What is the rule for adjective order?

I am re­mind­ed of how J.R.R. Tol­kien’s moth­er once fa­mous­ly cor­rect­ed him at a very ear­ly age when he said ‘a green great drag­on’. She told him that it had to be ‘a great green drag­on’, but when he asked her why, she couldn’t an­swer, there­by start­ing him down the road of puz­zling over mat­ters of philology (linguistics) his whole life long.

This top­ic is one of con­tin­u­ing re­search. Sim­ply goog­ling for ‘ad­jec­tive or­der­ing re­stric­tions’ (AOR) or ‘ad­jec­tive hi­er­ar­chy’ can un­cov­er some fas­ci­nat­ing re­search in this area.

In her 2006 pa­per on “Ad­jec­tive Order­ing Re­stric­tions Re­vis­it­ed” on pp 309–407 of the Pro­ceed­ings of the 25ᵗʰ West Coast Con­fer­ence on For­mal Lin­guis­tics, Alex­an­dra Te­o­dor­es­cu writes:

Ad­jec­tive or­der­ing re­stric­tions (AOR) have been wide­ly dis­cussed, but they are still not very well un­der­stood. For ex­am­ple, in lan­guages like English pre­nom­i­nal ad­jec­tives are strict­ly or­dered.

…

For ex­am­ple, ad­jec­tives that de­note qual­i­ty have been ar­gued to pre­cede ad­jec­tives con­vey­ing size, which in turn pre­cede ad­jec­tives con­vey­ing shape, and so on, in all lan­guages (5). Sim­i­lar claims have been made for oth­er ad­jec­tive types, and the re­spec­tive or­der­ing re­stric­tions are giv­en in (6).

• (5) Qual­i­ty > Size > Shape > Color > Prov­e­nance [Sproat and Shih (1991)]

• (6) a. Posses­sive > Speak­er-ori­ent­ed > Sub­ject-ori­ent­ed >Man­ner/The­mat­ic [Cinque (1994)]

•        b. Value > Di­men­sion > Phys­i­cal prop­er­ty > Speed > Hu­man Propen­si­ty > Age > Color [Dixon (1982)]

See Teodor­es­cu’s bib­li­og­ra­phy to chase down re­lat­ed work. You should al­so look for pa­pers that cite hers (Google Schol­ar finds 26 such ci­ta­tions to her work), like Lu­cas Cham­pi­on’s 2006 pa­per on “A Game-The­o­ret­ic Ac­count of Ad­jec­tive Order­ing Restric­tions”, which starts off with the Tol­kien ex­am­ple.

Build­ing then on Cham­pi­on’s work is this English-lan­guage pa­per by An­to­nia An­drout­so­pou­lou, Ma­nuel Es­pañol-Eche­va­rría, and Phil­ippe Pré­vost en­ti­tled “On the Ac­qui­si­tion of the Prenom­i­nal Place­ment of Eval­u­a­tive Ad­jec­tives in L2 Spanish”, from the 10ᵗʰ His­pan­ic Lin­guis­tics Sym­po­si­um in 2008. This one is in­ter­est­ing be­cause it looks at how sec­ond-lan­guage learn­ers ac­quire an un­der­stand­ing of ad­jec­tive or­der­ing when learn­ing a new lan­guage:

In this pa­per, we fur­ther in­ves­ti­gate knowl­edge of ad­jec­ti­val or­der­ing re­stric­tions in for­eign lan­guage learn­ing, by fo­cus­ing on L2 ac­qui­si­tion of eval­u­a­tive ad­jec­tives (EAs) in Span­ish by French learn­ers.

The most re­cent pro­fes­sion­al pub­li­ca­tion I could find on this is­sue is Katy Mc­Kin­ney-Bock­’s 2010 pa­per on “Ad­jec­tive Class­es and Syn­tac­tic Or­der­ing Re­stric­tions”, in which she writes:

There is a lack of con­sen­sus in the lit­er­a­ture as to which clas­si­fi­ca­tion of ad­jec­tives is di­rect­ly rel­e­vant for the ob­served syn­tac­tic re­stric­tions on their or­der­ing. In this pa­per, I ar­gue that ad­jec­tives are di­vid­ed in­to four class­es of rel­e­vance for syn­tac­tic or­der­ing. I pro­pose that ad­jec­tive or­der­ing re­stric­tions (AOR) are the re­sult of ad­jec­ti­val con­stit­u­ents rais­ing or not rais­ing in the struc­ture as a con­se­quence of their com­plex­i­ty, rather than stip­u­lat­ing that se­man­tic prop­er­ties cor­re­late to syn­tac­tic heads.

and whose ex­tend­ed ab­stract reads:

I ar­gue there are four class­es of ad­jec­tives rel­e­vant to syn­tac­tic or­der­ing: pred­ica­tive/in­ter­sec­tive, pred­ica­tive/non-in­ter­sec­tive, non-pred­ica­tive, clas­si­fy­ing (Sven­on­i­us 2008, Al­ex­i­a­dou et al 2007), and pre­vi­ous pro­pos­als have not iden­ti­fied the rel­e­vant se­man­tic di­men­sions. Among the prop­er­ties of grad­abil­i­ty, mass/count, and in­ter­sec­tiv­i­ty, on­ly in­ter­sec­tiv­i­ty is syn­tac­ti­cal­ly rel­e­vant. The four class­es of ad­jec­tives are mo­ti­vat­ed by the dis­tri­bu­tion of or­dered/non-or­dered ad­jec­tives, scope ef­fects with cer­tain ad­jec­tive-pairs, PP-mod­i­fi­ca­tion, N-drop­ping and com­par­a­tives (Bouchard 2002, Hig­gin­both­am 1985, Ken­nedy 1999). DP struc­ture in­volves 1) merg­ing the clas­si­fy­ing ad­jec­tive with pro­nounced N, 2) merg­ing in­ter­sec­tive ad­jec­tives with N, 3) merg­ing non-in­ter­sec­tive ad­jec­tives with a silent copy of N.

Fi­nal­ly, if you’re look­ing for some­thing slight­ly less pro­fes­sion­al — or at least, less aca­dem­ic — then in this blog post­ing, the wri­ter pos­its an or­der­ing of:

• eval­u­a­tion
• size
• shape
• con­di­tion
• hu­man pro­pen­si­ty
• age
• col­or
• ori­gin
• ma­te­ri­al
• at­trib­u­tive noun

And sum­ma­rizes with:

If there’s def­i­nite­ly a mean­ing dif­fer­ence be­tween dif­fer­ent ad­jec­tive or­der­ings, let that de­ter­mine how you or­der them, and don’t use com­mas. If you can’t find a mean­ing dif­fer­ence, don’t go try­ing to force there to be one. In­stead, go by the ad­jec­tive-or­der­ing hi­er­ar­chy, and don’t use com­mas. If more than one ad­jec­tive has the same kind of mean­ing in the hi­er­ar­chy, then use com­mas, or ands or buts if the ad­jec­tives have con­tras­tive mean­ings.

There’s a lot more out there on this top­ic.