The words are declaimed in the rhythm of natural speech with slight melodic variation with only the slenderest accompaniment, usually on a harpsichord or possibly a cello.

----------------

A recitative is a development of a type of singing done narratively. This type of singing is very old. It pre-dates or runs parallel to instrumental dance/work music. The term generally used is "lay" (German: leich, French: lai, Sc. Gaelic: laoidh, Ir Gaelic: Laoi, etc.).

Originally, people would declaim poetry. The rhythm was that of the language as it was declaimed. The pitches were chosen and discrete. This is in contrast to instrumental music where both pitch and rhythm are precisely defined. Therefore, only pitch in recitative is defined. The written notation is merely a backbone for the accompanist to follow. You do not sing recitative the way that it is written. The written form is a convention. This fact is being lost today as instrumentalists begin to instruct singers, which is counter to the historical record of singers being the model for instrumentalists. The tradition of narrative recitative is fading rapidly.

Italian recitative is a development of lay singing where instead of singing the words to a tune (and varying the tune to match the stresses of the lyrics), harmonic progressions force the melody. Caccini, Peri, et al. did not really create the form, but modulated it to suit their needs. Most of the plot of an opera is communicated through recitative; an aria was originally a highly wrought line of verse where the pattern of the poetry was very simplistic. This lended it to be set to the repetitive meter of instrumental music. Instrumentalists require repeating rhythm to know where they are. A singer does not require this. Therefore, the singer can rely upon the rhythm of the language. If there is more than one accompanist, meter must be imposed or the instrumentalists will strike their notes at different times. Singers do not require this constraint.

The manner by which a composer creates a recitative is by looking at a a sentence of the prose (an exhalation). The composer places all stressed syllables on the first and/or third beats. Unstressed syllables are placed between those stressed beats. In Italian, stressed syllables are spoken about twice as long as unstressed syllables. So the composer would look at groups of stressed and unstressed syllables and assign a meter that worked in that group. For example, one group might consist of an eighth note followed by two sixteenth notes, whereas the next group might consist of a quarter note followed by two eighth notes. They are performed to the same rhythm regardless of how it looks on the paper. Also, rests were added in order to make the beat count equal to 4, not because you are to pause at the rest.

Recitatives are declaimed. This means that you sing the rhythm of the words as you do when you speak loudly to many people in a large space. When you do this, your unstressed syllables come up in volume. This has the effect of making the sentences take more time as the words are pulled out like an Accordion. As an example, consider the speech of Martin Luther King Jr. and Barrack Obama. Mr. King was trained to speak at outdoor events. Mr. Obama was not. Mr. King's unstressed syllables are loud and long. He even has vibrato when he speaks. By comparison, Mr. Obama's unstressed syllables are very quiet. It takes almost twice as long for Mr. King to speak a sentence than it takes Mr. Obama.

Recitatives have the rhythm of declaimed speech, but use defined pitches. If there is one accompanying instrument, it is called "dry" (recitativo secco). This type of narrative singing with accompaniment is very, very old. There are now examples found on the island of Skye that date the existence of lyres (cruit, crowd, cryth, chorus, crotte, rotte, etc.) to 300 B.C. If there is more than one accompanying instrument, rhythm must be imposed. This forces the rhythm sung by the singer to be discrete and not in the pattern of declaimed speech. This is called recitativo accompagnato. The words are still narrative and not well-formed. Composers often used this type of recitative when bridging between a dry recitative and a chorus or aria.

Be careful when shifting languages. The French did not like the written form of recitatives because French does not have long, stressed syllables like Italian. Therefore, French recitatives do not follow the Italian model, but are much more indicative of how the recitative should actually be performed. English composers basically followed the written formulae of the Italians and made stressed syllables twice as long as unstressed syllables. This is not a feature of the English language and should not be the way that the recitative should be performed. The singer then has to make a choice. Should the singer follow what is written knowing that it is not the way that recitatives should be performed, or should the singer sing the rhythm of the words as one would declaim in English? The real problem is that recitative never had much history in England (Beowulf being an exception that died out long before recitatives were re-introduced in the late seventeenth century). The culture of instrumental music in England and Commonwealth countries has completely overwhelmed the notion of si canta come si parla. In essence, if the singer chooses to sing a recitative following the rhythm of declaimed English speech, the surrounding culture (instrumentalists, conductors, agents, managers, etc.) will think that the singer is uneducated and sloppy, not that the singer is more knowledgeable than they. It is a difficult choice.