when passing in back of the moon. During those times, microphones recorded sound, to be played back to base at high speed when the link returned. So in principle, all communications during the mission were public and there was no backstage. However, the recordings were selective: for a start, the recorder in the lunar module had only 10 hours of tape and could not rewind (Sangwan et al. 2013: 1136). Magnetic tape was expensive, in terms of weight carried – and in cost as reckoned by NASA. Some of the video tapes from the Apollo missions were later overwritten with satellite weather images to save money.
The master Apollo mission recordings were made on 30-track SoundScriber tape systems, long obsolete. To avoid having to digitize each stretch of the one-inch tape 30 times, stressing the tape media, a 2-track SoundScriber to be retrofitted with a newly made 30-track head (Hansen et al. 2018: 2758), There are also problems in aligning the audio without claps or timestamps, though communications with base are bounded by intro and outro beeps known as Quindar tones, which functioned as on-off tones for the worldwide network of transmitters. Telecommunications was expensive too.
A team at the University of Dallas completed the digitization of the Apollo 11 tapes in 2018. Recordings are available at the Internet Archive and at NASA, which also has an older page of samples). They are cryptically labeled at present; and the accompanying transcriptions, made soon afterward for engineering analysis, were groomed into written statements – not fit for any linguistic purpose. The Dallas team complained that the unfaithfulness of the transcripts made them 75% unworkable useful for forced alignment; see Sangwan et al. 2013: 1138). Recordings from the lunar module may survive only in a 15-minute excerpt squirreled away by a space buff, part of a team who curate a moon landing archive called the Apollo Lunar Surface Journalon NASA’s servers. This group, founded by a Los Alamos scientist who was inspired by scholarship on Captain Cook’s journals, seem a lot more organized than NASA and have provided corrected transcripts of the missions, which still are not linguistic transcripts.
The digital and pre-digital recordings have already been used in studies of pitch speaker recognition; delays in interaction; and transcription of connected speech among other things:
Pitch: Speech was compared for the same astronauts in phases of Earth, launch (Armstrong only), travel, and the Moon (Armstrong and Collins only). Fundamental frequencies increased by as much as 23% (for Armstrong on the moon). Speakers also sounded less like their Earth selves in space, according to a test with an automated recognition system. (Yu and Hansen 2017) The study has implications for speaker recognition, which is clearly important in long-distance multiparty conversations. The paper notes that a transcription tool from the Linguistic Data Consortium (a private resource holding group based at Penn) was used to improve transcriptions and timing, and the corrected transcripts “will be made public for the research community” (1606).
Turn taking: The time delays of space communication are hugely disruptive to turn-taking, and in long distance missions the delays will be extreme – as much as 20 minutes. Mission simulation experiments with time delays of 50 seconds (as it would be to Mars) and 300 seconds (as to Near Earth Objects such as asteroids) showed that team members overlapped, found their contributions overtaken by events, struggled to manage multiple ongoing threads of talk, and failed to give feedback cues or clearly communicate what was meant by “copy all” and other minimal responses. “Adherence to communication procedures that could facilitate grounding was surprisingly low, given the professional sample in this study, ” the researchers complained in a conference summary, though this observation seems absent from publications. (Fischer, Mosier & Orasanu 2013)
[wʌn smɔːl step f ɚ (ə) mæn] : Neil Armstrong says he said “one small step for a man, one giant leap for mankind.” Yet he was consistently heard as saying “one small step for man,” which is both contradictory and sexist language that would not pass muster today. Regardless of what he actually articulated in the moment, which we will never have access to, how did listeners arrive at their interpretation? Baese-Berk et al. (2016) say,” it is possible that an intended production of for and an intended production of for a may have identical, or nearly identical, acoustic signatures” as the article is regularly assimilated into the preceding word. Syntax is identical in both cases so is no help in distinguishing which was said.
The decisive factor is duration relative to the duration of syllables around the words. An examination of sentences from speakers who like Armstrong came from the Columbus, Ohio area (the Buckeye Corpus) showed that normalized duration of for and for a in similar sentences had only slightly different profiles, with Armstrong’s utterance being “slightly more compatible” with the for a distribution. In a second test, tokens of for a were presented to informants in slowed and unaltered contexts. They were more likely to perceive one word when surrounding words were slowed – they expected fewer syllables per second. If Armstrong said “for a,” those were the fastest syllables in his utterance so it was no wonder it was heard as “for.”
This post was inspired by the following Tweet:
Bonus: The Bedford Citizen is running a series of interviews with people in my hometown who worked on Apollo 11. Several parents of classmates there. I had no idea.