Here are examples of usage:
Almost immediately Mr Bartletop came shuffling out. (source)
I started violently when she came shuffling out. (source)
She stood in the court as the Germans came shuffling out. (source)
What is the grammatical function of shuffling? I can't seem to locate any category that clearly fits.
Update: It turns out that this sort of thing has indeed been studied, but apparently only relatively recently (2008). Having said that, there were precursors to it, e.g. in Quirk, Greenbaum, Leech, and Svartvik (QGLS), as I will explain below.
Here is a Master's thesis in which the author describes a pattern called
Path Verb plus Manner of Motion Verb in the Ing-Form.
This, apparently, should be distinguished from certain participle clauses, as explained here:
There exists an overlap between participle clauses and the sentences containing a path verb + a verb in the ing-form. The following sentences illustrate the difference between the construction ‘a path verb plus a verb in the ing-form’ and the participle clause:
(26) When her mother died, Black Irene gave out such a blood-curdling scream in the middle of the night that we all woke up with a jump. My son went running to Irene's house to see what was the matter.
(27) Any junkie or Bowery red-eye comes limping down the street, then five sombre fatboys with baseball bats and axe-handles stride out of the nearest trattoria.
(28) People kept pouring in; the women would come in chatting and laughing, then as soon as they saw the mourners they’d break into loud wails […].
(29) One Sunday he entered the chapel shouting and screaming and cursing and woke the snoring congregation.
In sentences (26) and (27), the verbs in the ing-form clearly define the way of movement. On the other hand, the ing verbs in sentences (28) and (29) are participial clauses denoting simultaneous actions. In the following chapters, it will be emphasized that especially with the ‘a path verb plus a manner verb denoting a non-kinetic type of activity’ construction it is sometimes difficult to draw a clear line between these two phenomena because the borderline is sometimes very thin.
As has already been pointed out, the possible disputability of some of the examples of the pattern under consideration is more common in the case of non-kinetic verbs.
(30) Meanwhile, the two boys went on their way laughing.
Example (30) presents one of the borderline examples between the VVingPP construction and participle clauses. But this example should be labelled as participle clause because, according to Goldberg (2006, 51), the directional complement should follow the verb in the progressive form.
(31) “Help, help! I can’t get down,” I hollered. Soon my rescuers arrived jostling, pushing, laughing, joking.
Also example (31) represents a case of participle clause because the path verb is followed by a series of non-motion verbs instead of just one. However, it is still possible to say that non-motion verbs depict the way the movement is carried out and that they do not merely represent simultaneous actions.
The text continues with the discussion of 'Path Verb plus Manner of Motion Verb in the Ing-Form' per se:
As has been mentioned earlier, this specific sub-type of the construction may be found in Kudrnáčová’s (2008) publication in chapter ‘The Dissociation of a Change of Location and the Manner of its Execution’. Kudrnáčová uses here the sentence John came running to the store to illustrate the structure of this construction.
“Although the motion per se (expressed in the path verb) and its manner (expressed in the manner of motion verb) are each given a separate expression, the prepositional phrase specifies the direction of both the movement expressed in the path verb and the movement expressed in the manner of motion verb” (Kudrnáčová 2008, 48). This statement is valid when the manner of motion verb is used in purely kinetic such as in the following sentence:
(32) Then a few days later, the other girl, Sally, came dancing across the playground and said she’d just got hers back.
This example shows that Sally both ‘came across the playground’ and ‘danced across the playground’, which proves that “both path verb and manner of motion verb in the ing-form form an integral part of the directed motion event in question. These two components represent a compact unit” (Kudrnáčová 2008, 48). This characteristic proves that Goldberg’s VVingPP construction does not fully correspond with this sub-type of construction because Goldberg (2006, 51) claims that “the directional is an argument of the main verb, not of the second verb” (Goldberg 2006, 51). This claim brings the VVingPP construction closer to the second sub-type of the construction and will be analysed later in the course of this thesis.
As has been mentioned in the first part of this thesis, the reader’s/listener’s perception of the sentence differs in accordance with the use of a path verb or a manner of motion verb. Their use in combination “enables the speaker to present the mere fact of the change of location [in telic events] as having a processual character” (Kudrnáčová 2008, 50). At the same time, the use of this construction helps the encoder to achieve more vivid presentation of the movement because the attention of the decoder is focused both on the change of location and on the way it is achieved. It is true, however, that the path verb serves mostly as an indicator of the direction (in versus out of, to versus from the defined point etc.) and the specific information about the motion is conveyed in the ing-verb.
There were precursors to this study. For example,
The existence of the phrases of ‘they came running’ type is also briefly mentioned in QGLS in sub-chapters 8.28 and 8.29 (pp. 506-507), which deal with adjuncts. They mention not only motion verbs come and go but also non-motion verbs sit and stand. They state that:
“This is particularly noticeable in the case of certain verbs of broad meaning in respect of posture or motion: sit, stand, come, go. These can take an obligatory adjunct of respect in the form of an –ing clause, with consequent weakening of the primary meaning of the main verb” (Quirk et al, 1985, 506).
As examples for these three verbs they provide the following sentences:
(10) He stood waiting (patiently).
(11) She sat reading (to the children).
(12) They went hurrying (breathlessly).
(13) She came running (in great haste).
The above seems to be about as detailed a discussion as we are likely to find...
Previous guesses
For what it's worth, here are some other things that came to mind, which I'm leaving for completeness:
1. Shuffling as a direct object (or possibly a verb phrase complement) of a catenative verb
The idea here is that came (out) is a catenative verb, and shuffling its object, or, possibly, its verb phrase complement. From Wikipedia:
Catenative verbs are verbs which can be followed within the same clause by another verb. This second subordinated verb can be in either the infinitive (both full and bare) or gerund forms. An example appears in the sentence He deserves to win the cup, where "deserve" is a catenative verb which can be followed directly by another verb, in this case a to-infinitive construction.
Some catenative verbs are used in the passive voice followed by an infinitive: You are forbidden to smoke in here.
These verbs are called "catenative" because of their ability to form chains in catenative constructions. For example: We need to go to the tennis court to help Jim to get some practice before the game.
(Note that the name catenative comes from Latin catena, meaning 'chain'.)
There is indeed some positive evidence for come being catenative here. In QGLS, on p. 146, they make the following comment:
Unlike main verb constructions such as expect (to), want (to), and attempt (to), catenative constructions are in no way syntactically related to transitive verb constructions in which the verb is followed by a direct object or prepositional object (cf 16.268, 16.38). Compare:
John appeared to attack the burglar.
John attempted to attack the burglar.But:
*John appeared an attack on the burglar.
John attempted an attack on the burglar.
Let's try with come:
John came talking over the phone.
John attempted talking over the phone.
*John came a talk over the phone.
John attempted a talk over the phone.
So, this is an indication that the relationship between come (OK, without out in these cases) and the -ing participle is not one of an action verb and a direct object. This does support the theory that come functions like a catenative verb here.
On the other hand, there are also some problems with this theory.
One problem is that the relevant examples in QGLS (on p. 147) seem somewhat different from our case:
The girl started (out)/kept (on)/went on working.
Why different? Well, first of all, it seems that in these cases, one can't put the out (or on) to the end:
The girl started working out.
?The girl kept/went working on.
The example with started completely changes in meaning, whereas the examples with kept and went now look dubious. (Don't they? One might ask, What about keep on looking and keep looking on? Well, note that these aren't simple rewrites of each other, because on in the latter no longer belongs to keep, but instead to looking: 'looking on'. Indeed, you can say keep on looking on.)
In contrast,
He came out shuffling
and
He came shuffling out
both seem fine, and both seem to mean the same thing, and out in both seems to be associated with came.
Secondly, in the examples from QGLS, the participle seems to be pretty much required: if you do say
The girl started (out)/kept (on)/went on
to me, at least, this sounds like an ellipsis.
In contrast,
He came out
works just fine without shuffling or anything else.
Another problem is that, although come does appear in lists of catenative verbs (like this one), it is listed as requiring a bare infinitive in such usage. I suppose this refers to imperative, as in come eat? But then what about She came to eat?
Despite all these problems, so far, this theory where come out functions as a catenative verb seems more promising than the rest, to which I now turn.
2. Shuffling as an adverbial phrase (AdvP)
First of all, many sources (e.g. CGEL) will tell you that the head of an AdvP must be an adverb (which shuffling is not). Second, shuffling doesn't behave the same way an actual adverb like slowly does: compare
Slowly, a man came out
A man slowly came out
A man came slowly out
A man came out slowly,
which are all grammatical, with
?Shuffling, a man came out
?A man shuffling came out
A man came shuffling out
A man came out shuffling,
where the first two are dubious at best (the second is perhaps akin to "a man on fire came out"), and anyway in them, shuffling refers to man, not came. In contrast, in the examples of slowly, that word always refers to came.
So, shuffling does not seem to be an AdvP.
3. Shuffling as a complement of came
Here I have the mind an analog of the following pattern in QGLS (p. 1206), which they call Object + -ing participle complementation. This pattern is relevant to verbs of perception (feel, hear, notice,...), verbs of encounter (catch, discover, find, leave), and two verbs of coercive meaning (get and have). The perception verbs can also come with the bare infinitive, though the meaning changes somewhat. Example:
Tim watched Bill mend/mending the lamp.
The -ing predication can normally be omitted without radically altering the meaning:
I saw him lying on the beach. [entails: I saw him]
The problem, of course, is that there is no object in come out shuffling. And I was unable to find a pattern in QGLS that matches come out shuffling exactly.
Moreover, I'm not sure if shuffling is really a complement of come, or just a modifier.
4. Shuffling as an elliptical clause
Some sources (e.g. this one, in Sec. 43.e) say that a sentence like
I saw the elephant roller-skating
should be understood not in terms of roller-skating complementing saw (like QGLS suggests), but rather in terms of being an elliptical clause, e.g of this sentence:
I saw the elephant: it was roller-skating.
The fact that this treatment contradicts the one in QGLS is already worrisome. However, if it is accepted, then I suppose that He came out shuffling could be understood as the elliptical phrase arising from something like
He came out; he was shuffling.
But in all honesty, this story with elliptical clauses sounds like a stretch.
First of all, if this really is an instance of an ellipsis, there should be strong, explicit arguments for that conclusion.
Second, QGLS gives five characteristics of ellipses (pp. 884-887):
To distinguish ellipsis from other kinds of omission, it is important to emphasise the principle of VERBATIM RECOVERABILITY that applies to ellipsis; that is, the actual word(s) whose meaning is understood or implied must be recoverable. Even so, like those of so many other grammatical categories, the boundaries of ellipsis are unclear, and it is best to recognize different degrees of 'strength' in the identification of examples of ellipsis. To be ellipsis in the strictest sense, an example must satisfy all the criteria specified [below]:
i. The ellipted words are precisely recoverable;
ii. The elliptical construction is grammatically 'defective';
iii. The insertion of the missing words results in a grammatical sentence
(with the same meaning as the original sentence;
iv. The missing word(s) are textually recoverable, and;
v. are present in the text in exactly the same form.
Looking at these, we see that none of them really apply to our case.
For starters, note that in the alleged reconstruction,
He came out; he was shuffling
we had to exchange the order of out and shuffling; this runs afoul of iii. Further, was is not present in the text at all; this violates iv. and v. Next, arguable, the following reconstruction is just as good:
He came out, and he was shuffling
which violates i. Finally, it's not at all obvious that our initial sentence is grammatically 'defective'; this violates ii.
5. Shuffle as an instance of semantic implication
QGLS recognizes other kinds of omission, e.g. "sematic implication" (p. 884). They give two examples of that:
i. Frankly, he is very stupid, about which they say, "The disjunct frankly implies a comment by the speaker on the way he is speaking. But there is no one set of missing words that can be supplied. We can expand frankly to (among many forms) I am speaking frankly when I say . . . or If I may put it frankly I would tell you....
ii. He's drunk, because I saw him staggering, for which " it is difficult to pin down in exact words what has been omitted," as any of the following could work: He's drunk, and I claim this because I saw him staggering, or and I know, or and I am sure of it, or and I am convinced of it, or and the proof is that...
But somehow shuffling seems much more intergated into He came shuffling out than the frankly and because I saw him staggering are into their respective sentences. One piece of evidence for this is that offsetting shuffling with commas looks weird: He came, shuffling, out. In contrast, in the QGLS examples, it is weird without the commas; even if it's rewritten as He is, frankly, very stupid, you'd still want those commas. So again our sentence doesn't seem to fit very well to the pattern described in QGLS.
A few extra observations
Finally, notice that certain other verbs are possible here, in addition to come (out), for example
He emerged shuffling.
He arrived shuffling.
Moreover, all this is very similar to the usage in the well-known idiom
He came out swinging.
In conclusion
So far, the correct analysis of the grammar of the word shuffling in the sentence He came shuffling out seems to be in terms of a construction called 'Path Verb plus Manner of Motion Verb in the Ing-Form'. This construction is not at all a part of standard grammar, since it appears to have only been studied relatively recently.
If anyone has a different opinion, I would very much appreciate hearing about it!