Respect all members: no insults, harassment, or hate speech.
Be tolerant of different viewpoints, cultures, and beliefs. If you do not agree with others, just create separate note, article or collection.
Clearly distinguish between personal opinion and fact.
Verify facts before posting, especially when writing about history, science, or statistics.
Promotional content must be published on the “Related Services and Products” page—no more than one paragraph per service. You can also create subpages under the “Related Services and Products” page and publish longer promotional text there.
Do not post materials that infringe on copyright without permission.
Always credit sources when sharing information, quotes, or media.
Be respectful of the work of others when making changes.
Discuss major edits instead of removing others' contributions without reason.
If you notice rule-breaking, notify community about it in talks.
Do not share personal data of others without their consent.
Preposition stranding or p-stranding is the syntactic construction in which a so-called stranded, hanging, or danglingpreposition occurs somewhere other than immediately before its corresponding object; for example, at the end of a sentence. The term preposition stranding was coined in 1964, predated by stranded preposition in 1949.[1][2] Linguists had previously identified such a construction as a sentence-terminal preposition[3] or as a preposition at the end.[4]
Wh-movement—which involves wh-words like who, what, when, where, why and how—is a syntactic dependency between a sentence-initial wh-word and the gap that it is associated with. Wh-movement can lead to P-stranding if the object of the preposition is moved to sentence-initial position, and the preposition is left behind. P-stranding from wh-movement is observed in English and Scandinavian languages. The more common alternative is called pied piping, a rule that prohibits separating a preposition from its object, for instances in Serbo-Croatian and Arabic languages. English and Dutch use both rules, providing the option of two constructions in these situations.
An open interrogative often takes the form of a wh- question (beginning with a word like what or who).
P-stranding in English allows the separation of the preposition from its object, while pied piping allows carrying the preposition along with the wh- object.[11] From the examples below, we can see the two options.
P-stranding in Danish is banned only if the wh-word is referring to nominative cases.[12] "Peter has spoken with <whom>", the wh-word <whom> is the accusative case. Therefore, p-stranding is allowed.
Wh-movement in Greek states that the extracted PP must be in Spec-CP,[14] which means the PP (me) needs to move with the wh-word (Pjon). It can thus be seen that Greek allows pied piping in wh-movement but not prepositional stranding.
Pied-piping is the only grammatical option in Spanish to construct oblique relative clauses.[15] Since pied-piping is the opposite of p-stranding, p-stranding in Spanish is not possible (* indicates ungrammaticality).
P-stranding in wh-movement sentences is normally banned in LA. However, a recent study found that a preposition seems to be stranded in a resumptive wh-question.[16]
Sluicing is a specific type of ellipsis that involves wh-phrases. In sluicing, the wh-phrase is stranded while the sentential portion of the constituent question is deleted. It is important to note that the preposition is stranded inside the constituent questions before sluicing. Some languages allow prepositional stranding under sluicing, while other languages ban it.[10][11] The theory of preposition stranding generalization (PSG) suggests that if a language allows preposition stranding under wh-movement, that language will also allow preposition stranding under sluicing.[17] PSG is not obeyed universally; examples of the banning of p-stranding under sluicing are provided below.
A number of common Dutch adpositions can be used either prepositionally or postpositionally, with a slight change in possible meanings. For example, Dutch in can mean either in or into when used prepositionally, but only mean into when used postpositionally. When postpositions, such adpositions can be stranded:
short-distance movement:
[...]
[...]
dat
that
hij
he
zo'n
such-a
donker
dark
bos
forest
niet
not
in
into
durft
dares
te
to
lopen
walk
[...]
[...]
[...] dat hij zo'ndonkerbos niet in durft te lopen [...]
[...] that he such-adarkforest not into dares to walk [...]
'[...] that he doesn't dare walk into such a dark forest [...]'
Another way to analyze examples like the one above would be to allow arbitrary "postposition + verb" sequences to act as transitive separable prefix verbs (e.g. in+lopen → inlopen), but such an analysis would not be consistent with the position of in in the second example. (The postposition can also appear in the verbal prefix position: [...]dat hij zo'n donker bos niet durft in te lopen[...].)
Pseudopassives (prepositional passives or passive constructions) are the result of the movement of the object of a preposition to fill an empty subject position for a passive verb. The phenomenon is comparable to regular passives, which are formed through the movement of the object of the verb to subject position. In prepositional passives, unlike in wh-movement, the object of the preposition is not a wh-word but rather a pronoun or noun phrase:
To standard French ears, all of those constructions sound quite alien and are thus considered barbarisms or anglicismes.
However, not all dialects of French allow preposition stranding to the same extent. For instance, Ontario French restricts preposition stranding to relative clauses with certain prepositions. In most dialects, stranding is impossible with the prepositions à 'to' and de 'of'.
A superficially-similar construction is possible in standard French in cases where the object is not moved but implied, such as Je suis pour 'I'm all for (it)' or Il faudra agir selon 'We'll have to act according to (the situation)'.
Some dialects permit
Tu connais pas la fille que je te parle de.
'You don't know the girl that I'm talking to you about.'
Dutch prepositions generally do not take the ordinary neuter pronouns (het, dat, wat, etc.) as objects. Instead, they become postpositional suffixes for the corresponding r-pronouns (er, daar, waar, etc.): hence, not *over het ('about it'), but erover (literally 'thereabout'). However, the r-pronouns can sometimes be moved to the left and thereby strand the postposition:[20]
Some regional varieties of German show a similar phenomenon to some Dutch constructions with da(r)- and wo(r)- forms. That is called a split construction (Spaltkonstruktion). Standard German provides composite words for the particle and the bound preposition. The split occurs easily with a composite interrogative word (as shown in the English example) or with a composite demonstrative word (as shown in the Dutch example).
For example, the demonstrative davon ('of that / of those / thereof'):
Standard German requires
Ich
I
kann
can
mir
me
davon
thereof
nichts
nothing
leisten.
afford.
Ich kann mir davon nichts leisten.
I can me thereof nothing afford.
'I can't afford any of those.'
Some dialects permit
Ich
I
kann
can
mir
me
da
there-[clipped]
nichts
nothing
von
of
leisten.
afford.
Ich kann mir da nichts von leisten.
I can me there-[clipped] nothing of afford.
'I can't afford any of those.'
Again, although the stranded postposition has nearly the same surface distribution as a separable verbal prefix (herbekommen is a valid composite verb), it would not be possible to analyze these Dutch and German examples in terms of the reanalyzed verbs *overpraten and *vonkaufen, for the following reasons:
The stranding construction is possible with prepositions that never appear as separable verbal prefixes (e.g., Dutch van, German von).
Stranding is not possible with any kind of object besides an r-pronoun.
Prefixed verbs are stressed on the prefix; in the string von kaufen in the above sentences, the preposition cannot be accented.
Also, pronunciation allows distinguishing an actual usage of a verb like herbekommen from a split construction her bekommen.
Although preposition stranding has been found in English since the earliest times,[21] it has often been the subject of controversy, and some usage advisors have attempted to form a prescriptive rule against it. In 1926, H. W. Fowler noted: "It is a cherished superstition that prepositions must, in spite of the incurable English instinct for putting them late [...] be kept true to their name & placed before the word they govern."[22]
The earliest attested disparagement of preposition stranding in English is datable to the 17th-century grammarian Joshua Poole,[3] but it became popular after 1672, when the poet John Dryden objected to Ben Jonson's 1611 phrase "the bodies that those souls were frighted from". Dryden did not explain why he thought the sentence should be restructured to front the preposition.[23][24] In his earlier writing, Dryden himself had employed terminal prepositions but he systematically removed them in later editions of his work, explaining that when in doubt he would translate his English into Latin to test its elegance.[4] Latin has no construction comparable to preposition stranding.
Usage writer Robert Lowth wrote in his 1762 textbook A Short Introduction to English Grammar that the construction was more suitable for informal than for formal English: "This is an Idiom which our language is strongly inclined to; it prevails in common conversation, and suits very well with the familiar style in writing; but the placing of the Preposition before the Relative is more graceful, as well as more perspicuous; and agrees much better with the solemn and elevated Style."[25] However Lowth used the construction himself, including a humorously self-referential example in this passage ("is strongly inclined to"), and his comments do not amount to a proscription.
A stronger view was taken by Edward Gibbon, who not only disparaged sentence-terminal prepositions but, noting that prepositions and adverbs are often difficult to distinguish, also avoided phrasal verbs which put on, over or under at the end of the sentence, even when these are clearly adverbs.[4][b] By the 19th century, the tradition of English school teaching had come to deprecate the construction, and the proscription is still taught in some schools at the beginning of the 21st century.[26]
However, there were also voices which took an opposite view. Fowler dedicated four columns of his Dictionary of Modern English Usage to a rebuttal of the prescription:
The fact is that the remarkable freedom enjoyed by English in putting its prepositions late & omitting its relatives is an important element in the flexibility of the language. [...] That depends on what they are cut with is not improved by conversion into That depends on with what they are cut; & too often the lust of sophistication, once blooded, becomes uncontrollable, & ends with, That depends on the answer to the question as to with what they are cut.
Criticizing the controversy over preposition stranding, American linguist Donald Ringe stated:
The original reason for the objection, apparently, was that Latin has no such construction (or, with a bit more sophistication, that few other languages have such a construction). In other words, people who objected to preposition stranding were insisting that English grammar should be like Latin. That's perverse - English isn't Latin and isn't even descended from Latin...
— Donald Ringe, An Introduction to Grammar for Language Learners, Epilogue
Overzealous avoidance of stranded prepositions was sometimes ridiculed for leading to unnatural-sounding sentences, including the quip apocryphally attributed to Winston Churchill: This is the sort of tedious nonsense up with which I will not put.[28]
Today, most sources consider it to be acceptable in standard formal English.[26][29][30] As O'Conner and Kellerman point out: "Great literature from Chaucer to Milton to Shakespeare to the King James version of the Bible was full of so called terminal prepositions."[29]Mignon Fogarty ("Grammar Girl") says, "nearly all grammarians agree that it's fine to end sentences with prepositions, at least in some cases."[31]
^ abcIn transformational approaches to syntax, it is commonly assumed that the movement of a constituent out of a phrase leaves a silent trace, in this case following the preposition:
Whati are you talking about ___i?
This bed looks as if iti has been slept in ___i.
This is the bookithati I told you about ___i.
^For more on the distinction between verbs with particles (called adverbs in older texts) and those with prepositional phrases, see English phrasal verbs#Types
^Algryani, A. (2012). He Syntax of Ellipsis in Libyan Arabic: A generative analysis of sluicing, Vp ellipsis, stripping and negative contrast (dissertation).
^O'Conner and Kellerman 2009. p. 22. "It's perfectly natural to put a preposition at the end of a sentence, and it has been since Anglo-Saxon times."
^Fowler, Henry Watson (1926). "Preposition at end". A Dictionary of Modern English Usage. OUP. p. 457. (cited from the revised ed. 1940). Similarly Burchfield in the 1996 version: "One of the most persistent myths about prepositions in English is that they properly belong before the word or words they govern and should not be placed at the end of a clause or sentence." Burchfield 1996. p. 617.
^Fogarty, Mignon (2011). Grammar Girl Presents the Ultimate Writing Guide for Students. New York: Henry Holt & Company. pp. 45–46. ISBN978-0-8050-8943-1.
Haegeman, Liliane, and Jacqueline Guéron. 1999. English Grammar: a Generative Perspective. Oxford: Blackwell. ISBN0-631-18839-8.
Hornstein, Norbert, and Amy Weinberg. 1981. "Case theory and preposition stranding." Linguistic Inquiry 12:55–91. Hornstein, N.; Weinberg, A. (1 January 1981). "Case Theory and Preposition Stranding". Linguistic Inquiry. 12 (1): 55–91. ISSN0024-3892. JSTOR4178205.
Koopman, Hilda. 2000. "Prepositions, postpositions, circumpositions, and particles." In The Syntax of Specifiers and Heads, pp. 204–260. London: Routledge. ISBN0-415-16183-5.
Takami, Ken-ichi. 1992. Preposition Stranding: From Syntactic to Functional Analyses. Berlin: Mouton de Gruyter. ISBN3-11-013376-8.
van Riemsdijk, Henk. 1978. A Case Study in Syntactic Markedness: The Binding Nature of Prepositional Phrases. Dordrecht: Foris. ISBN90-316-0160-8.
Fowler, Henry. 1926. "Preposition at end." A Dictionary of Modern English Usage. Oxford: Clarendon Press. Wordsworth Edition reprint, 1994, ISBN1-85326-318-4
Preposition stranding is a syntactic phenomenon in which a preposition becomes separated from its complement (typically a noun phrase) due to movement operations in a sentence, such as wh-movement in questions or relative clauses, leaving the preposition in its original position at the end of the clause.[1] This construction is exemplified in English by utterances like "Who did you talk to?" where the preposition "to" is stranded after the movement of its object "who" to the front of the clause.[2]In contrast to preposition stranding, the alternative pied-piping construction involves the preposition moving together with its complement, as in the more formal variant "To whom did you talk?"[3] Preposition stranding occurs not only in interrogatives but also in relative clauses (e.g., "the person who you talked to") and passives (e.g., "the issue was talked about"), though its availability varies by construction and context.[1] While frequently used in informal spoken English, stranding has historically been stigmatized in prescriptive grammar rules, which favor pied-piping as more elegant, despite its prevalence in everyday usage.[2]Cross-linguistically, preposition stranding is relatively rare among the world's languages, being permitted freely in English and the continental Scandinavian languages (e.g., Danish, Swedish) but disallowed in Romance languages such as French and Italian, where pied-piping is obligatory.[4] In languages like Dutch and German, it is more restricted, often limited to certain pronouns or contexts.[1] Linguists have debated its theoretical underpinnings, with analyses linking it to processing efficiency—stranding may reduce immediate cognitive load at gap sites in simpler clauses but increase overall complexity in embedded structures—and to parametric variations in syntactic rules across languages.[3] Acquisition studies indicate that English-speaking children master stranding around age 2;7, following related constructions like verb-particle phrases, highlighting its integration into early grammar development.[4]
Overview
Definition
Preposition stranding is a syntactic phenomenon in which a preposition is separated from its complement and left in a position at or near the end of a clause, typically without an immediately following object.[5] This occurs when the complement, such as a wh-word or noun phrase, is extracted to a higher position in the sentence structure, leaving the preposition "stranded." In contrast, pied-piping involves the preposition moving together with its complement to the front of the clause, as in the formal English variant "To whom did you speak?" rather than the stranding construction "Who(m) did you speak to?"[5][6]The basic mechanism of stranding arises during processes like wh-movement, where the complement is displaced, but the preposition remains in its original position due to language-specific syntactic rules.[5] For instance, in English questions, this yields constructions such as "What are you looking at?" where "at" is stranded after the extraction of "what." Similarly, in relative clauses, "the person who I was talking to" exemplifies stranding, contrasting with the pied-piped "the person to whom I was talking."[5] These examples illustrate how stranding contributes to the informal, analytic style prevalent in spoken English.[7]Typologically, preposition stranding is more common in analytic languages like English and Scandinavian languages, which rely heavily on word order and function words rather than inflection.[1] In contrast, it is rare or disallowed in more synthetic languages, such as those in the continental West Germanic family, where prepositions typically move with their complements or require alternative strategies like resumptive pronouns.[1] This variation highlights stranding as a feature tied to the degree of analyticity in a language's grammar.[1]
Historical Development
Preposition stranding has been a feature of English since the Middle English period (c. 1100–1500), evolving from earlier Germanic constructions, though limited in Old English.[8] The phenomenon attracted early critical attention from 17th-century grammarians, who viewed it as a deviation from classical norms. In 1672, poet and critic John Dryden famously condemned it as an "idiom which our language is strong in" but ungrammatical, citing examples from Ben Jonson's work in his Defence of the Epilogue to The Conquest of Granada.[8] This prescriptivist stance, lacking historical precedent in English, marked the beginning of debates over its legitimacy, influencing subsequent grammars.[9]By the 18th century, such criticisms persisted but began to incorporate descriptive elements. Bishop Robert Lowth, in his 1762 A Short Introduction to English Grammar, described stranding as a "common idiom" natural to spoken English and informal writing, though he advised preposition pied-piping for elevated styles to align with Latin models.[8] This period saw stranding increasingly documented in usage, yet prescriptive rules against it solidified in educational texts, reflecting broader efforts to standardize English along classical lines.In the 19th and early 20th centuries, linguistic scholarship shifted toward descriptive analysis, integrating stranding into accounts of English syntax. Otto Jespersen, in his 1924 The Philosophy of Grammar, emphasized English's analytic tendencies, contributing to the language's evolution from synthetic to more particle-like elements.[10] This work highlighted stranding as emblematic of English's pragmatic word order preferences over rigid inflectional rules.The modern study of preposition stranding emerged in the 1960s with the advent of generative linguistics, which formalized it as a rule-governed process. Noam Chomsky's 1977 paper "On Wh-Movement," published in Formal Syntax, analyzed stranding as integral to wh-movement operations, positing it as a language parameter distinguishing English-like languages (which permit stranding) from those requiring pied-piping, such as French.[11] This framework elevated stranding from a stylistic quirk to a core syntactic feature. The term "preposition stranding" itself crystallized in the linguistic literature of the 1970s, appearing prominently in generative analyses to denote the separation of prepositions from their complements.[11]Post-1980s research expanded into cross-linguistic comparisons, revealing stranding's variability across languages and its ties to broader parametric settings in Government and Binding theory. Studies like Hornstein and Weinberg's 1981 "Case Theory and Preposition Stranding" in Linguistic Inquiry explored its implications for case assignment, while subsequent works examined occurrences in languages such as Dutch and Scandinavian tongues. These investigations underscored stranding's role in universal grammar debates, fostering a richer understanding of syntactic diversity.
Theoretical Background
Generative Grammar Approaches
In generative grammar, preposition stranding is treated as a subtype of A-bar movement, specifically wh-movement, in which the complement of a preposition is extracted, leaving a trace bound to the stranded preposition. This analysis originates in Government and Binding (GB) theory, where the trace must be properly governed and case-licensed for the derivation to converge. In the Minimalist Program, the process is recast as internal Merge driven by feature checking, with the preposition unable to check the case feature of its complement, allowing extraction without violating locality constraints like the Phase Impenetrability Condition.A key parametric distinction accounts for cross-linguistic variation in stranding: languages like English permit it because prepositions do not assign (structural) case to their complements, enabling the trace to receive case from a higher verbal head through a reanalysis rule that incorporates the preposition into the verb's projection. In contrast, languages like French disallow stranding because prepositions function as case-assigners (proper governors), blocking extraction as the trace would remain uncased.[12] This parameter aligns with broader head-complement asymmetries, where English prepositions lack the government properties of their French counterparts.[12]Within GB theory, bounding constraints such as Subjacency further restrict stranding by limiting movement across certain barriers; for instance, extraction from a complex NP island yields ungrammaticality, as in *Who_i did you meet the woman that argued with t_i?, violating the bounding nodes S and NP. Similar island effects apply in the Minimalist framework via phase-based intervention, ensuring that stranded prepositions do not circumvent locality.Formally, stranding in English wh-questions is represented with a non-branching PP after reanalysis, where the preposition adjoins to the verb, allowing the complement to move freely:
CP
├── Spec: who_i
└── C'
└── IP
├── you
├── V': talked-to
└── PP: t_i (trace)
CP
├── Spec: who_i
└── C'
└── IP
├── you
├── V': talked-to
└── PP: t_i (trace)
This structure illustrates the trace's governance by the reanalyzed verbal complex, satisfying theta-role assignment and case requirements.
Alternative Syntactic Theories
In Construction Grammar, preposition stranding is analyzed as a holistic construction rather than a byproduct of movement rules, where the entire form-meaning pairing of stranding patterns emerges from usage-based generalizations in language data. This approach posits that stranding competes with pied-piping as distinct but related constructions, with factors like end-weight influencing acceptability; for instance, heavier complements following the stranded preposition improve grammaticality, as in "This chair feels to me so hard that it could be made of concrete," where the complex post-verbal phrase favors stranding over pied-piping. Adele Goldberg's work emphasizes how such constructions are entrenched through frequency and context, allowing productive use in English relative clauses and wh-questions without invoking universal parameters. Thomas Hoffmann further argues that this framework better captures cross-dialectal variation in British and Kenyan English, where stranding frequency correlates with corpus-based patterns rather than innate syntactic constraints.Dependency Grammar treats prepositions as non-head dependents of their nominal objects or verbs, with stranding arising through reattachment mechanisms that preserve hierarchical relations without movement. In pseudo-passives like "He was yelled at," the preposition "at" reattaches directly to the verb via a specialized "prep-strand" relation, ensuring valency satisfaction and avoiding dangling modifiers. For relative clauses, such as "The man we talked about," the stranded preposition reattaches to the antecedent noun, maintaining dependency integrity. This analysis extends to languages like Dutch, where R-pronouns (e.g., "waar" in "Waar heb je over gepraat?" – "What did you talk about?") enable stranding by fronting as clitics, reattaching the preposition to the verb while adhering to linear precedence constraints in dependency trees.Functionalist approaches view preposition stranding as driven by discourse-pragmatic needs, particularly in structuring information flow to highlight new or focal elements in spoken English. Stranding facilitates end-focus by placing prepositions after lighter extracted phrases, as in "Who did you talk to?" versus the more formal "To whom did you talk?", enhancing conversational naturalness and accessibility. Ken-ichi Takami's functional analyses underscore how stranding aligns with cognitive salience, where the extracted constituent carries discourse-new information, promoting efficient communication over rigid syntax.Critiques of generativist accounts highlight their lack of universality, as preposition stranding exhibits gradient acceptability across typological profiles rather than binary parameters. For example, while English freely strands prepositions, languages like French prohibit it entirely, and Dutch restricts it to R-pronouns, revealing pragmatic and historical influences over purported innate settings. Takami argues that syntactic theories fail to explain such variation, advocating functional explanations that prioritize usage and context for broader cross-linguistic insight.
P-stranding in Wh-movement
Languages Allowing P-stranding
Preposition stranding in wh-movement is fully permitted in English, allowing the wh-element to extract while leaving the preposition at the end of the clause. A canonical example is "Who(m) are you waiting for?", where "for" is stranded after the movement of "who(m)".[13] This construction became prevalent in Modern English following a historical shift from a preference for pied-piping in Old and Middle English, where stranding was limited to specific contexts like relatives with the invariant complementizer þe and gradually expanded to broader wh-movement by the Early Modern period.[14]In Danish, preposition stranding under wh-movement is similarly allowed and common, particularly in V2 questions that characterize the language's syntax. For instance, "Hvem taler du med?" ("Who are you talking to?") strands the preposition med ("with") after extraction of the wh-pronoun hvem.[15] This parallels English in its flexibility, occurring freely with prepositions denoting accompaniment, instrument, and place, and extending to other semantic categories like manner under certain conditions.[15]Dutch permits preposition stranding in wh-movement, though it is variable and more restricted than in English or Danish, often depending on register and the involvement of R-pronouns. The standard pied-piping form is "Met wie praat je?" ("With whom are you talking?"), but colloquial speech, especially in northern and western dialects, favors stranding as in "Wie praat je met?" ("Who are you talking with?").[16] R-pronouns like waar or waarmee facilitate stranding by optionally incorporating into the preposition (e.g., "Waar praat je over?" becoming "Wie praat je over?" in informal contexts), reflecting a syntactic mechanism tied to pronominal cliticization.[16]These languages share analytic traits as Germanic VO varieties, including the loss of rich case morphology on nouns and prepositions, which enables stranding by reducing the need for case agreement between the preposition and its extracted complement.[17] This parametric variation aligns with generative approaches positing differences in head movement or phase boundaries for prepositional phrases.[13]
Languages Disallowing P-stranding
In languages that disallow preposition stranding (P-stranding) during wh-movement, the preposition must obligatorily pied-pipe with the wh-phrase to the sentence-initial position, preventing the separation of the preposition from its complement. This restriction contrasts with P-stranding languages like English and is often attributed to parametric differences in syntactic structure, such as the inability of prepositions to assign case independently or the presence of PP islands that block extraction of the DP complement.[18]Greek exemplifies a strict ban on P-stranding in wh-movement, requiring pied-piping due to its rich case morphology, which demands that the wh-phrase match the case assigned by the preposition. For instance, the question "With whom is Anna talking?" is grammatical as Me poion milá i Ánna? but ungrammatical as Poion milá i Ánna me? (lit. "Whom talk-PRES.1SG the Anna with?").[18] This prohibition arises because Greek prepositional phrases function as islands at Logical Form (LF), preventing DP extraction, and E-chain uniformity requires categorical feature matching between the wh-phrase and its trace.[18]In Spanish, P-stranding is generally disallowed in formal registers of wh-movement, with pied-piping as the preferred strategy, though some dialects exhibit limited variation. The standard form for "With whom are you talking?" is ¿Con quién hablas? rather than the ungrammatical ¿Quién hablas con?.[19] This pattern stems from parametric properties like D-to-P incorporation, where the determiner-like wh-word incorporates into the preposition, forcing joint movement, and the absence of reanalysis rules that could reassign case to allow stranding in languages like English.[19]Standard Arabic similarly prohibits P-stranding in wh-movement, favoring pied-piping, with resumptive pronouns frequently employed as an alternative strategy to avoid extraction gaps. For example, "With whom do you speak?" is expressed as Maʿa man tatakallamu? (pied-piped) rather than the ill-formed Man tatakallamu maʿa?, or mitigated via resumption as man tatakallamu maʿa-hu? ("whom speak-2SG with-him?").[20] The disallowance is linked to Arabic's morphological case system on prepositional objects, which precludes stranding by disrupting case government, alongside broader parametric settings that treat PPs as bounding nodes for movement.[20]
P-stranding in Sluicing
Behavior in Allowing Languages
In languages that permit preposition stranding under wh-movement, such as English and Danish, this pattern extends to sluicing, where the elided clause preserves the stranding configuration without requiring pied-piping of the preposition.[21]In English, preposition stranding is preserved in sluicing constructions, allowing the wh-remnant to appear without the preposition, as in the example: "She talked to someone, but I don't know who [she talked to _ ]." This structure implies that the wh-phrase has moved from the object position of the preposition, leaving a trace in the elided tense phrase (TP), while the preposition remains in its base position within the deleted material.[21]Similarly, Danish retains preposition stranding under sluicing, with the preposition optionally realized before the remnant but typically omitted to reflect the stranded configuration, as in: "Hun talte med nogen, men jeg ved ikke hvem [med _ ]." Here, the stranded preposition "med" (with) is not deleted or required to pied-pipe with the wh-word "hvem" (who), mirroring the language's tolerance for stranding in non-elliptical contexts.[21]The underlying mechanism involves wh-movement of the remnant to the specifier of CP, followed by deletion of the TP, which contains the trace of the moved wh-phrase and the preposition in its governed position; the stranding survives as a remnant effect because the deletion targets the TP but leaves the CP structure intact.[21]Empirically, these stranding patterns in sluicing exhibit high acceptability among speakers of English and Danish, as evidenced by grammaticality judgments in linguistic studies, and they are frequently attested in corpora of natural language use, contrasting with the fuller, less elliptical wh-questions where the preposition is overtly visible.[21][1]
Behavior in Disallowing Languages
In languages that prohibit preposition stranding under wh-movement, such as Spanish and Arabic dialects, sluicing permits stranding in a manner that reveals a diagnostic asymmetry between overt extraction and ellipsis. This partial repair effect arises because the elided structure in sluicing licenses configurations that violate surface syntactic constraints, effectively "hiding" the stranded preposition within the deleted material. Such behavior underscores sluicing's role in probing underlying syntactic operations across languages. However, analyses differ by dialect, with some attributing apparent stranding to pseudo-sluicing (e.g., wh-clefts) rather than true ellipsis.In Spanish, stranding is disallowed in wh-questions, where pied-piping is obligatory (e.g., *¿Habló con quién? is ungrammatical, requiring ¿Con quién habló?). However, under sluicing, stranding becomes acceptable, as in María está hablando con alguien, pero no sé quién, where the wh-phrase appears without the preposition. This construction is often realized with inversion for naturalness (e.g., no sé quién), and acceptability shows dialectal variation, with some speakers preferring pied-piped forms like no sé con quién in certain regions. Child language data further support this repair, showing no significant preference between overt and omitted prepositions in sluices, consistent with ellipsis licensing the structure.[22]In Arabic dialects like Hijazi, wh-movement bans stranding (e.g., requiring pied-piping with prepositions like maʕa 'with'), yet sluicing allows it, as evidenced by empirical judgments showing higher acceptability for stranded sluices than non-elliptical counterparts. This pattern holds in varieties including Libyan and Emirati Arabic. In contrast, for Omani Arabic, apparent stranding in examples like Zaid raḥ maʕa ḥad, lakin ma aʕraf man ('Zaid went with someone, but I don't know who') is argued to derive from pseudo-sluicing structures rather than true preposition stranding. This emergent stranding in allowing dialects may be facilitated by focus prosody that highlights the wh-remnant, reducing perceptual constraints on the ellipsis site.[23][24]These patterns indicate that sluicing relaxes overt syntactic restrictions, enabling stranding as a tool to diagnose movement derivations. Merchant's diagnostics highlight such cross-linguistic asymmetries, where ellipsis repairs locality violations or case mismatches that block stranding elsewhere, though the exact mechanism—whether PF-deletion or alternative sources like clefts—remains debated in non-stranding languages.
P-stranding in Other Constructions
Relative Clauses
In English, preposition stranding is common in restrictive relative clauses, particularly when the relative pronoun functions as the object of the preposition, as in the example "The man (who) I spoke to _," where the preposition "to" is left at the end of the clause.[25] This construction is preferred over pied-piping (e.g., "The man to whom I spoke") in informal speech and spoken English, reflecting its higher frequency in casual registers compared to formal writing.[25][3]In French, preposition stranding in relative clauses is rare and generally disallowed in standard usage, which traditionally requires pied-piping, as in "L'homme à qui j'ai parlé" (The man to whom I spoke).[26] However, it emerges in colloquial varieties, especially in dialects like Quebec French or informal spoken registers, with examples such as "L'homme que j'ai parlé à _" (The man who I spoke to _), often under influence from English contact.[27]Preposition stranding typically occurs in object relative clauses, where the relative pronoun is the object of the preposition, whereas it does not apply in subject relative clauses, as the extracted element is the subject rather than a prepositional object (e.g., "The man who spoke _" involves no preposition stranding). Processing ease also plays a role, with stranding favored in simpler, less embedded structures to reduce cognitive load during comprehension.[3]Typologically, preposition stranding in relative clauses occurs more frequently in head-initial languages, such as those in the Germanic family (e.g., English), where verb-object order aligns with preposition-complement sequencing that facilitates stranding under movement.[4]
Pseudopassives
Pseudopassives are passive constructions derived from intransitive verbs followed by prepositional phrases, where the object of the preposition is promoted to subject position, stranding the preposition. In English, this is permitted, as in the sentence "This bed was slept in," where the preposition "in" is stranded after passivization of the intransitive verb "sleep."[28] Here, the prepositional phrase functions equivalently to a direct object, allowing the noun phrase to undergo A-movement to subject position.[29]Theoretically, pseudopassives are analyzed as passives of intransitive verbs combined with prepositions, where the verb-preposition complex undergoes reanalysis to form a unit eligible for passivization. Noam Chomsky's 1981 framework in Government and Binding theory accounts for this by positing a syntactic reanalysis rule that treats the V+P as a complex verb, enabling the preposition's object to move while stranding the preposition.[29][30] This analysis highlights the structural parallels between pseudopassives and standard passives, though it has been critiqued for not fully capturing semantic constraints.[28]Pseudopassives in English are restricted to certain verbs and contexts, particularly those implying affectedness or characterization of the subject. For instance, constructions with verbs like "live in" are acceptable when the subject (e.g., "The house was lived in") is directly affected by the action, but not with "arrive at," as in the ungrammatical "*Boston was arrived at," since arrival does not characterize or affect the location in the required way.[31] Unergative verbs (e.g., "sleep," "live") permit this stranding more readily than unaccusative ones (e.g., "arrive," "die"), reflecting lexical and semantic conditions on passivization.[31]In contrast, French marginalizes or disallows pseudopassives with preposition stranding, as seen in the ungrammatical "*Ce lit a été dormi dans," where speakers prefer active alternatives like "On a dormi dans ce lit." This absence stems from French's stricter constraints on preposition stranding and passivization of prepositional objects, differing from English's more permissive syntax.[12] Richard Kayne's analysis attributes this to parametric differences in case assignment and movement, where French prohibits the necessary reanalysis for stranding in such constructions.[12]
Directional and Split Constructions
In directional prepositional phrases (PPs), Dutch allows preposition stranding particularly with motion verbs, where the R-pronoun replaces the complement, leaving the preposition stranded. For instance, in the sentence Daar loopt hij naartoe ("There he walks to"), the directional preposition naar ("to") is stranded after the verb, with the R-pronoun daar fronted as an adverbial element.[32] This construction arises through adverbial stranding, where the R-pronoun inserts into the PP structure, reflecting the fine-grained syntax of spatial and directional elements.[32]Similarly, German exhibits split constructions in directional PPs, often involving R-pronouns and influenced by verb-second (V2) word order, which separates the preposition from its complement. An example is Da hinein läuft er ("There into he runs"), where the R-pronoun da fronts, stranding the directional element hinein ("into") after the finite verb in a main clause.[32] This mechanism relies on R-pronoun insertion, treating the stranded part as a remnant of a complex PP, and is common with motion verbs expressing directionality.[17]Compared to English, where preposition stranding is more broadly permitted but less tied to R-pronouns, continental Germanic languages like Dutch and German demonstrate greater flexibility in these directional splits due to their extended projection of adpositional structures and V2 constraints.[32][17] R-pronouns, such as daar/er in Dutch and da in German, facilitate this discontinuity by encoding locative or directional reference without full pied-piping of the preposition.[32]
Cross-linguistic Patterns
Germanic Languages
Preposition stranding, the syntactic phenomenon where a preposition is separated from its complement by movement operations such as wh-movement or passivization, is widely attested across Germanic languages, with patterns influenced by their shared Proto-Germanic origins and subsequent divergence in case systems and analytic tendencies.[33] The historical development of stranding tolerance traces back to early Germanic varieties, where it was often restricted to contexts with invariant complementizers like Old Englishþe or Old Icelandic er, but expanded in analytic descendants following the loss of inflectional case marking from Proto-Germanic, which diminished the morphological cues provided by prepositions.[33][34] This erosion, particularly pronounced in Middle English and continental North Germanic, facilitated reanalysis of prepositional phrases (PPs) as non-phasal units, enabling freer extraction.[33]In English and the Scandinavian languages, such as Danish, Norwegian, and Swedish, preposition stranding exhibits high acceptance rates in constructions like wh-questions, relative clauses, and pseudopassives, reflecting their predominantly analytic syntax with reduced case distinctions.[7] For instance, English allows stranding freely, as in "Who are you talking to?" rather than the pied-piped "To whom are you talking?" Corpus analysis from the British National Corpus reveals stranding in 59% of relevant cases overall, rising to 100% in spoken data and approximately 51% in written wh-movement contexts, underscoring its prevalence in informal and analytic environments.[2] Similarly, Danish and Swedish permit stranding in wh-constructions (e.g., Swedish "Vilka äpplen pratar du om?" – "Which apples are you talking about?") and pseudopassives (e.g., Danish equivalents of "The beds have been slept in"), though at somewhat lower frequencies than English due to residual synthetic elements.[7][35]West Germanic languages like Dutch and German display more restricted and variable patterns, often favoring pied-piping or resumptive strategies over direct stranding, though colloquial varieties show greater flexibility. In Dutch, standard usage avoids stranding through R-pronouns (e.g., "waarop" for "on which"), which incorporate the preposition to form complex wh-elements like "Waarop baseer je je mening?" ("On what do you base your opinion?"), effectively preventing PP separation.[16] However, spoken northern Dutch dialects commonly allow stranding without R-pronouns, as in "Wie praat je over?" ("Who are you talking about?").[16] German similarly prioritizes pied-piping in formal registers (e.g., "Womit hast du das gemacht?" – "With what did you do that?"), but permits colloquial stranding or "da"-compounds (e.g., "Worum geht's?" – "What's it about?") in spoken and dialectal contexts, with studies indicating stranding in under 10% of formal corpus instances but up to 40% in informal spoken data from northern dialects.[36][37] These divergences highlight how retention of case marking in West Germanic sustains synthetic preferences, contrasting with the analytic leniency in English and Scandinavian.[38]
Romance Languages
In Romance languages, preposition stranding is generally disallowed, with pied-piping being the predominant strategy in wh-movement constructions such as questions and relative clauses.[39] This restriction stems from the languages' morphological properties, where prepositions often fuse with wh-elements or pronouns, as seen in standard forms like French à qui ("to whom") or Spanish con quién ("with whom").[40] However, exceptions appear in colloquial varieties and contact situations, creating a gradient of acceptability across the family.In French and Spanish, pied-piping remains the norm in formal registers, but emerging stranding occurs in spoken dialects influenced by bilingualism. For instance, Quebec French permits stranding in questions like Quoi t'as mangé avec? ("What did you eat with?"), attributed to contact with English.[26] Similarly, in U.S. heritage Spanish among simultaneous English-Spanish bilinguals, stranding is accepted at rates up to 47% in acceptability judgments, as in ¿Quién bailó con? ("Who did they dance with?"), diverging from monolingual norms that reject it.[40] These patterns highlight how early bilingual exposure can introduce stranding into otherwise conservative systems.Italian and Portuguese exhibit similar blocks to stranding under regular wh-movement, with Italian fully disallowing it, as in the ungrammatical Che hai parlato di? ("What did you talk about?").[39] Portuguese follows suit in European varieties, but Brazilian Portuguese shows greater tolerance, particularly with lexical prepositions in colloquial speech, such as Que remédio que você não pode ficar sem? ("What medicine can't you do without?").[41] Both languages display higher tolerance for stranding under sluicing compared to overt wh-movement, often employing resumptive pronouns to evade the restriction.[27]This variation reflects the Romance family's Latin heritage, where rich inflectional morphology and preposition-pronoun contractions inhibit stranding, yet contact effects in bilingual regions—such as North American French or U.S. Spanish communities—increasingly promote it.[26] Typologically, Romance languages form a gradient scale, from conservative pied-piping in standard French and Italian to more permissive colloquial stranding in Brazilian Portuguese.[41]
Semitic and Other Languages
In Semitic languages such as Arabic, preposition stranding is generally disallowed under wh-movement, with prepositions required to pied-pipe the wh-phrase, as in standard non-stranding patterns observed across dialects like Hijazi and Saudi Arabic.[42] However, exceptions arise in sluicing constructions, where stranding becomes acceptable despite the prohibition in full wh-questions; for instance, Hijazi Arabic permits a stranded preposition in elliptical questions derived from clefts, aligning with resumption strategies common in the language.[23] This behavior is influenced by the clitic-like nature of prepositions in Arabic, which often affix to nouns or pronouns, restricting their independent stranding outside of ellipsis contexts.[43]Greek, an Indo-European language with Semitic-like adpositional features, predominantly requires pied-piping in constructions involving prepositions, as seen in examples like me ton opion ('with whom'), where the preposition cannot be stranded under wh-movement.[18] Standard Modern Greek adheres to the Preposition Stranding Generalization, disallowing stranding under sluicing unless the language permits it in overt wh-questions, though Cypriot Greek dialects show emerging tolerance for stranding in sluicing due to substrate influences or contact effects.[44] Limited evidence suggests slight increases in stranding acceptability in contemporary Standard Greek, potentially from English language contact in bilingual settings, but this remains marginal and ungrammatical in formal registers.[45]Among Slavic languages, preposition stranding is rare and typically avoided, with Russian exemplifying strict pied-piping requirements; however, certain "ambivalent adpositions" can postpose to create apparent stranding effects, as in idiomatic expressions like v otmestku ('in revenge'), where the preposition relocates after its dependent without true extraction.[46] In Asian languages like Japanese, which employ postpositions rather than prepositions, stranding is impossible due to the head-final structure and wh-in-situ strategies, ensuring postpositions remain attached to their hosts without movement-induced separation.[47]Research on preposition stranding in non-Indo-European families, particularly African languages, remains understudied, with typological surveys highlighting gaps in data from Bantu and Niger-Congo groups where adpositional systems vary widely but lack systematic analysis of stranding phenomena.[48] Post-2010s typological work has called for expanded cross-linguistic investigations to address these lacunae, emphasizing the need for comparative studies beyond Indo-European to refine universal constraints on stranding.[49]
Debates and Controversies
Prescriptivism in English
In the 18th century, prescriptive grammarians such as Robert Lowth in his A Short Introduction to English Grammar (1762) and Lindley Murray in English Grammar (1795) condemned preposition stranding as a vulgar or illiterate practice, drawing on Latin models where prepositions must precede their objects. Lowth illustrated the construction with examples he deemed faulty, arguing it deviated from classical norms, while Murray echoed this by altering examples to avoid stranding, reinforcing the view that it marred elegant prose.[50][51]The 20th century marked a significant shift toward tolerance, with H.W. Fowler's A Dictionary of Modern English Usage (1926) dismissing the ban as a "superstition" and defending stranding as idiomatic and preferable in many cases for clarity and rhythm. This perspective gained traction in style guides; for instance, The Chicago Manual of Style has permitted ending sentences with prepositions since its first edition in 1906 and explicitly endorses it in modern editions as natural English usage.[52][53]Today, preposition stranding is accepted across informal, journalistic, and most formal registers, with only hyper-formal or archaic contexts avoiding it; usage surveys among writers and editors indicate majority approval, around 55% in targeted polls, though broader linguistic consensus views it as standard.[54][55] This evolution has permeated popular culture through memes and the apocryphal quote attributed to Winston Churchill—"This is the sort of thing up with which I will not put"—which satirizes the rule's awkward enforcement and underscores its declining stigma.[56]
Theoretical Disputes
One major theoretical dispute in the analysis of preposition stranding concerns whether it constitutes a binary parametric choice or a more gradient phenomenon within generative grammar. Early parametric approaches, such as those in Chomsky's 1981 framework, treated stranding as a discrete option tied to broader principles like the Empty Category Principle (ECP), where languages either permit traces left by wh-movement to be governed appropriately (as in English) or require pied-piping to avoid ECP violations (as in French).[57] However, critiques of the ECP, notably from Kayne (1981) and Pesetsky (1982), argued that it inadequately captures cross-linguistic variation, proposing instead that stranding correlates with independent parameters like the absence of preposition cliticization or verb-particle constructions, rendering the ECP overly restrictive for explaining stranding without additional mechanisms.[58] More recent work challenges the binary view by advocating gradient parameters, where stranding acceptability varies continuously based on processing load, frequency in corpora, or probabilistic constraints, as evidenced in studies of English speakers' variable judgments on stranding versus pied-piping.[59][2]A related puzzle arises in sluicing constructions, where preposition stranding appears under ellipsis even in languages that prohibit it in non-elided wh-movement, prompting debates on the scope and mechanism of deletion. Merchant (2001) proposed the Preposition Stranding Generalization, asserting that stranding is permitted under sluicing if and only if it is allowed under overt wh-movement, attributing apparent repairs to wh-movement preceding TP-deletion, which conceals ECP-like violations at PF.[21] In contrast, Lasnik (2001) critiqued this by emphasizing PF deletion's role in hiding illicit structures but argued for narrower deletion scopes that do not fully repair island violations or stranding bans, suggesting that Merchant's generalization overpredicts cross-linguistic uniformity and fails to account for cases where sluicing does not ameliorate stranding in non-Germanic languages. This post-2000 exchange highlights ongoing tensions between movement-plus-deletion analyses and alternatives that limit repair to semantic or LF levels, with empirical tests from languages like Brazilian Portuguese showing partial counterexamples to Merchant's claim; for instance, studies on Emirati Arabic (as of 2014) demonstrate stranding under sluicing despite its prohibition in overt wh-movement, challenging the strict iff condition.[60][61]Cross-linguistic investigations reveal significant gaps in stranding patterns, particularly in underexplored families like Bantu, which challenge rigid universal grammar assumptions. Bantu languages such as Lubukusu and Kiswahili generally prohibit stranding, relying instead on resumptive pronominal clitics to license extraction from prepositional phrases, as direct traces would violate locality constraints akin to the ECP.[62][63] These patterns, absent in Germanic languages where stranding is robust, suggest that universal grammar may require revisions to incorporate resumptives as a core parametric option rather than a last-resort strategy, complicating binary head-complement parameterizations and calling for featural or micro-parametric refinements to accommodate Bantu's agglutinative morphology and agreement systems.[64][65]In the 2020s, advances in optimality theory have reframed stranding as an outcome of ranked constraints weighing movement costs against pied-piping faithfulness, offering a non-binary model for variation. For instance, analyses of wh-related prepositions in English posit that stranding emerges when constraints favoring remnant lightness outrank those prohibiting trace government, predicting gradientacceptability across contexts like interrogatives.[66] Such models, building on earlier OT applications, integrate cross-linguistic data by allowing language-specific rankings to penalize stranding in Romance or Bantu while permitting it in Germanic, thus addressing ECP critiques through violable universals rather than hard parameters.[67]
