Left dislocation
« previous post | next post »
A couple of days ago, Jim Bisso sent me a question:
I've been embattled by a bunch of peevologists over the grammaticality of sentences of the sort: "my mother(,) she is a good person". I have pointed out that many kinds of apposition are not only acceptable but flow from the pens of some of our finest writers, but they are having none of that. Somehow a construction like "we the people of the United States etc." is okay, but reverse the order of Pron + NP to NP + Pron and sparks start to fly. I say it's simply a stylistic matter and not a syntactic one, but who am I? What say you? (Do you know any monographs that I might delve into to fuel my argument?)
Executive summary: This construction goes back to Old English, and is still widely used in spoken English and in some regional varieties ; but its use in formal written English has been decreasing since about 1500, and is now either informal or archaic.
If I understand what Jim has in mind, the sentences at issue are not really appositives. In nominal appositives, a noun phrase is followed by another noun phrase that provides additional information about it, so to speak in parallel. Appositives are common in news writing — the lead sentence of the first article I picked in this morning NYT has two of them (indicated in boldface):
Warren E. Buffett, the country’s most famous investor and one of the world’s richest men, announced on Tuesday that he would invest $5 billion in Goldman Sachs, the embattled Wall Street titan, in a move that could bolster confidence in the financial markets.
Appositives used after third-person pronouns are usually editorial interpolations to clarify the reference ("They (the defense) got to the line and showed their hand"), since there's otherwise little reason not to use the appositive phrase in place of the pronoun. With first and second person pronouns — as in "we the people" — the situation is obviously different.
Jim's sentence ("my mother, she is a good person") is an example of a different construction, where a non-vocative noun phrase in initial position is set off from a following sentence that contains one or more pronouns coreferential with the initial NP. Since Haj Ross's 1967 dissertation Constraints on Variables in Syntax, this construction has been known as "left dislocation". Haj's examples (pp. 422-451) included:
The man my father works with in Boston, he's going to tell the police that the traffic expert has set that traffic light on the corner of Murk Street far too low.
My father, he's Armenian, and my mother, she's Greek.
My wife, somebody stole her handbag last night.
When the pronoun is sentence-initial, you might think that it's in apposition with the left-dislocated noun phrase. But this generally makes neither semantic nor prosodic sense. If the structure were really
My father – he – 's Armenian.
the phrasing would be different (and strange), and it's hard to see why one would want to add the pronoun, which would add none of the parallel information that appositives usually do.
And the third example ("My wife, somebody stole her handbag last night") illustrates the fact that the pronoun need not be adjacent to the left-dislocated noun at all. (Though examples with the pronoun in subject position are by far the commonest.)
Constructions of this general type are common across the languages of the world, and in so-called topic-prominent languages, they're the norm. The left-dislocation structure is often said to divide the sentence into topic and comment, or some similar sort of articulation of information. (This works when the initial item is a full noun phrase, referenced in the following sentence by a pronoun or pronouns, but it wouldn't work the other way around. There's a lot more to said about the pragmatics of LD, and there's a large literature discussing it, but I'll leave it there for now.)
Elizabeth Traugott ("Old English left-dislocations: Their structure and information status", Folia Linguistica. 41(3-4): 405–441, 2007) gives examples like this from Old English:
Se awyrigeda deofol syððan he ðone frumsceapenan man beswac.
syððan he hæfde anweald ofer ungelyfedum mannum.
that cursed devil after he that first-created man deceived
afterwards he had control over unbelieving men
In a corpus search, she found "a total of 280 left-dislocated main clauses, out of 39,654 main clauses (0.7%), and a total of 53 left-dislocated subordinate clauses out of a total of 34,866 subordinate clauses (0.15%)".
This rate appears to decrease through the history of (written) English. Javier Pérez Guerra & David Tizón-Couto, "‘These hands, they are apt enough to dislocate and tear thy flesh’: On Left Dislocation in the Recent History of the English Language", 2004, give counts of left-dislocations per 1,000 words in a corpus study of late Middle English, early Modern English (16th-17th century), late Modern English (18th century), and Present-Day English (20th century):
lME | eModE | lModE | PDE |
1.19 | 0.39 | 0.14 | 0.10 |
Unfortunately, Traugott doesn't give word-count information for the material she searched, but she does list the works, and by reference to the original corpus documentation I tentatively estimate the total as about 458,000 words, which would yield a left-dislocation rate of about 0.72 per 1,000 words. Whether or not this is accurate, it seems likely that the LD rate in Old English was higher than the historical estimates for texts since 1500.
In the earlier materials, left dislocation is found in formal prose as well as in less formal contexts such as letters. In fact, Pérez Guerra and Tizón-Couto argue that "LD cannot be characterised as a strategy applying to exclusively informal language since most of the corpus examples are found in instructive and narrative text types, which clearly instantiate formal English". This argument is not as compelling for their present-day English sample, which comes from the LOB corpus, where the narrative texts include several genres that are likely to be written in a quasi-conversational style, even leaving aside the high proportion of dialogue (mysteries, science fiction, westerns, romance).
Within the past century or so, LD has clearly remained a lively part of spoken English. According to Michelle Gregory and Laura Michaelis ("Topicalization and left-dislocation: a functional opposition revisited", J. Pragmatics 33(11): 1665-1706, 2001), searching part of the Switchboard corpus, found 187 instances of LD in 32,805 statements and questions, for a rate of 0.6%, which is not very different from the rate of 0.7% that Traugott found for LD in main clauses in Old English.
They give examples like
The Saturns, you can get air bags in them.
And heavy metal, it's noisy.
Well, my car, it's an eighty six.
However, the use of LD in formal written English has become quite rare over the past century or so. Even in poetry, examples tend to occur in consciously archaic writing, e.g. Walter de la Mare, "The Cage":
Fret now no more; be still. Those steadfast eyes,
Those folded hands, they cannot set you free;
Only with beauty wake wild memories—
Sorrow for where you are, for where you would be.
Or in regional varieties of English, e.g. Hugh McDiarmid, "To Circumjack Cencrastus" (1930):
My love she is the hardest thocht
That ony brain can ha'e,
And there is nocht worth ha'en in life
That doesna lead her way.
Or in quoted speech, e.g. Robinson Jeffers, "The Loving Shepherdess" (1928):
Fern lagged and lagged,
Dibbling the dust with the mere points of the hoof
Of the hurt fore-leg, and rolling up to her shepherdess
The ache of reproachful eyes. "Oh Fern, Oh Fern,
What can I do? I'm not a man, to be able to carry you.
My father, he could have carried you." Tears from Clare's eyes
Fell in the roadway; she was always either joyful or weeping.
Examples in contemporary news text are almost always in quotes or in informal passages of opinion pieces. Even in the 18th century, my impression is that LD is commoner in e.g. first-person narratives than in more formal expository prose. Thus Jane Barker, Exilius (1719):
Valerius, though a little Opposite at first, yet, upon his Mother’s pressing, and repeating how far my Happiness was the Object, if not the whole End of the Undertaking, he at last consented, and this my forced Marriage was resolv'd on that coming Day.
And I just read 400 sentences of Gibbon's Decline and Fall without finding any examples of LD — this suggests that he used this construction at a rate lower than the 1 in 175 that Gregory and Michaelis found for contemporary American English conversational speech.
So I'd answer Jim's question as follows. Left dislocation is certainly grammatical in English. Up to 1500 or so, roughly one in every 100 or 200 sentences had this form, even in formal writing, and a similar frequency of use continues in spoken English to this day. Over the past few centuries, the frequency of this construction in standard written English has been declining, and it's now quite rare except in archaic styles, in representations of speech, or in informal styles that use spoken-language norms. Therefore Jim's claim that the issue is one of style rather than grammaticality is correct, although his identification of LD as a form of apposition is misleading, as is his citation of "we the people …" as a relevant example.
Maria said,
September 24, 2008 @ 7:41 am
So left dislocation is not as common as it used to be. Can we say that the times they are a-changing?
bfwebster said,
September 24, 2008 @ 7:45 am
Posts like this, they're why I enjoy Language Log so much. ..bruce..
Mark Liberman said,
September 24, 2008 @ 7:59 am
Maria: Can we say that the times they are a-changing?
Nice.
Actually, LD seems to be especially common in ballads. This is partly because it's a traditional part of the style, but also, I think, because it helps with the meter.
Benjamin Zimmer said,
September 24, 2008 @ 8:11 am
And then there's the old Peggy Lee song, Mañana (Is Soon
Enough for Me)…
The faucet she is dripping and the fence she's fallin' down…
The car she needs a motor so I can't go anywhere…
The window she is broken and the rain is comin' in…
(Or was that supposed to mimic a stereotypical feature of native Spanish speakers?)
Peter said,
September 24, 2008 @ 8:13 am
I'm not sure I follow your reasoning here. The presence of these LD constructions in spoken English and in some written forms of English surely mean that English is a topic-prominent language. If English were classified as a topic-prominent language, then these constructions would be grammatical, a fortiori. On what basis, then, is English not classified as a topic-prominent language?
Mark Liberman said,
September 24, 2008 @ 8:22 am
Peter: On what basis … is English not classified as a topic-prominent language?
Jeannette K. Gundel and Thorstein Fretheim, "Topic and Focus", chapter 8 (p. 175-) in L. Horn and G. Ward, Eds. The Handbook of Pragmatics, 2004, put it this way:
These structures exist in English, but are marked even in speech (around 1% of sentences or less). The implication of the taxonomic category is that there are languages where such structures are much commoner, though I don't know what the frequencies actually are.
Maria said,
September 24, 2008 @ 9:44 am
Benjamin: (Or was that supposed to mimic a stereotypical feature of native Spanish speakers?)
As a native speaker, but a non-linguist, I don't think it's a feature of Spanish. We tend, if anything, to omit subjects when their inclusion is not necessary – verb declinations do most of our work for us. So I'd put my money on Mark's explanation (style + meter).
John Cowan said,
September 24, 2008 @ 10:04 am
I think that we call a language topic-prominent when even what look like straightforward grammatical subjects are capable of being reinterpreted as non-subject topics: in Chinese (which is SVO like English) "Fish eat here on Sundays" can mean, depending on context, "Fish, [people] eat [them] here on Sundays", an entirely different sense. English can't do that: there always has to be a subject quite distinct from any dislocated topic (though of course in both English and Chinese the subject can be, and frequently is, the topic).
This leads naturally to the notion that left- and right-dislocations are part of a process whereby free word order languages can arise. In modern French, dislocation is so straightforward that we can have a series of sentences like these (respelled):
jdetèst mari (VO)
mari jladetèst (OV)
mwa jdetèst mari (SVO)
mari jladetèst mwa (OVS)
mwa mari jladetèst (SOV)
mari mwa jladetèst (OSV)
jladetèst mwa mari (VSO)
jdetèst mari mwa (VOS)
all plausible in slightly different contexts, and all meaning "I loathe Mary" and also:
jladetèst (V)
jladetèst mwa (VS)
mwa jladetèst (SV)
likewise all plausible, and all meaning "I loathe her". You can't get much freer word order than that: S, V, and O in any order, S and O both optional, with a required subject-agreement prefix on the verb followed by a object-agreement prefix that is omitted when the object directly follows the verb.
John Atkinson said,
September 24, 2008 @ 10:45 am
O course, as Maria says, the kind of LD used in that song of Peggy Lee's isn't a feature of Spanish itself. The question was, is it a feature of broken English as spoken by native Spanish-speakers? Whether this is actually the case or not, it's certainly typical of the way authors,actors, and comedians represent the English of southern Europeans (Italians and Greeks as well as Hispanics).
It seems to me that the use of excess pronouns when speaking English (if it actually happens) is likely due to the fact that the Spanish speaker does usually omit subject pronouns in Spanish, their function being taken up by declensional endings, which are there whether or not there is an explicit subject. That is, they automatically translate "co'men" (say) by "they eat" whether or not there is a noun subject present. If this is what's going on, I don't think it can be called "left dislocation" in their variety of English.
Dislocation, both left and right, _is_ very common in colloquial Erench, so much so that some reckon it's the unmarked form.
Laurent C said,
September 24, 2008 @ 10:46 am
John Cowan,
I agree this is typical of spoken French. As in Marcel Pagnol: Aubagne, c’est là où mon frère il est né.
However I don't agree with your classification (or with the statement htat subject and object are optional): "Je la déteste" for instance, that you spelled jladetèst, is an SOV construction. "Marie, je la déteste" (mari jladetèst) would rather be OSOV, with a repeated object.
Ralph Hickok said,
September 24, 2008 @ 10:48 am
My teachers inveighed against this construction from the first grade right through senior year in high school (ca 1946-1955). Maybe it was unusually common in Northeastern Wisconsin at that time?
Aaron Davies said,
September 24, 2008 @ 10:52 am
Is this part of that "French is polysynthetic" meme I run across every now and then?
Paul Frederick said,
September 24, 2008 @ 11:01 am
There is one situation where I don't know how to avoid using LD. Consider:
Nilima and me, our car has good mileage.
Without LD, I would have to say one of the following, all of which sound bad to me.
Nilima's and my car has good mileage.
Nilima and my car has good mileage.
Me and Nilima's car has good mileage.
Skullturf Q. Beavispants said,
September 24, 2008 @ 11:21 am
This may be a digression, but Paul Frederick's example there, it has long perplexed me.
Parallelism suggests "Nilima's and my car", though I agree with his intuition that that still sounds bad. Realistically, in a somewhat formal context, I would simply rewrite, and go with something like "The car that Nilima and I share".
Karen said,
September 24, 2008 @ 11:45 am
I think I'd say "My and Nilima's car", myself.
Benjamin Zimmer said,
September 24, 2008 @ 11:46 am
John Atkinson: O course, as Maria says, the kind of LD used in that song of Peggy Lee's isn't a feature of Spanish itself. The question was, is it a feature of broken English as spoken by native Spanish-speakers? Whether this is actually the case or not, it's certainly typical of the way authors,actors, and comedians represent the English of southern Europeans (Italians and Greeks as well as Hispanics).
Yes, I think that's right. I'm trying to recall dramatic/comedic examples, but all I can think of is this line from an old Firesign Theatre sketch ("Temporarily Humbolt County," from the 1968 album "Waiting for the Electrician or Someone Like Him"):
Steve said,
September 24, 2008 @ 11:50 am
I'd say 'Mine and Nilima's car. . .' Maybe that's a Brit thing.
Left dislocation « ESOL World News said,
September 24, 2008 @ 12:51 pm
[…] I've been embattled by a bunch of peevologists over the grammaticality of sentences of the sort: "my mother(,) she is a good person". I have pointed out that many kinds of apposition are not only acceptable full story […]
rkillings said,
September 24, 2008 @ 2:04 pm
You define LD as non-vocative NP followed by a pronoun. Does the same analysis apply when the NP is replaced by what I know only as a disjunctive pronoun (but suspect that such thing may not be admitted to exist in English)? It seems to me that in speech if not in writing, constructions like "Me, I'm . . ." are not uncommon.
In writing, you will often see the initial pronoun punctuated with a question mark ("Them? They don't . . ."), but it's debatable whether this is a necessary or even a correct interpretation.
Scott said,
September 24, 2008 @ 2:18 pm
In my college Linguistics classes I've sometimes heard similar constructions referred to as "Yiddish fronting" (examples given are often along the lines of "My son, he's studying to be a doctor" – not to stereotype too much!). Do you all know anything about this, and is it the same thing?
Hannah said,
September 24, 2008 @ 2:20 pm
John Atkinson: Whether this is actually the case or not, it's certainly typical of the way authors,actors, and comedians represent the English of southern Europeans (Italians and Greeks as well as Hispanics).
The first example I thought of was Borat: "My wife, she die."
Benjamin Zimmer said,
September 24, 2008 @ 2:34 pm
Scott: This isn't quite the same as Yiddish fronting, which is exemplified by:
My brother-in-law he wants to be.
A German car I should buy?
(R.L. Trask, A Dictionary of Grammatical Terms in Linguistics)
Compare:
My son, he's studying to be a doctor. (LD)
A doctor, my son is studying to be. (YF)
Cephi said,
September 24, 2008 @ 3:29 pm
Peevologists — you gotta love 'em.
That this is not a version of apposition is also apparent from the fact that these constructions don't take nominative case. "Me, I'm voting for Kucinich" not "I, I'm voting for Kucinich". The latter sometimes pops up but is pretty clearly a different phenomenon.
Faith said,
September 24, 2008 @ 3:33 pm
What came to my mind is a certain historical kind of Canadian poetry called "habitant" poetry. It was written by Anglophones supposedly reproducing French-Canadian speech (in English). The one that was still in use when I was a school child was William Henry Drummond's "The Wreck of the Julie Plante," which begins:
On wan dark night on Lac St. Pierre,
De win' she blow, blow, blow,
and gets worse from there. (see full text http://rpo.library.utoronto.ca/poem/735.html). As far as I know, these poetic excrescences are no longer taught in schools, which is certainly good for the students' love of literature, but leaves them without the ability to turn to their partners in later life, on a rainy, windy, wuthering night, and say casually, "de win she blow on Lac St. Pierre"–a fine Canadian tradition.
dr pepper said,
September 24, 2008 @ 4:06 pm
I've mostly seen this as a device to indicate that the speaker normally speaks a romance language.
W.E.Howard said,
September 24, 2008 @ 5:04 pm
"The state, it is I" – Louis XIV(?)
"Me, I want a hula hoop." – Alvin
Trent said,
September 24, 2008 @ 5:29 pm
I'd construct the context in such a way that I could write "Our car …".
Peter said,
September 24, 2008 @ 5:35 pm
Thanks, Mark, for your reply (comment #6).
Mark Liberman said,
September 24, 2008 @ 5:42 pm
rkillings: Does the same analysis apply … [to] constructions like "Me, I'm . . ." … ?
Yes, sentences like that would be considered examples of LD.
Nathan Myers said,
September 24, 2008 @ 8:02 pm
Trent: So, "Nilima and me, our car, it gets good mileage"? Works for me. LD, it's fun, but double LD, it's funner.
Cephi said,
September 24, 2008 @ 9:04 pm
Ha yes, a much underappreciated strain of gardenpath sentences!
"The guy in the corner, his house, its architect, her car, it's red."
Dave said,
September 24, 2008 @ 10:18 pm
I've mostly heard LD's used as a rhetorical device to emphasize the subject of the sentence, sometimes to indicate that the speaker is introducing a new topic, but more often to allow the speaker to pause after the subject to set it off. "My mother, she'd never approve" highlights the subject in a way that "my mother would never approve" just doesn't.
Nathan Myers said,
September 25, 2008 @ 12:49 am
I guess nowadays it would be, "The guy in the corner? His house? The architect? Her car? It's full of blood."
What is the device in, "The blood of the martyrs? You're soaking in it" called?
Rachael said,
September 25, 2008 @ 5:51 am
LD is common in lolcat-speak. You often get sentences like "Skillz, I has dem."
Benjamin Massot said,
September 25, 2008 @ 7:45 am
I'll rely my comment on Lambrecht's work on Information Structure (IS) (1981, 1987, and especially 1994) (full references on this page: http://web.austin.utexas.edu/frit/utfrit/faculty_cv/lambrechtcv.pdf).
If I sum up him right, English LD doesn't have the property French LD has: it's not required when the subject is the topic of the sentence in English, whereas it is in French. In functionalist terms, English marks topical subjects by letting them unaccented, and non-topical ones by accenting them. Because the sentence prosody of French doesn't allow accented subjects (no, it does not), topical and non-topical subjects have to be marked by other means (syntactical ones):
a. LD (or RD) for topical subjects, b. have-clefts for sentences without any topic, c. it-clefts for focal subjects (and d. SV(O) for subjects with low topicality/background sentences):
1)a. Ma voiture elle est en panne.
b. J'ai ma voiture qu'est en panne.
c. C'est ma voiture qu'est en panne.
d. Ma voiture est en panne.
All these cases are achievable with English sentence prosody (Lambrecht's examples):
2)a. My car broke DOWN.
b. My CAR broke DOWN.
c. My CAR broke down.
d. My car broke DOWN. (ambiguous with a.)
English is subject prominent is the sense that IS doesn't have to trouble SV(O) word order. French is (maybe) topic prominent because IS always takes priority over argument structure (what I assume to guide SV(O), and what Lambrecht just calls syntax).
In this sense, English LD is a marginal syntactic device to mark topical subjects (if it really is its function).
Faldone said,
September 25, 2008 @ 8:06 am
Cephi: Ha yes, a much underappreciated strain of gardenpath sentences!
"The guy in the corner, his house, its architect, her car, it's red."
I wouldn't call this a gardenpath sentence since there's really no path to go down. It's more like a nested-door sentence. Open one door and there's another right behind it.
zmjezhd said,
September 25, 2008 @ 9:33 am
Thanks, Mark. It's nice to have a name to put to the phenomenon. It seems similar to what Jespersen calls extraposition in his Analytic Syntax. He treats it, along with apposition, as kinds of adjuncts.
Barry Nordin said,
September 25, 2008 @ 11:14 am
As Jerry Thomas says at Wordcraft.com, "The old gray mare, she ain't what she used to be . . . . etc."
David Waugh said,
September 25, 2008 @ 1:21 pm
In my own English (Scottish) I would be much more likely to say "She's a good person, my mother" and so on for most of the other examples given. On the face of it this seems to be the exact opposite, but really it's not that different. Is this kind of thing recognised? Does it have a name?
[(myl) Yes, that one is called "right dislocation".]
Trent said,
September 25, 2008 @ 9:17 pm
Here's a famous literary one:
"–I tell the tale that I heard told.
Mithridates, he died old."
Mark Liberman said,
September 26, 2008 @ 6:40 am
zmjezhd (Jim Bisso): It seems similar to what Jespersen calls extraposition in his Analytic Syntax.
Yes — the examples of "extraposition" that Jespersen gives (Chapter 12 of Analytic Syntax) include
But these days, the term "extraposition" is generally used to refer to a phenomenon that is different from LD in several ways: the extraposed material is shifted to the right, relative to its canonical location; and the extraposed material is usually not a noun phrase, but rather some sort of clause.
When a relative clause is extraposed, there is no pro-form in the canonical place:
But extraposed that-clauses may connect to a sort of dummy it, especially in subject position:
So if you call LD "extraposition", you'll confuse contemporary syntacticians; and if you want to find out what they have to say about examples of LD, you have to search for it under the name that they use.
Nathan Myers said,
September 27, 2008 @ 1:29 am
We've seen a few examples of a similar construction:
"The blood of the martyrs? You're soaking in it."
"Skills, I haz dem"
What are they called?
Mark Liberman said,
September 27, 2008 @ 6:02 am
@Nathan Myers:
"Skills, I haz dem" seems to be an ordinary case of LD, just one where the pronoun is in object position rather than the commoner subject position.
"The blood of the martyrs? You're soaking in it." is a construction that is exactly like ordinary LD, except that the initial NP is questioned, with the speaker or writer apparently acting on behalf of an interlocutor to echo a question that sets the stage for the comment that follows.
Whether this is just a particular kind of LD, or a different but related construction, is not clear to me. I'll ask around and get back to you.
Nathan Myers said,
September 28, 2008 @ 7:13 pm
Thanks, Mark. It seems left subject dislocation (if I may invent a name) is most useful for improving scansion, and sometimes to help construct a complicated noun phrase ("Nilima and me, our car, it …"). Left object dislocation seems the more powerful, enabling reordering for emphasis adding nothing beyond a pronoun.
Anonymous Cowherd said,
September 29, 2008 @ 8:49 pm
"My wife… I think I'll keep her."
Coordinate possessives « ESOL World News said,
October 13, 2008 @ 12:32 pm
[…] LANGUAGE LOG–Comments on Mark Liberman's Left Dislocation posting drifted for a while into the vexed question of how to express possession when two (or more, though […]
John Cowan said,
October 20, 2008 @ 2:06 am
Laurent C: I know that's the traditional reading, and etymologically the bound prefix j- does descend from the Latin pronoun ego. But I argue that in French it is now a prefix, serving the same function as the Latin suffix -o: namely, first person singular agreement, whether the subject is actually expressed (using mwa) or not. In addition, French now has object agreement markers, which Latin did not.
The other Romance languages are on the boundary line in this respect: their bound pronouns are clitics rather than prefixes.
Grammar fail fail « Motivated Grammar said,
December 1, 2010 @ 1:36 pm
[…] whoever … here isn't complete.* A comma is indeed correct here; this is an example of left-dislocation, rare in written English but common in spoken English and many other languages**. In […]
Xmun said,
October 13, 2014 @ 1:06 pm
@John Cowan
Please explain your distinction between prefixes and clitics. I'd have thought the j in jdetèst etc. to be a clitic or more precisely a proclitic (judging by the definitions of these terms in the Oxford Dictionary of English Grammar).