Relative Sizes of Inexact Quantitative Measures

A research proposal.


There are a great many inexact and colloquial units of measurement used in everyday life by native English speakers. Often they are used in situations where their meaning can only be metaphorical—“a handful of people turned up”, “I have a truckload of work to do next week”, “I was stuck in traffic for an eternity”.

This research project would aim to answer a number of fundamental questions about such inexact measurements:

  1. Are the relative sizes of the measurements universally understood between native English speakers? For example, does everyone have roughly the same idea of how much “a hell of a lot” is, compared to (say) “an obscene amount”?

  2. Do they have a well-ordered ranking? That is, if unit A is generally deemed larger than unit B, and unit B is larger than unit C, is it guaranteed that unit A is generally felt to be larger than unit C?

  3. Is there a definite deep semantic content to inexact quantities, such that they can be understood without conscious reasoning? Is this what gives them their literary expressive power?


The project would be divided into four phases, outlined below.


The initial phase would consist of collecting an extensive list of inexact or colloquial quantitative measures. The principal criterion for inclusion on the list would be that the measurement in question is ill-defined; hence the list could include “miles”, in the sense of “the bus stop is bloody miles away”.

This initial research phase would probably take several months.


Next, the units would be categorized according to their dimensionality. Unlike in exact unit systems, colloquial units can have multiple values for their dimension. For example, a “handful” can be either a dimensionless quantity (“a handful of people showed up”), a unit of currency (“a handful of spare change in my pocket”), a unit of volume (“a handful of rice to throw at the bride and groom”), or a unit of weight (“a handful of potato chips”).

This phase would also be quite time-consuming; it would take a bunch of months. During this time, software development for phase three would begin.


A group of randomly chosen test subjects, all native English speakers, would be asked to answer computer-administered questions, presented via the web in the following form:

    <input type="radio" name="x" value="less" checked> is less than<br /><input type="radio" name="x" value="greater"> is greater than
    Unit B
Unit A

The central choice could be made using either a drop-down box or a radio button.

The subjects would thus be presented with single pairs of units in isolation, chosen from the list, and asked to indicate their immediate unthinking ranking of the pair of items. Only inexact units with compatible dimensionality would be chosen for comparison.

This method is superior to the obvious approach of simply asking subjects to rank units in a list for two reasons. Firstly, it allows us to investigate the second question above (whether the units are well-ordered), without assuming an answer. Secondly, it allows the subjects to record their instinctive rather than reasoned response, and makes the questioning process less of an intellectual process.

To put it another way: Most people would have trouble ranking a set of ten approximate units they were given in a list. However, almost anyone can say which of “a heck of a lot” and “a shitload” is the largest unit, without having to think very hard at all. Remember, the intent here is to record the deep semantics of the units, the meaning we are not conscious of.


Statistical analysis would be performed on the collected data, and appropriate conclusions drawn.

Followup Work

There is clearly potential for further research along these lines, once the initial work has been done. The effect of modifying adjectives on the rankings of inexact measures could be evaluated, or the adjectives themselves possibly ranked in power—if “forever” is shorter than “an eternity”, is “fucking forever” still shorter than “a bleeding eternity”?

Adjectives and modifying phrases could also be measured, using similar methodology. This would allow us to answer questions such as: is ”complete and utter fuckwit of the first order” more or less insulting than “24 karat solid gold fuckwit”?


Funding for this project could be modest. It wouldn’t need screaming fast computer equipment or cavernous laboratory space. Most of the budget for the early phases would be spent on aids to dissociative cogitation, probably consumed in a public house or similar venue.

Value Of The Research

Beyond intellectual curiosity, this research has the potential to yield a great steaming crapload of valuable data. Thesaurus compilers could include numerical indications of the size of inexact quantitative measures, allowing people to choose (say) a volumetric metaphor of appropriate capacity when describing how much they drank the previous night. Dictionaries and textbooks could carry tables of inexact units ranked in order, to assist non-native speakers in learning the “deep meaning” quantity indicated by each phrase. The information might also be trouser-wettingly valuable to researchers in natural language comprehension in the fields of artificial intelligence and computational linguistics.