*The Language of Food: A Linguist Reads the Menu*

That is the new and excellent book by Dan Jurafsky, due out this September, and I found it interesting throughout.  Here is just one bit:

In fact, the more Yelp reviewers mention dessert, the more they like the restaurant.  Reviewers who don’t mention a dessert give the restaurants an average review score of 3.6 (out of 5).  But reviewers who mention a dessert in their review give a higher average review score, 3.9 out of 5.  And when people do talk about dessert, the more times they mention dessert in the review, the higher the rating they give to the restaurant.

This positivity of reviews, filled with metaphors of sex and dessert, turns out to be astonishingly strong.

That is another reason not to trust customer-generated restaurant reviews.

And how exactly do Americans conceive of dessert?

Americans usually describe desserts as soft or dripping wet…US commercials emphasize tender, gooey, rich, creamy food, and associate softness and dripping sweetness with sensual hedonism and pleasure.

This association between soft, sticky things and pleasure isn’t a necessary connection.  For example, Strauss found that Korean food commercials emphasize hard, textually stimulating food, using words like wulthung pwulthung hata (solid and bumpy), coalis hata (stinging, stimulating), thok ssota (stinging), and elelhata (spicy to the extent one’s nerves are numbed).

How can you resist a book with sentences such as these?

The pasta and the almond pastry traditions merged in Sicily, resulting in foods with characteristics of both.

Here is a previous MR post on Jurafsky, including a link to his blog, and concerning “Claims about potato chips.”


'Americans usually describe desserts as soft or dripping wet'

Is this another example of the feminization of American culture, compared to the 'solid and bumpy' stimulation associated with Korea?


Or is it infantilisation?

Yes and yes.

Isn't Jurafsky comparing American dessert adjectives to Korean non-dessert adjectives? If so, obviously the signal words differ.

In an American context, women like desserts and men do not, so to some extent this would just map in real terms to the "feminization" of American women.

The sexes both like confectionery more equally , but that's often chewy and crunchy at least as much as soft and gooey (taffy, bubble gum, cookies).

I was watching some Antony Bourdain recently where he remarked on how gelatinous textures common in the particular Asian cuisine he was eating (Bai people food, but I believe gelatinous textures are fairly general to Asian food) were despised by Americans, who he described by contrast as liking anything crisp.

It seems there's a type of soft and yielding that mainstream Americans like, the sort of textures associated with soft white bread, cheese and butter, and unfamiliar kinds of soft textures, examples being the sort common to tofu, rice flour and other Asian bean jellies, that they do not in general like.

The higher ratings given by dessert mentioning diners may just be that dessert fans are positive young women, and dessert loathers are curmudgeonly older men.

The dessert/good review relationship seems highly prone to data selection problems.

Since dessert is a sequential decision, you don't know if you are ordering until you've already enjoyed the appetizers and main dish. In contrast, how often do you order an appetizer, and decide that the service was not good enough or quality too low to justify an entree? Similarly, no one ever said "wow, that steak was mediocre, but I bet they will make up for it with a great piece of pie!"

In other words, it should be obvious that reviews that involve desserts were better appetizers/entrees/service/experience and the possibly interesting question is: Do restaurants understand they need great pastry chefs even when (say) 1:10 diners order a dessert?

Precisely. Survivor bias run rampant.

I agree completely about the survivor bias.

But as to pastry chefs, having worked at a couple of mid range restaurants that were actually well known for deserts, and being friends with quite a few friends still in the business. A lot of places have cakes and pastries prepared, or at the very least prepped off the premises, often by a contractor who supplies multiple restaurants. This extends into the high end as well.

I agree 98%.

Some restaurants are known for fabulous desserts which is why people choose them. They might deliberately go light on the meal and expectations holding out for dessert.

Also, some people are just more enthusiastic about dessert. They are satisfied with mediocre food and an excellent dessert.

Try eating dessert at the beginning of the meal:

"If you believed you were drinking the indulgent shake, she says, your body responded as if you had consumed much more."


Even supposing that we lived in a universe where this didn't hold, this result can be anticipated via the peak-end rule. I predict that feeding two groups a different meal order (one with dessert first and one with dessert last) and then eliciting ratings would reproduce the phenomenon.

I wonder if the reason dessert mentions are correlated with higher ratings is that people who have liked the meal so far are more likely to order dessert. The decision to order a dessert comes towards the very end of the meal. If you haven't enjoyed your food so far, or you don't find the service and atmosphere pleasant, or you feel like the meal has been overpriced, dessert doesn't sound like a great idea. And the converse is also true. In fact, I would posit that "would you like to order dessert?" is a pretty good indicator of overall satisfaction. I wonder if any restaurants keep track of this.

The people more likely to eat and rave over desserts are the morbidly obese and emotionally fragile who are elated at every opportunity to feed the holes in their heads.

I'm a morbidly obese diner who seconds AShepard's comments. Ordering dessert is never a given.

The price of the Kindle edition is $23.41 when Amazon (.com each time) thinks you're in the UK, $13.88 when it thinks you are in Australia, and $12.99 for Japan.

Do Netflix / IMDB movie reviews that mention the ending / climax give a higher rating?

The value of reviews is not a review but all of them. The composite score, the mix of positive and negative ratings, the consistency of comments. Reviews are a wisdom of the crowd mechanism.

And since the cost of poor reviews is only the price of a meal and transportation, some indigestion and frustration, I will hazard the few crowds with perverse tastes. Besides, I can usually spot over-enthusiastic fanboyz.

Yes, this is rampant mood affiliation and cognitive dissonance on the part of Tyler. "How can other people not like the best taco restaurant in the entire world that happens to be in my back yard attached to a gas station? Must be that ratings are totally unreliable, I could never be wrong."

As @charlie said:

"If you are reading Yelp for the food reviews you are doing it wrong."

Maybe it is just the restaurants that I go to or the area where I live, but I find dessert to typically be the weakest course at most restaurants. For example, home prepared pie made from scratch with fresh ingredients is nearly always tastier than the day old refrigerator pies and cakes that most restaurants sell. The highest end restaurants might be producing really good stuff, but they also are certainly not the modal restaurant type being reviewed on Yelp.

In America I think dessert has been supplanted from restaurants into a separate industry of its own. Consequently most regular restaurants don't give dessert as much attention.

If you are reading Yelp for the food reviews you are doing it wrong.

"How can you resist a book with sentences such as these?"


The function of yelp has changed over the years, roughly parallel to the changes in Facebook after it allowed non-college members.

Originally, yelp consisted of a relatively homogenous group of young urban people that actually liked cool/good/swpl restaurants and had a decent sensitivity to quality. Now that the general public writes most of the reviews, things like (a) how big the portions sizes are (b) whether or not they ended up picking a fight with the waitstaff and (c) how cheap the bottomless breadsticks deal is have become controlling factors.

This is why the commenters above are right that you can now only look at the aggregated number of stars and infer that a place is "probably horrible" or "probably not horrible."


It's a simple application of the economic theory of clubs. In its earlier days, yelp was the optimal size to produce a public good for urban foodies. Now, it's too big and is no longer very useful for that group.

"Business networking" consultants recommend meeting new "networking prospects" for coffee. Each person get buzzed on the coffee and thinks, "Wow, this is a great person I just met!" Actually they are just mildly euphoric from the coffee, but that's okay--it will motivate them to do business together! It's the same with sugar--a known drug-like substance (and a most unhealthy one!) A sugar addict will be feeling mighty fine after a good dessert, and rate the restaurant accordingly. Also it has been shown the beginning and ending of an experience make the strongest emotional impression, so a sugar addict who enjoys a good dessert at the end of a meal is almost guaranteed to give the restaurant a rave review.

Comments for this post are closed