Note durations and emphasis patterns in early Gregorian chant were not notated. This does not mean that they were not present. It further means that singers were not limited by what notation could express. There is potential here for creative reconstruction.
Guido of Aressa, 11th century, in Micrologus, writes:
I speak of chants as metrical because we often sing in such a way that we appear almost to scan verses by feet, as happens when we sing actual meters--in which one must take care lest neumes of two syllables persist excessively without an admixture of some of three or four syllables. For just as lyric poets join now one kind of foot, now another, so composers reasonably juxtapose different and various neumes. Diversity is reasonable if it creates a measured variety of neumes and phrases, yet in such a way the neumes answer harmoniously to neumes and phrases to phrases, with always a certain resemblance. That is, let the likeness be incomplete, in the manner of the outstandingly lovely chant of St. Ambrose.
The parallel between verse and chant is no slight one, since neumes correspond to feet and phrases to lines of verse. Thus one neume proceeds like a dactyl, another like a spondee, and a third in iambic manner; and you see a phrase now like a tetrameter, now like a pentameter, and again like a hexameter, and many other such parallels.
Let the subdivision [partes] and phrases of both the neumes and the words end at the same time, and do not let a long stay [tenor] on any short syllables or a short stay on long ones create an impropriety, though this will rarely demand attention.
Let the effect of the song express what is going on in the text, so that for sad things the neumes are grave, for serene ones they are cheerful, and for auspicious text exultant, and so forth.
We often place an acute or grave accent above [the vowels in the text for] the notes, because we often utter them with more or less stress, so much so that the repetition of the same note often seems to be a raising or lowering.
Towards the ends of phrases the notes should always be more widely spaced as they approach the breathing place, like a galloping horse, so that they arrive at the pause, as it were, weary and heavily. Spacing notes close together or widely apart, as befits, is a good way to indicate this effect [in writing].
Translation from "Hucbald, Guido and John on Music: three medieval treatises", Warren Babb, 1978.
Update: As you might expect of something involving the Catholic Church, this has been the source of a rather bitter debate for the last 100 years or so. The quote above supports the "mensuralist" position. The alternative is the "equalist-accentualist" position, which supposes notes to all be of the same length, differentiated by emphasizing certain notes. The style most people associate with Gregorian Chant is equalist-accentualist. Even within the equalist-accentualist camp there is bitter division.
Chant does seem to have degenerated into notes of the same length, at about the same time as polyphony was rising, and indeed "plainchant" or "plainsong" came to mean exactly this. The early neumatic notation seems to have been somewhat able to represent duration. This came to be replaced by systems that could precisely specify the note within the scale, but did not specify duration. It also seems that the simple polyphonic style of "organum", which was slower than the original monophonic chant, made it harder to maintain durations.
See eg "Gregorian Chant: a history of the controversy concerning its rhythm", John Rayburn, 1964.