In our everyday conversations we tend to speak, hear, read, and write stuff like “ladies and gentlemen”, “bread and butter”, or “coffee or tea”, without giving much thought to the mechanisms that make these sequences of words palatable to the ear or eyes. This pattern of words transpires down to order of names, as in the Shakespeare plays like “Romeo and Juliet”, or game characters like Mario and Luigi. Occasionally, one does ponder why the reverse order of many of these sequences sound weird yet grammatical; if you said “gentlemen and ladies”, or “span and spick”, you’d most likely be corrected with the socio-linguistically “correct” order. I had been thinking about this feature for quite some time, and it is time to write a short piece on some of the linguistic workings behind these expressions.

Linguistically speaking, the presentation of such sequences of words is termed as “binomials”, or “frozen binomials”, where nouns or nominalisations are linked by a conjunction like “and” or “or”. These noun pairs are usually of the same word class. Sometimes these word orders are subject to various linguistic constraints, thereby impeding the reversibility of the binomial. For example, it is acceptable for the binomial “day and night” to be reversed, although the given context may be associated with the reversal, as with “I worked day and night to get this piece of assignment done”, compared to “the difference between this piece and that one is like night and day”. The reversibility of binomials, in English at least, has been a topic of research and debate for a rather long while, with many pieces of literature published detailing the various constraints which may account for why such binomials are presented as they are in today’s sociolinguistic context. I don’t want to make it too much like a dissertation nor an epiphany, but more rather, to break this down to a simpler, hopefully more concise account.

The paper written by Sandra Mollin in 2012 has put forward several ways how such binomials form, and how some leeway in reversibility may be granted in some cases. These largely include semantics, phonology and word frequencies, although some factors are more explicitly observed in some binomials than others. The paper also covers the history of rationalising this pattern of binomials, where different angles are proposed, from morphology, to semantics and phonetics.

One of the first works covering frozen binomials was done in the early 20th century, with das Gesetz der wachsenden Glieder (The law of increasing constituents), written by Otto Behaghel in 1909. Here, he proposed that the shorter words tend to be placed first in the binomial, using a German corpus to substantiate this, and finding 72 events that follow this rule, and 7 exceptions.

This led to the 1950 paper by Richard D. Abraham, who suggested that syllable count and semantics are important factors driving frozen binomials. This included semantic stuff like:

• Desirable before undesirable
• Important before less important
• Light before dark
• Masculine before feminine
• Positive before negative
• Principal before subsidiary
• Larger before smaller
• Near before far
• Top before bottom
• Present before future

Yakov Malkiel also suggested word length might be a factor as well, although it does not seem to follow Abraham’s definition of word length by number of syllables. Additionally, the semantic and pragmatic factors he suggested included those of time, those of society, and those of relative strength of two polar characteristics (like larger before smaller). In essence, the semantic and pragmatic factors proposed by Malkiel largely grouped Abraham’s proposed semantic factors into three main themes.

These three semantic themes were coalesced into a single general rule in a publication by William E. Cooper and John R. Ross in 1975, termed the ‘Me First’ principle. In this principle, the first element of the irreversible binomial contains factors that would describe the prototypical speaker, or ‘Me’, though in the anglosphere, that pretty much means the present, adult, men, positive, animate, agentive, living, friendly, and patriotic elements usually precede the other word. But there are also phonetic factors proposed by the authors, organised into 7 rules organised into a single hierarchy. In an ‘A and B’ setting, this hierarchy is presented as such:

• A has fewer syllables than B
• A has a shorter vowel than B
• A has fewer initial consonants than B
• A has a more obstruent consonant than B, if the two words have one initial consonant
• A has a vowel which is more front than that in B
• A has fewer final consonants than B
• A has a less obstruent final consonant than B, if the two words have one final consonant

This introduced a new paradigm of the structure of irreversible binomials, and it would continue to be worked on for the later 20th century. John R. Ross, for example, proposed in a 1980 publication that A has more final consonants than B, and suggesting additional factors in the vowel space that could affect priority, rather than the frontness or backness of a vowel. The late 20th century also introduced more quantitative studies to test these theories, using data from lists or corpus texts. However, the earlier ones have been criticised for only assessing frozen binomials, rather than using corpus texts that could help discern if a binomial truly is irreversible. This sampling also prevented any comparison made for binomials that may be reversible, which could also have yielded interesting findings.

In Mollin’s 2012 publication, she conducted an empirical study on these binomials, assessing them for compliance with four broad themes of constraints — semantic, metrical, non-metrical but phonological, and relative word frequency. This drew on the factors proposed by previous studies conducted over the past century, and adding or modifying certain factors such as “avoidance of lapse”, “avoidance of ultimate stress”, “iconicity”, “perceptual markedness”, and “formal markedness”. Her data were from the British National Corpus, which contained 100 million words.

From this corpus, Mollin found that semantics best predicted the order of binomials, with iconicity correctly predicting 95.38% of the orders of binomials in a corpus. This was followed by metrical factors, and lastly, came the phonological ones. Word frequency came in the middle of the pack though. This led Mollin to suggest the following hierarchy, with the top one taking priority:

• Iconicity, markedness
• Power
• Syllable count, avoidance of lapse and ultimate stress, word frequency
• Syllable weight, vowel length, syllable openness, vowel backness, final and initial consonants
• Vowel height, initial and final sonority

How this hierarchy works is, given two elements that make up a binomial, if they can be ordered on iconicity or markedness, the order of binomials will follow that order. Otherwise, some factor lower down on the hierarchy could be used to predict the order, and so on. Additionally, Mollin also found that elements that are ordered based on power are more likely to make up more irreversible binomials.

Mollin also suggested the mechanism in which binomials may become increasingly irreversible. Given a preference for a binomial like ‘A and B’ instead of ‘B and A’, if the former is encountered more frequently than the latter, the more likely the former is internalised by speakers and processed as a single unit of a binomial. This kicks off a positive cycle of internalisation, wherein a binomial spreads within a speaking population, until speakers only process and parse the binomial ‘A and B’. This end state could be interpreted as a binomial being ‘completely frozen’.

There are some exceptions that this hierarchy could not explain, such as ‘back and forth’. Here, the elements of the binomial violate the semantic factors in the hierarchy, but satisfies phonological factors instead. Despite violating the semantic factors underlying the order of elements that make up the binomial, ‘back and forth’ is said to be irreversible, as ‘forth and back’ would sound extremely weird to English speakers. Other binomials this hierarchy could not explain include ‘City and Guilds’ and ‘postage and packing’.

Further Reading

Abraham, R. D. (1950) ‘Fixed order of coordinates: A study in comparative lexicography’, Modern Language Journal, 34, pp. 276-287.

Behaghel, O. (1909) ‘Beziehungen zwischen Umfang und Reihenfolge von Satzgliedern’, Indogermanische Forschungen, 25, pp. 110-142.

Benor, S. B. & Levy, R. (2006) ‘The chick or the egg? A probabilistic analysis of English binomials’, Language, 82, pp. 233-277.

Cooper, W. E. & Ross, J. R. (1975) ‘World order’, Papers from the parasession on functionalism, pp. 63-111.

Fenk-Oczlon, G. (1989) ‘Word frequency and word order in freezes’, Linguistics, 27, pp. 517-556.

Malkiel, Y. (1959) ‘Studies in irreversible binomials’, Lingua, 8, pp. 113-160.

Mollin, S. (2012) ‘Revisiting binomial order in English: ordering constraints and reversibility’, English Language and Linguistics, 16(1), pp. 81-103.

Ross, J. R. (1980) ‘Ikonismus in der Phraseologie. Der Ton macht die Bedeutung’, Semiotik, 2, pp. 39-56.