Natural language understanding requires the ability to map language such as color descriptions, spacial instructions and gradable adjectives to real-world physical properties. In the paper, 'Good Night at 4 pm?! Time Expressions in Different Cultures', Dr. Vered Shwartz focuses on temporal grounding, particularly mapping time expressions such as "morning" and "evening" to hours in the day.
In this Q&A with Language Sciences, Dr. Shwartz discusses what variables impact time expressions, how the seasons may lead to varying interpretations of time expressions, and more!
How might the seasons lead to varying interpretations of time expressions? What are some other variables that impact time expressions?
The interpretations of time expressions may vary across seasons because of the differences in sunrise and sunset times. Think about how people tend to greet “good night” earlier in the day during the winter, when it gets dark early in the northern hemisphere. I'd be really surprised if someone greeted me with "good night" at 6 pm during the summer, but it would seem perfectly normal at 4:30 in December.
I conducted the survey only once, so the claim that individual interpretations differ across seasons is only based on the comments I got from the survey respondents. With that said, I did find some match between the average sunset time in the capital of each country at the time of the survey (October) and the average night start time. For example, the average sunset time in the US was 6:30 pm and the average night start time among US-based survey respondents was 6:59 pm. In India, it was 5:52 pm vs. 4:49 pm, and in Italy it was 6:30 pm vs. 6:22 pm. I would have liked to have a finer-grained analysis based on more specific location (e.g. states, provinces, etc.), but this was a small-scale survey.
There are many other variables that impact time expressions. Age might be one of them. For example, younger people typically go to bed later, which might affect their interpretation of "night". The day and hour in which I published the survey might have also introduced some sampling bias, that is, the results don’t perfectly reflect the opinions of the general population. I published the survey during my working hours, which might have been outside working hours for some countries. An early riser answering a survey at 5 am or a night owl that answers it at 2 am might not be representative of the population. Finally, people might consider "morning" to start later if they were asked about it during the weekend rather than the weekday. This study resulted in many followup questions which I may research in the future.
Can you discuss some variations you found for time expressions relating to "afternoon" or "evening"?
Let me start with a personal anecdote that was one of the inspirations for this study. A few years ago, I went to a conference and my American friend and I planned to go sightseeing together on the day after the conference ended. In the morning, he texted me that we can meet in the afternoon. I responded with “why so late?”. How much sightseeing can one get done in one day if you only start at 4 or 5 pm? But he literally meant that we could meet any time after 12 pm, which is a reasonable time to leave your hotel room after a week of exhausting conferencing. The reason I misinterpreted “afternoon” was because I was biased by the norm associated with the literal translation of “afternoon” to my native language, Hebrew, which is typically used later in the day.
In the study, I found that in many languages, "noon" indeed refers to a short period of time after 12 pm, but there was more variation in the end of the afternoon and the beginning of the evening. I'm assuming it's partly because unlike "night" and "morning" that depend on sunrise and sunset times, "afternoon" and "evening" are dictated mostly by culture norms such as dinner time. Dinner time in North America is much earlier than in, for example, Spain.
Another reason for variation is that some languages divide time into different slots. For example, both Spanish and Portuguese don't have a word for "evening". So even among the survey respondents from Brazil there was a lot more variation in the interpretation of afternoon and evening.
How might using a 12-hour clock, as compared to a 24-hour clock, impact expressions of time?
In 12-hour clock, 7:00 may refer to either 7 am or 7 pm. The am / pm marker helps disambiguate this time mention, but it might be omitted when it could be inferred from the context. For example, given the sentence "I wake up every morning at 7:00", people rely on their commonsense to eliminate the alternative interpretation that this person wakes up at 7 pm every day. The speaker who decides to omit the am / pm marker assumes that the listener shares roughly the same mapping from time expressions to times and would be able to understand. The 24-hour clock is unambiguous so it doesn't rely on any cultural commonsense.
While conducting this research, did you come across any new, surprising, or unique expressions of time?
Yes! The survey also allowed people to suggest additional time expressions that exist in their native language. Many suggested an expression that spans the time between midnight and sunrise, which they referred to as “midnight”, “after midnight”, “late night”, “early morning”, and “dawn”. Other suggestions included “twilight” (6-7 pm, India), “sunrise” (5-6 am, Italy), “late morning” (11-11:59 am, Italy), “after lunch” (1:15-2 pm, Italy), and “late afternoon” (3-4 pm, Italy). The use of “midnight” to indicate a range of time was surprising to me because I consider midnight as the exact time 12 am, although I realize I’m inconsistent with my interpretation of noon. Maybe it was clearer if it was more common to call it “midday” instead of “noon”.
Click here to read the full paper.
Written by Kelsea Franzke