Thomas Kyd is traditionally accepted as the author of The Spanish Tragedy (1587), Soliman and Perseda (1588), and Cornelia (1594), a translation of Robert Garnier’s French drama Cornélie (1573) (I have utilized Wiggins and Richardson 2012 and 2013 for the dates of first performances; all quotations in this article for The Spanish Tragedy, Soliman and Perseda, and Cornelia are from the Boas 1901 edition). In this essay I argue that the anonymous plays, The True Chronicle History of King Leir (1589) and Arden of Faversham (1590), can also be assigned to Kyd on the basis of parallelisms of thought and language, versification characteristics, and overall dramaturgy (all quotations in this article for King Leir come from Greg 1907; all quotations for Arden of Faversham come from Wine 1973). In particular, I focus on one aspect of Kyd’s drama: his use of ominous dreams. Before presenting the evidence for Kyd’s authorship of these plays, I shall need to elaborate on the methodologies I employ for identifying shared phrases between these texts.

Following Brian Vickers, who contends that “an author’s individuality will be more visible if we can identify his preferred groupings of words” (2008, 13), I originally employed anti-plagiarism software called WCopyfind (2018, developed by Lou Bloomfield, Professor of Physics at the University of Virginia), in order to highlight strings of words shared between old spelling versions of these plays, downloaded from Literature OnLine (2019), or LION. Heinz Dreher points out that although there are “many systems doing […] string-of-text matching,” WCopyfind is one of the most “readily available” and is “computationally very efficient” (2007, 604). This software has the advantage of producing results that are “objectively identified” (Jackson 2012, 161), as well as “complete, and replicable by any other scholar, all qualities that constitute properly scientific procedures” (Vickers 2018, 458). It can be set to highlight any specified n-gram (consecutive word sequence) length within a pair of electronic documents, from two words upwards, and can even identify – within a fraction of a second – approximate matching utterances (for example, some sequences might differ slightly in terms of syntactical arrangement or spelling) through adjusting the “Minimum % of Matching Words” and “Most Imperfections to Allow” values. Dennis McCarthy and June Schlueter also favour this software because it enables “the researcher to expose shared collocations and phrases that may have otherwise escaped notice” (2018, 12). The software, originally designed to expose student plagiarism, is not without disadvantages when it comes to the purpose of examining drama, however: it is impossible to compare results for all plays of the period concurrently and it is a painstaking process to check the rarity of each highlighted phrasal structure by using the search functions in Literature OnLine.

Fortunately, there has been a recent innovation, which has the potential to revolutionize corpus linguistic approaches to early modern drama: Pervez Rizvi (2018) has developed a publicly accessible electronic corpus of 527 play texts dated between 1552 and 1657, titled Collocations and N-grams. Searches of the normalized and lemmatized texts – drawn from Martin Mueller’s (2014) corpus Shakespeare His Contemporaries (subsequently renamed Early Modern Print. “Early Modern Print” 2018) and the Folger Shakespeare Library Editions (2018) website – allow a wider range of matches to be discovered than by searches using original spelling or unlemmatized forms of words. Rizvi’s results are fully automated and enable scholars to check for every contiguous word sequence (including lemmas) shared between plays. Rizvi’s database also enables users to search for all discontinuous word sequences (i.e. matching phrases separated by intervening words) within ten-word windows, something that WCopyfind is apt to miss, given that the software is designed to detect shared n-grams. It has all of the advantages of plagiarism software in terms of its objectivity and reproducibility, but it does not require time-consuming searches of databases like Literature OnLine. I was therefore able to supplement the results drawn from using WCopyfind and Literature OnLine with recourse to Rizvi’s Collocations and N-grams. The word sequences I present throughout this essay are rare in that they occur no more than five times within plays first performed on the London public stages during the period 1580–1600. These parallels adhere to David Lake’s categorizations of “combinations of more than one word” and “grammatical or semantic” patterns (1975, 14). Studies of “an author’s grammatical and especially syntactic practice” (Love 2002, 111) have, of course, long been a feature of this field and have led to many successes.

The methods described above help us to identify “idiolect markers,” which are “combinatorial, embedded in an author’s long-term memory, and repeated. We recognize them by unconscious pattern matching similar to what enables us to quickly make out a face in a crowd” (Lancashire 2010, 4). Vickers explains that:

Vickers’s parallel-checking method has been criticized by several scholars involved in The New Oxford Shakespeare edition (Jackson 2008; Egan 2012; Taylor 2014), which includes, erroneously in my view, Arden of Faversham (Taylor, Jowett, Bourus, Egan 2016; Taylor and Egan 2017). Here I aim to show that these attributions are more compelling than detractors would have us believe, and that arguments for an “expanded” Kyd canon have been anticipated and developed by several scholars. My analysis involves both quantitative and qualitative research methods, rather than the purely statistical approaches that can be found in many recent essays concerning authorship attribution studies. We should bear in mind the caveat that statistical analysis, like literary analysis, can aspire to an objective viewpoint, but it also relies upon subjective interpretation.

Edmond Malone was the first scholar to suspect that Kyd was “the author of the old plays of Hamlet, and of King Leir” (Boswell 1821, 2.316), while later, in 1891, F. G. Fleay proposed Kyd and Thomas Lodge as authors of King Leir (1891, 2.52). John Mackinnon Robertson asserted in 1914 that there was “some reason to think” Kyd (albeit recasting a play written by Lodge, as suggested by Fleay in 1891) was the play’s author (1914, 109). Robertson expanded on his attribution in 1924, arguing that the “play is ascribable to Kyd on the score” of “the naturalness of the diction […] the orderly planning and complication of the action throughout,” and “the frequent parallelism both in action and in phrase to those of Kyd’s ascertained plays” (1924, 387).

William Wells also argued for Kyd’s authorship of King Leir, for it is “a play of simple, undisguised realism, with few flights of fancy. Its sentiment is extraordinarily naive, in content and expression, and yet, in its way, powerful. This accords with Kyd’s characteristics” (1939, 434). We might question the somewhat pejorative tone of Wells’s statement here, but he was right to dismiss arguments for Lodge as part author, for “the style of Leir is uniform throughout, one poet alone is involved” (437). He pointed out that King Leir is “abounding in feminine endings, and this points directly to Kyd, for none but he, among the pre-Shakespearian dramatists, wandered far from the normal ten-syllable line” (438). In his 1931 study of eleven-syllable verse lines, Philip Timberlake recorded an average of 10.8% feminine endings in King Leir, which corresponds to the 10.2% for Soliman and Perseda, and 9.5% for Cornelia (61–2). Kyd is the only known dramatist preceding Shakespeare who comes close to the proportion of feminine endings in King Leir; incidentally, Arden of Faversham averages 6.2% feminine endings. Timberlake’s figures therefore rule out other authorship contenders, such as Robert Greene (whose plays have a range of 0.1–1.6), George Peele (1.5–5.4), and Christopher Marlowe (0.4–3.7), whose percentages, within their dramatic works, are manifestly too low. The Russian-American prosodist Marina Tarlinskaja points out that “King Leir could be attributed to Kyd on the basis of feminine endings alone” (2014, 105).

Paul V. Rubow, having identified numerous parallels of thought, language, and corresponding plot features, also ascribed King Leir to Kyd in 1948 (145–55). Recently, Vickers has utilized anti-plagiarism software to highlight “any identical three-word sequence” shared between King Leir and The Spanish Tragedy. Vickers “checked each match against” a “corpus of 54 plays performed in the public theatres before 1595” (2015). In total, he has identified 96 collocation (word association) matches that first appear in King Leir and the works of Kyd, which provides linguistic evidence for common authorship. Gabriel Egan (2009), citing MacDonald P. Jackson’s (2008) paper “New Research on the Dramatic Canon of Thomas Kyd” has called Vickers’s method of collecting highly distinctive collocational clusters of words “useless”:

However, Jackson’s statistics are based on Vickers’s crude counts for matches, during his preliminary researches in 2008. For example, Jackson notes that Vickers counts 31 Kyd matches with King Leir (Jackson 2008, 114), but Vickers has since amassed a far more impressive number of unique matches between King Leir and The Spanish Tragedy alone. Here I should like to dispute Jackson’s claim that “[t]he canonical Kyd plays are utterly different from the plays putatively Kyd’s” in terms of quantities of verbal matches (2008, 121).

Martin Mueller’s electronic corpus Shakespeare His Contemporaries consists of over 500 plays dated between 1552 and 1662. In 2009, Mueller applied a series of statistical tests to the putative Kyd texts that convinced him: “Vickers is right about the Leir play” and “Arden” (2009a). In a blog post entitled “N-grams and the Kyd canon: a crude test,” on his (then) website Digitally Assisted Text Analysis, Mueller explained that he “ran an experiment on 318 early modern plays in the MONK corpus” and “extracted lemma n-grams from bigrams to heptagrams that were repeated at least once.” He computed “their distribution across plays” and discovered that King Leir and Soliman and Perseda are placed above the median (the number separating the higher half of Mueller’s data from the lower half) – with a percentage of 96.5 – for play pairs suggesting “characteristic patterns of authorial usage” (2009b), in comparison to all other candidate playwrights. We might note that this percentage is higher than that found for the accepted Kyd play pair Soliman and Perseda and Cornelia (93.5%). Mueller also demonstrated that Soliman and Perseda and Arden of Faversham are placed “in the top quartile for shared two-play n-grams by the same author,” with a percentage of 99.7, while Arden of Faversham and King Leir are given a percentage of 99, which provides compelling quantitative evidence for common authorship of these texts (Mueller 2009b).

In another blog post titled “Vickers is right about Kyd,” Mueller applied “Discriminant Analysis to lemma trigrams” (three-word sequences) “that occur at least 500 times in 318 early modern plays,” which “misclassifies 50 or 16% of 318 plays. It gets 84% right. Of 37 plays by Shakespeare, it gets 34 right” (2009a). Discriminant Analysis, which establishes “variance between groups on the basis of the combined effect of multiple variables,” assigned The Spanish Tragedy to Kyd with a 96.1% chance, while Soliman and Perseda and Cornelia were given percentages of 85.3 and 79.7 respectively. Mueller also applied these tests to plays newly attributed to Kyd: Discriminant Analysis assigned King Leir to Kyd with a 99.3% chance, while Arden of Faversham was given a 97.4% chance of having been written by Kyd, once again in comparison to all other candidate dramatists of the period, including Shakespeare. Mueller concluded that “Discriminant Analysis very strongly confirms” that these plays come “from the same stable” as the three accepted Kyd plays, and “If you combine my evidence from common trigrams” with the evidence “from rare shared repetitions, you would have to be very sceptical about the power of quantitative analysis not to acknowledge the fact that the claim for an expanded Kyd canon rests on quite solid evidence” (2009a).

Mueller (2014) has also pointed out that plays by the same author are likely to share twice as many rare n-grams consisting of at least four words as plays by different authors. Indeed, Arden of Faversham shares more dislegomena (phrases that recur in only one other play) of this type with Soliman and Perseda than any other play in Mueller’s corpus, and King Leir shares more unique tetragrams-plus (four-word units or higher) with the domestic tragedy than any Shakespeare text (Freebury-Jones 2015; Mueller 2018).1 In order to further determine whether the contested plays are, as Jackson and Egan have suggested, unlike Kyd’s accepted plays in terms of quantities of verbal matches, I tested these texts for contiguous sequences of four or more words and adjusted my figures according to composite word counts. However, using Jackson’s own criteria, I recorded only tetragrams-plus that occur “not more than five times in drama of the period 1580–1600” (2014, 219). In my initial round of analysis, I used WCopyfind to highlight strings of words shared between old spelling versions of these texts, drawn from Literature OnLine, and checked the rarity of the highlighted phrasal structures by searching plays written during the specified time period in LION. I subsequently double-checked results by searching Pervez Rizvi’s publicly accessible electronic corpus, Collocations and N-grams, so that these findings are entirely reproducible. Rizvi points out in a paper published on his website titled “Which N-grams are the Best?,” that “unique 3-grams” and “4-grams are the best authorial markers” (Collocations and N-grams 2018). Rizvi tested uncontested plays in his corpus and discovered that tetragrams “correctly attribute 82 out of 86 plays, while 5-grams are not far behind, with 80 out of 86 correct.” In another document on his website titled “Arden of Faversham and the Extended Kyd Canon,” Rizvi discovered that the three accepted Kyd plays and Arden of Faversham as a whole are all assigned to Kyd according to quantitative analyses of both unique trigrams and tetragrams (Rizvi 2018).

I discovered 21 rare tetragrams-plus between The Spanish Tragedy and Cornelia (adjusting this total according to the combined word count of these texts gives us a percentage of 0.06 matches); 12 rare word sequences between Soliman and Perseda and Cornelia (0.03 matches); and 36 sequences shared by The Spanish Tragedy and Soliman and Perseda (0.09 matches).2 By testing the accepted Kyd texts, I provided a negative check that enabled me to compare results for the disputed plays. My tests demonstrate that King Leir is at one with the other Kyd plays in terms of the quantity and quality of matching phrases. King Leir more or less shares the same number of matches with The Spanish Tragedy (0.09 matches) and Cornelia (0.04 matches) as Soliman and Perseda does. King Leir also shares 30 tetragrams (0.07 matches) with Soliman and Perseda.

Significantly, Arden of Faversham shares a remarkable 47 rare tetragrams-plus with Soliman and Perseda (0.12 matches); 32 tetragrams with The Spanish Tragedy (0.07 matches); and 14 tetragrams (0.04 matches) with Cornelia. Furthermore, the Kentish tragedy shares 41 rare word sequences with King Leir (0.09 matches). Below, I rank these putative Kyd play pairs according to the percentages for rare shared tetragrams-plus:

The Spanish Tragedy, Soliman and Perseda, and Cornelia are essentially Senecan tragedies, whereas King Leir (a comedy) and Arden of Faversham (a domestic tragedy) are largely based on Raphael Holinshed’s Chronicles of England, Scotland, and Ireland (1587). It is also worth noting that Kyd’s Turkish tragedy is an extended version of the play-within-the-play from The Spanish Tragedy, and the verbal evidence I have collected suggests that Kyd recycled phrases from this moment in his earlier tragedy when he came to write Soliman and Perseda. It is therefore striking that verbal relations between Arden of Faversham and Soliman and Perseda exceed any other play pair in the “extended” Kyd canon, and that the number of matches between King Leir and Arden of Faversham exceeds that of The Spanish Tragedy and Soliman and Perseda. The low rankings for play pairs involving Cornelia are accounted for by the fact that that play is a rather free translation from Garnier’s French, but it is worth noting that Arden of Faversham and King Leir share more matches with the closet drama than Soliman and Perseda does, despite generic differences. Readers may consult these matching phrases, the large majority of which have never been highlighted before, in the appendix (“Rare Tetragrams Plus Shared between Plays Attributable to Kyd”) to this essay.

The battery of computational tests above encompasses short word strings (such as two-word units, known as bigrams), longer strings of words (tetragrams-plus), as well as both high-frequency and rare word sequences. Fundamentally, each parameter of analysis suggests it is linguistically conceivable that these five texts were written by the same author, which contradicts Jackson’s claim that “The canonical Kyd plays are utterly different from the plays putatively Kyd’s” (2008, 121).

Jackson (2014) has argued in his monograph Determining the Shakespeare Canon: Arden of Faversham & A Lover’s Complaint that Arden of Faversham is a collaborative play written by Shakespeare and an unknown dramatist who was probably not Kyd.3 Having spent several years testing the Kyd ascriptions and scrutinizing Jackson and his New Oxford Shakespeare colleagues’ counterarguments, I have reached the conclusion that the domestic tragedy was written solely by Kyd, and that verbal links with Shakespeare’s plays are indicative of Kyd’s influence on Shakespeare, as opposed to Shakespeare’s authorship. For instance, in a study conducted in 2016 I tested Scene Eight, which Jackson ascribes to Shakespeare, for collocations shared with the remainder of the play (Freebury-Jones 2016). Plagiarism software highlighted a considerable number of distinct parallelisms of thought and language, many of which were unparalleled in Shakespeare’s corpus, but could be detected in Kyd’s plays. I argued that these findings were indicative of Kyd’s sole authorship. More recently, while exploring the notion of Shakespeare’s aural, or “actor’s memory,” I demonstrated that verbal relations (specifically tetragrams-plus) between Shakespeare plays and Arden of Faversham were in accordance with other texts ascribed to Kyd (Freebury-Jones 2017), thus countering Jackson’s assumption that the domestic tragedy was “more influential than all the other plays in which” Shakespeare may have “acted” (Jackson 2014, 24). Furthermore, having adjusted Jackson’s (2008) figures for trigrams shared between Henry VI Part Two (1591), The Taming of the Shrew (1592), and scenes he assigns to Shakespeare and an unknown co-author in the Kentish tragedy, I showed that the distribution of unique matches does not support Jackson’s divisions of authorship, nor his claim that the play is co-authored. Elsewhere, I have shown that the pausation practices in Arden of Faversham; as well as the figures for feminine endings; the distribution of exclamations and colloquialisms; the prosodic features of high-frequency words; and the frequency and inventiveness of compound adjectives, reveal homogeneity between the portions Jackson assigns to Shakespeare and an unknown co-author (quite unlike the results I have obtained from controlled experiments on Shakespeare collaborations, as part of these studies), and are commensurate with Kyd’s sole authorship (Freebury-Jones 2018a, 2018b, Forthcoming 2019).

The case for Kyd’s authorship of Arden of Faversham has been made by generations of scholars. F. G. Fleay proposed Kyd as the play’s author in 1891 (29), as did Charles Crawford in 1903. Crawford observed that the play “echoes all parts of Kyd’s work; and, therefore, it is a difficult thing to make choice of illustrations, there being such an abundance of material to substantiate his claim to the play” (1906, 120). Having listed 50 verbal matches between the domestic tragedy and The Spanish Tragedy and Soliman and Perseda, he concluded that “the parallels from Marlowe and Lyly are of an entirely different character from those I have adduced from Kyd himself. I assert, then, that Kyd is the author of Arden of Feversham” (Crawford 1906, 130). In 1907 Walter Miksch, having studied the stylistic, metrical, and rhetorical features of Arden of Faversham in comparison to The Spanish Tragedy and Soliman and Perseda, ascribed the play to Kyd. He listed almost a hundred verbal matches between these texts (Miksch 1907, 19–29). The following year, C. F. Tucker Brooke agreed with Fleay and Crawford that “there are more parallels in feeling and expression between” the “play and the tragedies of Kyd than coincidences will account for” (1908, xv). H. Dugdale Sykes, in 1919, identified additional collocation matches, including some with Kyd’s Cornelia. He argued that “this play has rightly been assigned to Kyd,” for “the resemblances between Arden and the unquestioned work of Kyd extend to the most trivial details of phrasing and vocabulary, and the whole weight of the internal evidence supports the conclusion that it is the product of Kyd’s own pen” (48–9).

I now extend these studies through exploring ominous dreams in Kyd’s three undoubted plays, and the verbal and dramaturgical correspondences between these passages, before comparing the use of ominous dreams in King Leir and Arden of Faversham. I suggest that my phraseological findings (many of which have not been discovered by previous scholars) give us an insight into Kyd’s cognitive processes, and thus provide additional evidence for his authorship of these texts.

Prophetic, premonitory, or precognitive dreams and/or visions can be found in the works of Seneca: Agamemnon opens with the ghost of Thyestes foreseeing Agamemnon’s death, while Andromache has an ominous dream in which she is visited by the ghost of Hector in Troades. Conversely, many Elizabethan writers argued that there was little to no validity in the notion that dreams could impart “special knowledge” (Alwes 1996, 153). For example, Thomas Nashe, in his pamphlet The Terrors of the Night (1594), contended that such dreams were “nothing else but a bubbling scum or froth of the fancy” (Steane 1972, 218). Francis Bacon asserted that the tales of divine prophecies throughout history had been “by idle and crafty brains merely contrived and feigned, after the event past” (Vickers 1996, 414). Prophetic dreams, however, are given some credibility in Kyd’s plays.

For instance, William Wells noted in 1939 that “Kyd’s characters almost invariably anticipate disaster by dreams or premonitions,” and that it is “probable that it was from Garnier that Kyd derived the feature” (435–6). Indeed, Kyd seems to have been influenced by Garnier long before he came to translate his closet drama; the General’s Speech in Act One Scene Two of The Spanish Tragedy was modelled partly on Garnier’s description of the battle of Thapsus in Cornélie. Garnier was in turn influenced by Seneca’s revenge tragedies, while Kyd, as Lukas Erne points out, “claims Seneca emphatically as his ancestor” (2001, 81). Kyd’s drama, like Seneca’s, adheres to a providential design. Miles S. Drawdy explains that “providentialism refers to the belief” that “the earth—indeed the cosmos—is divinely ordered and that this order, when sensitively interpreted, reveals, at least in part, the divine will” (2014, 2). Prophetic dreams afford Kyd’s characters an insight into the fates prescribed to them by cosmic forces. In The Spanish Tragedy, the Viceroy, who has been told (falsely) by Villuppo that his son has been slain in battle, says, “I, I, my nightly dreames have tolde me this” (Sp. T., 1.3.76). Similarly, Bel-imperia anticipates her lover Horatio’s murder in the line, “my hart foretels me some mischaunce” (2.4.15). Horatio attempts to placate Bel-imperia by dismissing her premonition: “Sweet, say not so; faire fortune is our freend” (Sp. T., 2.4.16). Kyd repeated this process (while drawing from his individual store of verbal formulae, in order to fulfil the same dramatic purpose) of having a character imparting a dream or vision to a confidant, who subsequently dismisses their interpretation of the premonition, throughout his dramatic career.

In Soliman and Perseda, the play’s heroine, Perseda, upon hearing of her lover Erastus’ murder, repeats the Viceroy’s line from The Spanish Tragedy:

almost verbatim:

Verbal formulations of this kind, repeated in recurring dramatic situations, are indicative of Kyd’s idiosyncratic lexicon of collocations. As this example suggests: “A writer’s memory can recall extensive linguistic units together with their prosodic frame. Kyd often recalled the physiognomy of a verse-line, so to speak, the placing of key words at the beginning and/or ending” (Vickers 2018, 452–3). Verbal patterns that recur in the same prosodic positions can therefore provide useful evidence for Kyd’s hand.

In the third act of Cornelia, the eponymous character relates a dream in which the ghost of Pompey visited her. The dream is prophetic in that Pompey tells Cornelia her father is also dead, which is confirmed by the Messenger in the play’s final act. Erne suggests that Kyd’s “aim seems to have been to impregnate his text with a stranger sense of the supernatural than he found in the original” (2001, 210). Cornelia cries, “My fearfull dreames do my despairs redouble” (Corn., 3.1.61), to which the Chorus respond, “Why suffer you vayne dreames your heade to trouble?” (Corn., 3.1.62). Like Horatio, the Chorus attempt to conciliate the dreamer with a dismissal of the dream’s significance. Nevertheless, the dream has such an impact on Cornelia that she still thinks she can see the ghost of Pompey, even when awake:

Kyd (by my argument) would also employ this device of having a character suffer from hypnopompic delusions in King Leir and Arden of Faversham.

At the beginning of King Leir, the eponymous character announces that “me thinks, my mind presageth still/I know not what; and yet I feare some ill” (KL, 3.216–7). These lines parallel, in both thought and language, Bel-imperia’s speech: “I know not what my selfe:/And yet my hart foretels me some mischaunce” (Sp. T., 2.4.14–15). As Muriel St. Clare Byrne put it in a classic essay: “Quality is all-important” in studies of verbal matches, and “parallelism of thought coupled with some verbal parallelism” provides stronger evidence for common authorship than “verbal parallelism” alone (1932, 24). Close study of shared phraseology therefore supplements the quantitative arguments above.

In Scene Nineteen, the King wakes from a prophetic dream, in which his:

Leir, like Bel-imperia and Cornelia, imparts his premonition “to a friend,” who dismisses “any deductions” he “may have drawn” (Wells 1939, 436). He tells Perillus that “with the feare of this I did awake,/And yet for feare my feeble joints do quake” (KL, 19.1500–1), which is comparable to the effect Cornelia’s dream has on her: “When drousy sleep, that wak’d mee at unwares,/Dyd with hys flight unclose my feareful eyes/So suddainly, that yet mee thinks I see him” (Corn., 3.1.123–5). Leir’s dream is fairly accurate, for Ragan has sent the Messenger to slay him. Nevertheless, Perillus reassures him, “Feare not, my lord, dreames are but fantasies,/And slight imaginations of the brayne” (KL, 19.1481–2), which is akin to the Chorus’s response to Cornelia: “Why suffer you vayne dreames your heade to trouble?” (Corn., 3.1.62). Leir’s premonitions are symptomatic of the “divine theatre in which” Kyd’s characters “enact the roles they have been prescribed by God” (Erne 2001, 96). Leir is only saved from his dream becoming total reality by the intervention of thunder, which causes the Messenger to flee. Leir and Perillus attribute this storm to a divine power: the Messenger exits and Perillus says, “Let us give thanks to God” (KL, 19.1759). Kyd’s characters can often be seen “testifying to a just and omnipotent God,” which follows a “providential logic” (Drawdy 2014, 38).

We can see, therefore, that the dream sequence in King Leir offers structural and dramaturgical parallels with Kyd’s accepted plays. My evidence suggests that this scene also derives from the same author’s linguistic resources. For example, Leir says, “I marvell, that my daughter stayes so long” (KL, 19.1476), which recalls, “I wonder that his Lordship staies so long” (Sp. T., 3.4.30). We find the common phrase, “God graunt,” in Leir’s line, “God graunt we do not miscarry in the place” (KL, 19.1478), but there is a unique association of this line-opening with the topic of dreams, shared with Cornelia’s line, “God graunt these dreames to good effect bee brought” (Corn., 3.1.65). We might compare Leir’s description of his daughters “Ech brandishing a Faulchion in their hand” (KL, 19.1490) with the Messenger’s report of the Battle of Thapsus in Cornelia, and the Roman nobles “with their fauchins in their fists” (Corn., 5.1.307). We also find the phrase, “bleeding wounds,” shared between Leir’s account of the dream, in which his daughter Cordella “Came with a boxe of Balsome in her hand,/And powred it into my bleeding wounds” (KL, 19.1496–7), and Hieronimo’s account of finding his son’s corpse: “Within the river of his bleeding wounds” (Sp. T., 4.4.124). The singular form, “bleeding wound,” occurs in Cornelia, in the line, “And launc’d hys bleeding wound into the sea” (Corn., 4.1.24), which, following the grammatical pattern in King Leir, is accompanied by the line-opening conjunction “And,” a transitive verb, as well as a possessive pronoun (as in The Spanish Tragedy), and the preposition “into.” The line thus appears to be a variant of the same writer’s verbal formula. Having combined digital methodologies with close reading of these plays, I suggest that King Leir should be restored to Kyd’s canon.

I will shortly present the linguistic and dramaturgical links I have discovered between Arden of Faversham and the use of dreams in Kyd’s three accepted plays, as well as King Leir. Firstly, however, I should point out that supporters of Shakespeare’s part authorship of Arden of Faversham might object that the use of ominous dreams is not exclusive to Kyd, for there are comparable passages in Shakespeare plays. For instance, Clarence anticipates his death through a dream in Act One Scene Four of Richard III (1593), but the pattern is quite unlike Kyd’s plays, for Brakenbury commiserates with Clarence, rather than dismissing any deductions. Jackson suggests that Arden’s dream shares “a psychological subtlety” with Clarence’s monologue (2014, 55), and yet he records just one rare verbal link between these passages (237–9). Shakespeare seems to have been deeply conscious of Kyd’s use of dreams when he came to write this scene in Richard III. This play is obsessed with the topic of dreams: there are 26 occurrences of the word “dream” or its variants in the text, more than any other play in Shakespeare’s corpus. Act One Scene Four presents a fascinating example of Shakespeare’s infusion of source material. The scene appears to draw from Thomas Sackville’s The Mirror for Magistrates; Ovid’s Metamorphoses; Seneca’s tragedies; Spenser’s The Fairie Queen; Homer’s The Odyssey; and Virgil’s Aeneid, as well as Kyd’s plays, but not from any of Shakespeare’s historical sources. The dream-world that Clarence describes is reminiscent of Andrea’s prologue in The Spanish Tragedy, while the verbal texture of this scene recalls King Leir and Arden of Faversham, thus supporting Erne’s evaluative statement that “Shakespeare, perhaps more than anyone else, seems to have specifically profited from Kyd’s works” (2001, 5). As I have demonstrated elsewhere, plays that Shakespeare composed around this period, such as Henry VI Part Three (1591) and Titus Andronicus (1592), share a striking number of verbal correspondences with Kyd’s corpus (Freebury-Jones 2017). The emphasis on dreams in Richard III could therefore be the result of Shakespeare’s closely attending to Kyd’s dramaturgy in works written around this time. Shakespeare is unlikely to have written Richard III immediately after Henry VI Part Three, given the “remarkable difference in dramatic power and stylistic control” (Wells and Taylor 1987, 115) between these plays. The composition of Shakespeare’s revenge drama, Titus Andronicus, which is heavily influenced by Kyd’s dramaturgy, prior to Richard III, could therefore help to explain why the history play can be considered “a study in Kydian methods” (Wells 1940, 243).

Another similar Shakespearean example would be Act Two Scene Two of Julius Caesar (1599), in which Caesar speaks of Calpurnia’s menacing dream, only to have it interpreted favourably by Decius Brutus. However, the importance of chronology must be emphasized: Shakespeare’s play was written around a decade after Kyd’s accepted plays, which are near-contemporary to King Leir and Arden of Faversham. Also, the account in Shakespeare’s play appears to derive from Valerius Maximus, Plutarch, or Suetonius, and Jackson (2014) records no rare links between Arden and Calpurnia’s dreams.

Early in the Kentish tragedy, the servant Michael, speaking very much in the voice of Kyd’s Cornelia, has a “fearful dream that troubled me” (AF, 4.93), in which he “was beset/With murderer thieves” (4.94–95). These lines parallel Kyd’s heroine thus: “The fearefull dreames effect that trouble me” (Corn., 2.1.209). Franklin, disturbed by Michael’s cry, complains, “What dismal outcry calls me from my rest?” (AF, 4.87). This line replicates Hieronimo’s famous cry: “What out-cries pluck me from my naked bed” (Sp. T., 2.5.1). We might consider this to be a direct allusion to The Spanish Tragedy, but the fact that the author of Arden of Faversham couples this line with Hieronimo’s later image of the “dismall out-cry” (Sp. T., 4.4.109) suggests self-repetition on Kyd’s part. This line is also matched in, “O was it thou that call’dst me from my bed?” (Sp. T., 3.2.16), and in Soliman and Perseda: “What dismall Planets guides this fatall hower?” (S&P, 1.5.78). These four unique matches with a single line in Arden of Faversham lead me to concur with Sykes, who pointed out that Kyd was most likely “producing, in the phraseology natural to him, just such an incident as he had already used in his earlier play” (1919, 67). Michael’s declarative, “My trembling joints witness my inward fear” (AF, 4.95), gives us another unique match with the dream sequence of King Leir: “And yet for feare my feeble joints do quake” (KL, 19.1501). Thomas Arden is dismissive here: “So great a cry for nothing I ne’er heard” (AF, 4.97).

However, Arden anticipates his own murder through an ominous dream in Scene Six, in which “a toil was pitched” (AF, 6.7) by an evil “herdman” (6.17), and he compares his frightening nightmare to “one” (6.21) who “sees a lion foraging about” (6.22). This passage shares a similar hunting image thread to Haleb’s dialogue in Soliman and Perseda: “The ones a Lyon almost brought to death,/Whose skin will countervaile the hunter’s toile” (S&P, 1.5.41–42).4 Indeed, Scene Six of Arden of Faversham shares a large number of rare verbal links with Kyd’s undoubted plays and King Leir, unacknowledged by Jackson (2014) in his monograph. Jackson uses the database Literature OnLine to test the rarity of utterances that he considers worth recording. Jackson concedes that this process of determining “whether a parallel is close enough to be recorded” involves “an element of subjectivity” and that “no doubt some relevant data have been accidentally overlooked” (2014, 19). This process of picking out potentially significant phrases in each line is evidently time-consuming in comparison to the use of plagiarism software, which might account for why Jackson examines less than ten percent of the play in his monograph. My findings suggest that Jackson has “overlooked” more than just “some relevant data” (2014, 19).

Like Leir, Arden is being pursued by murderers and anticipates this attempt on his life through a dream. Wells noted that “They both dream that they are attacked by two persons, one of whom, in each case, carries a falchion” (1939, 436). If we examine the verbal details of this moment in Arden of Faversham, we once again find the line-opening, “God grant,” uniquely associated with prophetic dreams in Arden’s line, “God grant this vision bedeem me any good” (AF, 6.31), which parallels, “God graunt we do not miscarry in the place” (KL, 19.1478), and, in particular, Kyd’s Cornelia: “God graunt these dreames to good effect bee brought” (Corn., 3.1.65). We should also note the unique co-occurrence of the (plural) noun “visions” in Kyd’s closet drama (Corn., 3.1.63). Arden tells his companion, Franklin:

Leir tells Perillus:

The three-word unit, “I did awake” (employed as a formulaic line-ending in both examples), is unique to these plays, yet Jackson fails to cite it in his monograph (2014, 237–9). Here we see verbal parallels combined with similar character relationships. Franklin reassures Arden that “This fantasy doth rise from Michael’s fear,/Who being awaked with the noise he made,/His troubled senses yet could take no rest” (AF, 6.32–34). Franklin, like Perillus in King Leir and the Chorus in Cornelia, “makes the wrong diagnosis” of the dream, “for it is truly predictive” (Presson 1967, 239). Arden responds to his companion, “It may be so, God frame it to the best:/But oftentimes my dreams presage too true” (6.36–37). These lines parallel the Messenger’s line in the dream sequence of King Leir: “Confesse, that dreames do often prove too true” (KL, 19.1484). Jackson makes no mention of this unique collocation in his analysis of the scene (2014, 237–9). Franklin’s speech, “To such as note their nightly fantasies,/Some one in twenty may incur belief” (AF, 6.38–39), recalls King Leir’s, “dreames are but fantasies,/And slight imaginations of the brayne” (KL, 19.1481–2), as well as The Spanish Tragedy’s, “I, I, my nightly dreames have tolde me this” (Sp. T., 1.3.76), and Soliman and Perseda’s, “my nightly dreams foretold me this” (S&P, 5.3.25), through linking the premodifier “nightly” with the topic of ominous dreams. Garnier’s “detailed influence on Kyd” is unmistakeable in these passages (Baldwin 1959, 181). The French tragedian calls dreams “Qu’vn vain semblant, qu’vn fantôme, une image/Qui nous trompe en dormant, et non pas un présage” in his Hippolyte (1573), i.e. ‘A vain pretense, a ghost, an image/Who deceives us while sleeping, and not an omen’.

Arden’s line, “But in the pleasure of this golden rest” (AF, 6.12), matches Erastus’s interrogative, “What thinks Lord Brusor of this strange arrest?” (S&P, 5.2.17). This example could have been prompted by the phonetically similar lexical choices “rest” and “arrest,” for stored units may be manipulated and processed mentally according to both meaning and sound. We also find a unique formulaic line-opening shared between Arden’s line, “With that he blew an evil-sounding horn” (AF, 6.16), and Piston’s line in Soliman and Perseda: “With that he purst the gould, and gave it us” (S&P, 5.2.52). Jackson (2014) focuses on the phrase, “blew an evil-sounding horn,” in his monograph, whilst overlooking the unique parallel with Kyd. Indeed, Jackson (2014) detects just one rare link between Arden’s dream and Kyd’s corpus (with Cornelia), which suggests that the number-specific results obtained from his LION method are unrepresentative.

I hope that my analysis of the dream sequences above has given readers an insight into Kyd’s dramatic methods, and the links of thought and language that could help to distinguish his oeuvre. With a forthcoming edition of Kyd’s plays that will include King Leir and Arden of Faversham – as well as further justifications for the presence of these texts in The Works of Thomas Kyd (Vickers Forthcoming) – it seems fair to say that a new era in Kyd studies, which will enhance our knowledge of his contributions to early modern drama as a whole, has begun.

The additional file for this article can be found as follows: