Language Log " Big bad modifier order

Language Log

Home

About

Comments policy

Big bad modifier order

September 4, 2016 @ 8:08 am<br>· Filed by Mark Liberman under Grammar, Logic, Rhetoric

" Days of the week in Nagoya | Annals of literary vs. vernacular, part 2 "<br>--><br>" previous post | next post "

Things native English speakers know, but don't know we know: pic.twitter.com/Ex0Ui9oBSL

— Matthew Anderson (@MattAndersonNYT) September 3, 2016

This is a quote from Mark Forsyth's book The Elements of Eloquence: Secrets of the Perfect Turn of Phrase. And Nicholas Feinberg asks

This claim seems iffy to me, but it's interesting – have you heard of this before? Do you know of anything related that I could read, or anyone else I should ask?

Mark Forsyth's book starts its discussion of this topic with an odd claim about the English language, and an interesting anecdote about Tolkien:

Hyperbaton is when you put words in an odd order, which is very, very difficult to do in English. Given that almost everything else in the English language is slapdash, happy-go-lucky, care-may-the-Devil, word order is surprisingly strict. John Ronald Reuel Tolkien wrote his first story aged seven. It was about a "green great dragon." He showed it to his mother who told him that you absolutely couldn't have a green great dragon, and that it had to be a great green one instead. Tolkien was so disheartened that he never wrote another story for years.

The reason for Tolkien's mistake, since you ask, is that adjectives in English absolutely have to be in this order: opinion-size-age-shape-colour-origin-material-purpose Noun. So you can have a lovely little old rectangular green French silver whittling knife. But if you mess with that word order in the slightest you'll sound like a maniac. It's an odd thing that every English speaker uses that list, but almost none of us could write it out. And as size comes before colour, green great dragons can't exist.

The bit about green great dragons is correct, I think, and it's also correct that there are often strong preferences for prenominal modifier order. But it's not so easy to characterize the preferences — "big bad wolf" is right and "bad big wolf" isn't, but that seems to be size-opinion rather than opinion-size.

And many cases work both ways — COCA has:

1+2<br>2+1

1=big

2=beautiful<br>43 (73%)<br>16 (27%)

1=little

2=beautiful<br>9 (3%)<br>345 (97%)

1=big

2=ugly<br>69 (95%)<br>4 (5%)

1=little

2=ugly<br>25 (18%)<br>116 (82%)

According to Forsyth (I think), those all should be opinion-size, i.e. 2+1 as in the second column — but all of them are mixed, and some go pretty strongly the other way.

Furthermore, there are some strong modifier-order preferences about which Forsyth's scale says nothing:

1+2<br>2+1

1=long

2=tall<br>18 (90%)<br>2 (10%)

1=big

2=huge<br>54 (73%)<br>20 (27%)

1=big

2=enormous<br>2 (25%)<br>6 (75%)

And some of these seem to have changed over time:

There's certainly a literature on such things — you could take a look at Richard Sproat and Chi-lin Shih, "The Cross-Linguistic Distribution of Adjective Ordering Restrictions", except for the fact that Springer wants to charge you a bit more than a dollar a page to read it. More cheaply, you could look through some of the 264 works that cite that chapter, including e.g. Alexandra Teodorescu's  "Adjective Ordering Restrictions Revisited". [Update — or you can find the Sproat and Shih article on the LLOG web site…]

Another interesting place to look would be Stefanie Wulff, "A multifactorial corpus analysis of adjective order in English", International Journal of Corpus Linguistics, 2003:

This paper is concerned with the question of which factors govern prenominal adjective order (AO) in English. In particular, the analysis aims to overcome shortfalls of previous analyses by, firstly, adopting a multifactorial approach integrating all variables postulated in the literature, thereby doing justice to the well-established fact that cognitive and psychological processes are multivariate and complex. Secondly, the phenomenon is investigated on the basis of a large corpus, rendering the results obtained more representative and valid of naturally occurring language than those of previous studies. To this end, corpus-linguistic operationalizations of phonological, syntactic, semantic and pragmatic determinants of AO are devised and entered into a Linear Discriminant Analysis, which determines the relative influence of all variables (semantic variables being most important) and yields a classification accuracy of 78%. Moreover, by means of the operationalizations developed in this analysis, the ordering of yet unanalyzed adjective strings can be predicted with about equal accuracy (73.5%).

The fact that her algorithm was only able to predict the ordering of two adjectives 73.5% of the time — where chance would be 50% — suggests that all of the relevant factors are not yet accounted for. Presumably Forsyth's template would do quite a bit worse, though of...